BASEL III:

A GUIDE TO THE
MAIN PROVISIONS

2022 Edition
 Contents

Contents

Introduction 1

1. OVERVIEW 5

1.1 Regulatory Capital 5

1.2 The Basel Committee 5

1.3 The 1988 Capital Accord 6

1.4 Reasons for Change 7

1.5 The Structure of Basel II 7

1.6 Lessons from the Global Financial Crisis 8

1.7 The Three Pillars of Basel 10

1.8 The Building of Basel III 12

2. CONSOLIDATED SUPERVISION 14

2.1 Supervision of Groups 14

2.2 Home/Host State Responsibilities 17

3. COMPONENTS OF AND DEDUCTIONS FROM CAPITAL 18

3.1 Definition of capital 18

3.2 Common Equity Tier 1 18

3.3 Additional Tier 1 Capital 20

3.4 Tier 2 Capital 27

 Contents

3.5 Minority Interests 29

3.6 Deductions from Capital 29

4. THE STANDARDISED APPROACH TO CREDIT RISK 36

4.1 Standardised Credit Risk approach under Basel III 36

4.2 Sovereign Exposures 37

4.3 Public Sector Entities 38

4.4 International Organisations 39

4.5 Bank exposures 39

4.6 Covered Bonds 42

4.7 Exposures to Securities Firms and other Financial Institutions 43

4.8 Corporate Exposures 43

4.9 Subordinated Debt and other Capital Instruments 45

4.10 Equity Exposures 46

4.11 Qualifying Holdings 47

4.12 Retail Exposures 47

4.13 Real Estate Exposures 49

4.14 Summary 55

4.15 Risk Weight Multiplier for Exposures with a Currency Mismatch 57

4.16 Exposures Giving Rise to Counterparty Credit Risk 57

4.17 Credit Derivatives 57

 Contents

4.18 Defaulted Exposures 57

4.19 Other Assets 58

4.20 Off-balance Sheet Items 59

4.21 External Credit Assessment Institutions (Rating Agencies) 60

4.22 Transitional Arrangements for Equity Exposures 63

5. INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB) 64

5.1 Introduction 64

5.2 Roll Out of IRB across a Given Asset Class 66

5.3 Risk Components 67

5.4 Minimum Requirements 68

5.5 No IRB Treatment 70

5.6 Foundation IRB for Corporate Exposures 70

5.7 Risk weighted assets 71

5.8 Advanced IRB for Corporate Exposures 71

5.9 Sovereign Exposures 72

5.10 Exposures to Banks and certain Securities Firms 72

5.11 Retail Exposures 73

5.12 Common Definition of Default 76

5.13 Corporate Specialised Lending 78

5.14 Defaulted Exposures 81

 Contents

5.15 Equity Exposures 82

5.16 Purchased Receivables 82

5.17 Treatment of Expected Losses 85

5.18 Transitional Arrangements for Equity Exposures subject to an IRB

treatment under Basel II 86

6. CREDIT RISK MITIGATION 88

6.1 Introduction 88

6.2 Legal Certainty 88

6.3 Credit Risk Mitigation under the Standardised Approach 89

6.4 Credit Risk Mitigation under IRB 97

6.5 Other Credit Risk Mitigation Techniques under the Foundation IRB
Approach 101

6.6 Credit Risk Mitigation under Advanced IRB 102

6.7 Credit Risk Mitigation for Retail Exposures 103

6.8 Purchased Receivables 104

7. SECURITISATION 105

7.1 Introduction 105

7.2 Background 105

7.3 Types of Securitisations 106

7.4 Definitions 107

7.5 Operational requirements for the recognition of a securitisation 110

 Contents

7.6 Due Diligence 114

7.7 Capital Requirements 115

7.8 Hierarchy of Methods 117

7.9 Re-securitisation Exposures 118

7.10 Non-Performing Securitisations 118

7.11 Implicit Support 119

7.12 Recognition of Credit Risk Mitigation 119

7.13 Simple, Transparent and Comparable (STC) Securitisations 119

7.14 The Standardised Approach to Securitisation Exposures 133

7.15 External Ratings-based Approach 137

7.16 The Internal Ratings-based Approach 141

7.17 The Internal-assessment Approach to ABCP programmes (SEC-IAA) 148

7.18 Securitisations of Non-performing Loans 150

8. UNSETTLED TRANSACTIONS AND FAILED TRADES 151

8.1 Failed Trades 151

8.2 Free Deliveries 152

9. EQUITY INVESTMENTS IN FUNDS 154

9.1 Exclusions 154

9.2 The Look-through Approach (LTA) 154

9.3 The Mandate-based Approach (MBA) 155

 Contents

9.4 Capital Calculation under LTA or MBA 156

9.5 Fall-back Approach 156

9.6 Funds-of-Funds 156

9.7 Partial Use of an Approach 156

10. COUNTERPARTY CREDIT RISK 157

10.1 Introduction 157

10.2 Definitions relevant to the CCR Framework 160

10.3 Calculation of Capital Charges 161

10.4 The Standardised Approach to CCR 162

10.5 Internal Models Method 173

10.6 CCR in the Trading Book 178

10.7 Exposures to CCPs 179

11. MARKET RISK 187

11.1 Introduction 187

11.2 The 1996 Market Risk Amendment 188

11.3 The Concept of Market Risk 191

11.4 The Basel Committee Responds to the Financial Crisis 205

11.5 The Basel Committee’s May 2012 Proposals 210

11.6 The Second Consultative Document 213

11.7 Basel III - The Definition of the Trading Book 220

 Contents

11.8 Basel III - The New Market Risk Framework for Banks Using Internal
Models 224

11.9 Basel III - The Standardised Approach to Credit Risk 241

11.10 Basel III - Final Standardised Approach to Market Risk 248

11.11 Basel III - The Simplified Standardised Approach 261

12. CREDIT VALUATION ADJUSTMENT 275

12.1 Introduction 275

12.2 Hedging 277

12.3 Banks using the IRB Approaches for Credit Risk 278

12.4 The Basic Approach to CVA 278

12.5 The Standardised Approach for CVA 280

13. OPERATIONAL RISK 284

13.1 Introduction 284

13.2 Definition 284

13.3 The Standardised Approach 284

14. LEVERAGE RATIO 291

14.1 Introduction 291

14.2 Exposure Measurement 293

14.3 The Leverage Ratio for G-SIBs 302

15. LARGE EXPOSURES 304

15.1 Introduction 304

 Contents

15.2 Scope 305

15.3 Definition of a Large Exposure 305

15.4 Limit on Large Exposures 308

15.5 Measuring Exposures 309

15.6 Trading Book Exposures 311

15.7 Sovereign Exposures 312

15.8 Inter-bank exposures 312

15.9 Covered Bonds 312

16. LIQUIDITY COVERAGE RATIO 316

16.1 Introduction 316

16.2 Calculation of the LCR 317

16.3 Scope of Application 319

16.4 High Quality Liquid Assets 320

16.5 Alternative Approaches to HQLA with Insufficient HQLA 329

16.6 Cash Inflows and Cash Outflows 333

16.7 Cash Inflows 347

16.8 Transitional Provision 350

16.9 Summary 350

17. NET STABLE FUNDING RATIO 355

17.1 Introduction 355

 Contents

17.2 Basic Principles 356

17.3 Definition of Available Stable Funding (ASF) 357

17.4 Required Stable Funding 360

17.5 Summary 369

18. MARGIN REQUIREMENTS FOR NON-CENTRALLY CLEARED DERIVATIVES 372

18.1 Introduction 372

18.2 Approaches and application 372

18.3 Margin Requirements 376

Conclusion 384

Glossary of words and expressions 389

Endnotes 401
 Contents

Introduction

 Back to contents
 Introduction

Introduction

Welcome to our guide to Basel III.

We have prepared this guide for banks and their regulators because we have found introductory
reference guides to Basel III difficult to find. There is, in our view, a need for a guide to Basel
III that assumes some knowledge of bank capital structures and business models but assumes
no great knowledge of the foundational rules on which they are based. The enduring
importance of the work of the Basel Committee on Banking Supervision (the Basel Committee)
has, in our view, increased this need.

Basel III does not actually apply directly anywhere. The nature and extent of its application is
dependent on its implementation by governments and regulators. But that does not diminish
its importance as a set of fundamental standards of prudential regulation in the banking sector.
There are many situations when knowledge of expectations based on Basel, in addition to
binding rules in specific jurisdictions, is important. Most obviously, the Basel regime
determines, or at least influences, the legislation and rules on bank prudential supervision that
governments and regulators make. In addition, when agreeing to modify or waive its own rules,
a banking supervisor may not wish to deviate from Basel standards. Banks requesting such
waivers or modifications are therefore well advised to understand what the Basel standards
are. Basel also plays an important role in the degree of deference that regulators are prepared
to give to the regimes of regulators in other jurisdictions: substantive adherence to Basel has,
in short, become a mark of seriousness and credibility in banking supervision. This has
numerous consequences, from the willingness of a regulator to cede the lead (group)
supervisory role over a banking group to a regulator in another jurisdiction following a merger,
to the willingness of a regulator to permit a foreign bank to operate a branch in its jurisdiction.
These are but some of the factors that are leading many banks to attach greater importance
over time to knowledge and understanding of Basel requirements when engaging with their
regulators.

Global finance has been seriously challenged since the global financial crisis of 2007-9, but
arguably the main factor keeping it together has been the maintenance of global standards.
These standards are, in effect, evaluated by regulators assessing each other when they consider
the risks to banks that they supervise interacting with, or being in the same groups as, banks
established in other jurisdictions. Basel provides regulators with a benchmark against which to
assess the credibility of the regimes operated by their counterparts in other jurisdictions.

Against this backdrop, we hope that this guide will be of interest and use to banks, their
regulators and anyone with an interest in prudential regulation.

The latest substantive episode in the Basel story, Basel III, has been a very long time in coming.
The original consultation documents published after the 2007-2009 financial crisis were issued
in December 2009 and finalised as standards in 2010, addressing certain urgent lessons that the
Basel Committee drew from that crisis. A major focus of these standards was on the definition
of capital, a feature of the Basel regime that had been almost unchanged since 1988, and on
liquidity. Since then, the Basel Committee has worked out new approaches to capital
requirements for credit risk, market risk, leverage, operational risk, large exposures, as well
as the supervisory review process and market disclosure that constitute a total replacement of
the Basel II Capital Accord.

 Back to contents 2
 Introduction

In some areas Basel III builds on, and refines, Basel II, such as for the standardised approach to
credit risk or the foundation internal ratings-based approach (F-IRB) to credit risk. In others,
it restricts the availability of advanced models for calculating capital requirements permitted
under Basel II, such as the advanced internal ratings-based (A-IRB) approach for credit risk, or
the internal measurement approach for operational risk, which is withdrawn. In others, a
wholly new and more risk-sensitive approach has been adopted, as under the new market risk
framework. Originally, Basel III was expected to be implemented in full by 1 January 2022,
although this was put back to 1 January 2023 due to the economic dislocations caused by the
coronavirus pandemic. The United Kingdom and the EU have announced that implementation
of the final elements of Basel III will be delayed until 2025.

The purpose of this guide is to provide a summary of the main aspects of Basel III as is intended
to be implemented on 1 January 2023. The intention is to explain, without indulging in
unnecessary mathematics, how the new rules work and what the principal relevant
requirements are. However, we do not go into the details of all the new rules and this guide
is no substitute for reviewing Basel III itself, which is very long and complex. Given the genesis
of Basel III over the past 13 years, instead of referring to individual standards released by the
Basel Committee since 2010, we refer instead to the comprehensive text of Basel III published
on the Committee’s website, and revised in 2021. This will facilitate reference to individual
requirements, where appropriate, although it cannot exclude future changes made by the Basel
Committee and incorporated subsequently into the consolidated text.

Basel III, insofar as it has not yet been implemented already, will make major changes to the
regulatory capital requirements for banks. The intention is to ensure that banks’ capital
requirements correspond more closely to the risks incurred in the course of their business, and
with their internal allocation of economic capital.

Basel III does not change the minimum capital ratio, which remains set at 8%, although national
supervisors have the power to set higher ratios for banks if they consider this to be prudentially
justified. However, it would be inaccurate to think that the old 8% ratio remains the same, as
under Basel III banks must, additionally, meet a capital conservation buffer and (in certain
circumstances) a counter-cyclical capital buffer, as well as satisfying a leverage ratio. Also,
both the composition of capital and the minimum requirements for common equity Tier 1
capital, additional Tier 1 capital and Tier 2 capital changed under the 2010 standard, as has
the calculation of risk-weighted assets, so it is not particularly meaningful to compare capital
ratios under Basel III with those under Basel II.

This guide omits discussion of the requirements for total loss absorbing capital (TLAC) imposed
on global systemically important banks. These requirements, which were inspired by the
financial crisis, were issued by the Financial Stability Board (FSB). TLAC is only relevant to G-
SIBs and (at national discretion) to domestic systemically important banks, and therefore is of
concern only to such institutions and their groups. The application of the TLAC regime varies
significantly around the world.

IMPORTANT NOTE: This guide is intended to provide assistance in understanding certain aspects
of Basel III. It should not be relied upon as a substitute for legal advice which should be sought
as required. Basel III is long and contains many technical provisions, and its application to
specific situations or particular transactions will require careful consideration.

At Slaughter and May we have a long-established prudential regulatory practice, advising

international and domestic banks on matters as varied as capital structure, domicile,
structuring cross-border M&A transactions, capital issuances, risk weightings and risk mitigation
transactions, liquidity, booking models and governance requirements. We are frequently
involved in making technical submissions to regulators on these issues.

 Back to contents 3
 Introduction

If you would like to discuss anything in this guide, or any aspect of the prudential regulation of
banks, please contact a member of the team:

Jan Putnis
Partner
T: +44 (0)20 7090 3211
M: +44 (0)7887 540 490
E: [email protected]

Nick Bonsall
Partner
T: +44 (0)20 7090 4276
M: +44 (0)7887 540 492
E: [email protected]

Tolek Petch
Associate
T: +44 (0)20 7090 3006
M: +44 (0)7795 656 682
E: [email protected]

Tim Fosh
Senior Counsel
T: +44 (0)20 7090 3791
M: +44 (0)7917 585 831
E: [email protected]

Kristina Locmele
Senior Counsel
T: +44 (0)20 7090 3963
M: +44 (0) 7825 006 724
E: [email protected]

David Shone
Associate
T: +44 (0)20 7090 5242
M: +44 (0)7917 267 092
E: [email protected]

Emily Bradley
Senior Professional Support Lawyer
T: +44 (0)20 7090 5212
M: +44 (0)7917 426 538
E: [email protected]

Selmin Hakki
Senior Professional Support Lawyer
T: +44 (0)20 7090 5153
M: +44 (0)7825 313 093
E: [email protected]

 Back to contents 4
 OVERVIEW

1. OVERVIEW

1.1 Regulatory Capital

1.1.1 The purpose of regulatory capital is to support the day-to-day operations of a

bank. It serves as a permanent fund which is able to absorb losses arising in
the ordinary course of a bank’s operations, and therefore enables a bank to
carry on trading in times of economic adversity. Capital also serves important
prudential purposes. Firstly, it helps to insulate a bank from insolvency, and
thereby protects the bank’s depositors if the bank becomes insolvent.
Secondly, through reducing the risk of bank failures, capital helps to reduce
systemic risk within the financial system. This is the risk that the insolvency
of a single bank may, owing to the exposures to it of other banks and financial
institutions, trigger further insolvencies, potentially causing serious problems
for the financial system. This risk became of global significance following the
bankruptcy of Lehman Brothers in September 2008.

1.1.2 Because of the importance of capital in protecting depositors and reducing

systemic risk, banking regulators in all developed countries have specified
minimum requirements for regulatory capital. The regulation of a bank’s
capital therefore supplements other aspects of prudential regulation, such as
minimum cash or liquidity ratios, and approval of a bank’s shareholders and
senior management to ensure that they are “fit and proper”.

1.1.3 In the UK banks are required under the Financial Services and Markets Act
2000 to maintain financial resources that are adequate in relation to the
regulated activities that they carry on. Detailed rules define what constitutes
capital for regulatory purposes and set out the amount of capital that a bank
must hold to cover specified risks. Currently, the requirements for UK
authorised banks are mainly set out in various pieces of EU legislation that
were applicable when the UK left the European Union at 11 pm on 31
December 2020, supplemented by the rulebook of the Prudential Regulation
Authority (“PRA”). A major current topic in UK financial regulation is the
extent to which the EU-inspired rules and requirements may suitably be
amended to promote the position of London as a global financial centre. As
the UK has stated its intention to abide by international standards, it seems
highly likely that in the area of prudential regulation the UK will be guided by
the Basel standards (which, as a Committee member, it has had a role in
formulating).

1.2 The Basel Committee

1.2.1 The Basel Committee was established by the central bank governors of the
Group of Ten countries at the end of 1974 following serious disturbances in
the international currency and banking markets. Its membership has
expanded over time and currently comprises senior officials with bank
regulatory and financial supervisory responsibilities from central banks and
banking regulators in 28 jurisdictions. The chairman is Pablo Hernádez de
Cos, who is also head of the Bank of Spain. The Committee now reports to an

 Back to contents 5
 OVERVIEW

oversight body, the Group of Central Bank Governors and Heads of Supervision
(“GHOS”), which comprises central bank governors and (non-central bank)
heads of supervision from member countries. The current chair of the GHOS
is François Villeroy de Galhau, Governor of the Banque de France.

1.2.2 The members of the Basel Committee are Argentina, Australia, Belgium,
Brazil, Canada, China, the European Union, France, Germany, Hong Kong SAR,
India, Indonesia, Italy, Japan, Korea, Luxembourg, Mexico, the Netherlands,
Russia, Saudi Arabia, Singapore, South Africa, Spain, Sweden, Switzerland,
Turkey, the UK and the United States. In addition to member organisations, a
number of institutions currently hold observer status. These include the
following country observers: the Central Bank of Chile/Banking and Financial
Institutions Supervisory Agency, the Central Bank of Malaysia, and the Central
Bank of the United Arab Emirates; and the following supervisory groups and
international agencies or bodies: the Bank for International Settlements, the
Basel Consultative Group, the European Banking Authority, the European
Commission and the International Monetary Fund. Its Secretariat is located
at the Bank for International Settlements (BIS) in Basel, Switzerland.

1.2.3 The stated mandate of the Basel Committee is to strengthen the regulation,
supervision and practices of banks worldwide with the purpose of enhancing
financial stability. Its focus has traditionally been on internationally active
banks, although the Committee’s standards have been applied more widely,
particularly in the European Union.

1.2.4 The Basel Committee formulates standards and guidelines, and recommends
statements of best practice. The rules and guidance adopted by the Basel
Committee have no legal force and their authority derives from the
commitment of banking supervisors in member countries (and, increasingly,
non-member countries) to implement the requirements agreed by the
Committee. The Committee has adopted standards on a wide range of issues
relevant to banking supervision, including banks’ foreign branches, core
principles for banking supervision, core principles for effective deposit
insurance, internal controls, supervision of cross-border electronic banking
and risk management guidelines for derivatives.

1.2.5 However, in recent decades, the Basel Committee has devoted most of its
attention to regulatory capital. It has also been active in the important areas
of liquidity risk and developing frameworks for the recovery or orderly wind-
down of internationally active banks that get into financial difficulties. The
Basel Committee is also a forum for consultation on aspects of banking
supervision. Its objective is to improve the quality of banking supervision
through exchanging information on national supervisory arrangements,
improving the effectiveness of techniques for supervising internationally
active banks and setting minimum supervisory standards.

1.3 The 1988 Capital Accord

1.3.1 The first Basel Capital Accord was adopted in 1988. The 1988 Capital Accord
was based on four core principles which are retained by Basel III. These are:

A common definition of regulatory capital.

 Back to contents 6
 OVERVIEW

The determination of the amount of capital that a bank is required to

hold through applying a risk weight to particular assets and off-balance
sheet items (“risk weighted assets”).

The setting of a minimum ratio of regulatory capital to risk weighted

assets (the “risk asset ratio”).

Consolidated supervision of banking groups.

1.4 Reasons for Change

1.4.1 After 1988 there were major changes in financial markets with the
development of new financial products, trading strategies and risk mitigation
techniques. A review of these developments led the Basel Committee to
identify several defects in the 1988 Capital Accord. In particular the Basel
Committee considered that:

Capital ratios were not always a good indicator of a bank’s financial

condition.

The approach of the 1988 Accord to risk weights provided a crude

indicator of risk.

A lack of sensitivity to credit risk provided an incentive for banks to

exploit differences between economic capital and regulatory capital (for
example, by increasing lending to poor credit quality borrowers, which
attracted a higher rate of interest, but the same capital charge).

There were inadequate incentives for banks to use credit risk mitigation
techniques, in particular collateral and credit derivatives, as such
techniques were often not recognised.

There were insufficient incentives for banks to develop more accurate

internal measurements of risk as this would not result in any capital
savings.

1.4.2 The Basel Committee concluded that these difficulties resulted from a lack of
sensitivity in the 1988 Capital Accord and that it should therefore be replaced.

1.5 The Structure of Basel II

1.5.1 The structure of Basel II differed significantly from that of Basel I. It was
considerably more complex and in many areas provided a choice of different
approaches for determining capital requirements (which, in some significant
areas, have been narrowed by Basel III). For example, Basel II set out three
different ways of calculating credit risk and three (or four) ways of
determining the capital charge for operational risk (now only one). Generally,
banks were free to choose between more complex methodologies, with the
potential for capital savings, and simpler approaches, that generally lead to
a higher capital charge, but with lower operational and systems costs.

1.5.2 The focus of Basel II (like Basel III) was on internationally active banks.
However, the Basel Committee considered that the principles developed in

 Back to contents 7
 OVERVIEW

Basel II might be suitable as an international benchmark. Overall, Basel II was

considerably more risk sensitive than Basel I. It also marked a shift in favour
of greater reliance on banks’ internal models and methodologies, and external
credit ratings. Both of these developments would come under considerable
scrutiny in the aftermath of the financial crisis leading to Basel III.

1.6 Lessons from the Global Financial Crisis

1.6.1 Before Basel II could be fully implemented on 1 January 2008 the global
financial crisis hit. The explanations for why it happened look set to be
debated by economists and historians in the decades to come. Nonetheless,
certain clear lessons were apparent early on, and guided the Basel Committee
in its decade-long work on Basel III.

1.6.2 Given its delayed implementation, Basel II clearly cannot be said to have
caused the financial crisis, although it may be fair to say that, had the Basel
II requirements been fully implemented earlier, or had measures adopted by
the Basel Committee (such as those relating to securitisations) been fully
applied at the relevant time, the crisis might not have played out in quite so
severe a manner. Hindsight, of course, facilitates such analyses.

1.6.3 However, even if Basel II had been fully implemented in 2007 there is no
reason to doubt that the market's appetite to accept risk in the pursuit of
yield, balance sheet arbitrage, rapid financial innovation and complexity of
financial product design, would nevertheless have occurred, possibly in much
the same way.

1.6.4 Extensive ex post analysis of the financial crisis led to a broad consensus
among governments, regulators and market participants that there was a
fundamental failure in market discipline as much as in the regulation of
markets, and that the causes of this failure went far beyond mere
inadequacies in bank capital requirements. Nevertheless, a consensus also
rapidly developed that there were a number of deficiencies in the Basel II
regulatory framework that needed to be addressed.

1.6.5 A fundamental failing demonstrated by the financial crisis was that the
financial sector did not hold enough capital. More highly capitalised
institutions would have been better placed to absorb losses without requiring
government support or enforced mergers with stronger institutions. The
quality of much financial sector capital also proved to be inadequate and did
not absorb losses in the crisis. For example, in the run up to the crisis banks
continued paying dividends and coupons on preference shares and hybrid
securities for fear otherwise of signalling financial weakness. Holders of
subordinated debt did not suffer significant losses when insolvent institutions
were rescued by taxpayers. In fact, holders of all these forms of capital
benefited from public sector equity injections which ranked behind other Tier
1 and Tier 2 capital instruments.

1.6.6 The lack of emphasis on liquidity in the Basel II framework was also striking.
Capital requirements were – and are – concerned with solvency, and aim to
enable an institution to continue trading in times of financial adversity.
However, a bank can also fail as a result of insufficient liquidity. In its initial

 Back to contents 8
 OVERVIEW

stages the financial crisis manifested itself through a lack of liquidity at such
institutions as Northern Rock in the UK and Bear, Stearns in the United States.

1.6.7 Another feature of the Basel II regime that received criticism was that it did
not impose restrictions on leverage. This gave rise to incentives for banks to
engage in riskier trading activities in the relatively benign economic
conditions that prevailed prior to the summer of 2007, which boosted
revenues and profits in that period, but at the same time increased systemic
risk and the possibility of bank failures. By way of example, if a bank is
leveraged 30:1 then a fall in the value of its assets of 3.4% will generate losses
greater than the amount of its common equity resulting in the bank becoming
balance sheet insolvent.

1.6.8 The focus of Basel II on capital also resulted in regulators overlooking the
growth of systemic risk as they concentrated on the position of individual
institutions. Ultimately, the interconnectedness of large and small
institutions created through complex webs of OTC derivatives and the rapid
growth of an unregulated “shadow banking sector” resulted in a situation
where the failure of a moderately-sized investment bank, Lehman Brothers,
risked bringing down the global financial system.

1.6.9 In addition, the capital requirements for banks’ trading books and
securitisations failed to reflect the real level of risk in those areas. Financial
institutions were, therefore, incentivised to book transactions in the trading
book, many of which were illiquid assets such as the infamous collateralised
debt obligations (CDOs), which were securitisations of securitisations, or even
more complex instruments. In the absence of a ready market, institutions
marked those assets to model, but little actual trading took place. Once the
crisis broke, firms experienced increasingly large losses in their trading
portfolios, which, together with a lack of liquidity, was the proximate cause
of Bear, Sterns’ and Lehman Brothers’ failure. Rating downgrades to assets
also became a significant cause of mark-to-market losses but, unlike credit
defaults, the Basel II framework did not take this problem into account.

1.6.10 The financial crisis also demonstrated structural flaws in the value at risk
(VaR) models used by financial institutions to calculate regulatory capital
requirements for market risks: short observation periods combined with
historically low volatility in market prices (with limited data sets that did not
include data from a severe economic downturn), models systematically
underestimating the significance of low frequency high impact events,
overlooking the importance of systemic risk and the presence of uncertainties
that are not capable of being modelled. According to VaR measures, risk was
low in spring 2007; in fact, the system was fraught with huge systemic risk.

1.6.11 Perhaps most damning of all was that many of the senior managers of the
institutions most at risk did not understand, and in many cases were not in a
position to understand, these matters1. This resulted in an overreliance in
many firms on technical staff who effectively determined matters of critical
importance to the stability of the institutions concerned. In many cases banks’
trading book portfolios, or parts of them, proved to be extremely difficult to
value once liquidity evaporated, the models previously employed by the banks
having broken down and senior management of many banks having lost

 Back to contents 9
 OVERVIEW

confidence in those models and their seemingly indecipherable mathematical

complexity.

1.6.12 The result was a series of unprecedented government bail-outs worldwide2,

or facilitated mergers of failing financial institutions. As this amounted to the
socialisation of private sector losses, as governments around the world sought
to rescue their financial systems, it is wholly understandable that banking
regulators would seek to improve the solvency and liquidity of financial
institutions in future to mitigate moral hazard.

1.7 The Three Pillars of Basel

1.7.1 Basel II adopted two complementary pillars to the minimum risk-based capital
requirements (Pillar 1). The second pillar (Pillar 2) refers to the supervisory
review process, which essentially describes how supervisors should regulate
internationally active banks in their jurisdiction. In particular, it required that
banking supervisors should have the power to compel banks to hold capital in
excess of the 8% minimum ratio where this was justified. Standards were also
adopted for the control of interest rate risk in a bank’s loan portfolio, and to
capture other risks not specifically covered under Pillar 1 (for example,
certain risks arising out of securitisations).

1.7.2 The third pillar (Pillar 3) relates to market disclosures of information. The
intention was that pressure from a bank’s counterparties, analysts and rating
agencies would serve to reinforce the minimum capital standards and ensure
that banks carried on their business prudently. This goal was demonstratively
not met in the run up to the global financial crisis.

Outline of Pillar 2

1.7.3 The purpose of Pillar 2 “is intended not only to ensure that banks have
adequate capital and liquidity to support all the risks in their business, but
also to encourage banks to develop and use better risk management
techniques in monitoring and managing their risks”3. National supervisors are
expected to evaluate how well banks are assessing their capital needs relative
to their risks and to intervene, where appropriate. This is intended to create
a dialogue between banks and supervisors to ensure that where deficiencies
are identified, prompt and decisive action can be taken to reduce risk or
restore capital4. In the event of deficiencies in an individual bank’s risk
management and internal controls increased capital requirements should not
be seen as the only option and other means, such as strengthening risk
management, applying internal limits, strengthening the level of provisions
and reserves, and improving internal controls, must also be considered.
Furthermore, capital should not be regarded as a substitute for addressing
fundamentally inadequate risk control or management5.

1.7.4 According to the Basel Committee:

“There are three main areas that might be particularly suited to treatment
under Pillar 2: risks considered under Pillar 1 that are not fully captured by
the Pillar 1 process (e.g. credit concentration risk); those factors not taken
into account by the Pillar 1 process (e.g. interest rate risk in the banking book,

 Back to contents 10
 OVERVIEW

business and strategic risk); and factors external to the bank (e.g. business
cycle effects)”6.

1.7.5 Pillar 2 is based on the following four key principles which are set out below
without the detailed elaboration in the Basel III text:

Principle 1: banks should have a process for assessing their overall capital
adequacy in relation to their risk profile and a strategy for maintaining
their capital levels7.

Principle 2: supervisors should review and evaluate banks’ internal

capital adequacy assessments and strategies, as well as their ability to
monitor and ensure their compliance with regulatory capital ratios.
Supervisors should take appropriate supervisory action if they are not
satisfied with the result of this process8.

Principle 3: supervisors should expect banks to operate above the

minimum regulatory capital ratios and should have the ability to require
banks to hold capital in excess of the minimum 9.

Principle 4: supervisors should seek to intervene at an early stage to

prevent capital from falling below the minimum levels required to
support the risk characteristics of a particular bank and should require
rapid remedial action if capital is not maintained or restored 10.

1.7.6 These principles are then followed with detailed requirements in respect of:

risk management11. This requires banks to carry out an internal capital

adequacy assessment process (ICAAP) which is then reviewed by
supervisors. There are also provisions on risk concentration, reputational
risk, valuation practices, stress testing and liquidity risk management;

interest rate risk in the banking book. This is defined as the “the current
or prospective risk to the bank’s capital and earnings arising from adverse
movements in interest rates that affect the bank’s banking book
positions”. When interest rates change, the present value and timing of
future cash flows change. This in turn changes the underlying value of a
bank’s assets, liabilities and off-balance sheet items and hence its
economic value. Changes in interest rates also affect a bank’s earnings
by altering interest rate-sensitive income and expenses, affecting its net
interest income12. There are three main types of such risks: (1) gap risk
arising from the term structure of banking book instruments; (2) basis
risk reflecting the impact of relative changes in interest rates for
financial instruments that have similar maturities but are priced using
different interest rate indices; and (3) option risk from derivative
positions or from elements in a bank’s assets, liabilities and off-balance
sheet items with optionality13. Basel III sets out 12 principles and
prescribes interest rate shock scenarios to be undertaken by banks. This
is supplemented by guidance on how to apply the requirements 14;

risks not fully taken into account under the frameworks for credit risk
(including counterparty credit risk and securitisation) 15, market risk16 and
operational risk17;

 Back to contents 11
 OVERVIEW

supervisory review of bank compensation practices including bonuses18;

risk data aggregation and risk reporting 19; and

liquidity monitoring metrics that measure other dimensions of a bank’s

liquidity and funding risk profile than is done under the liquidity
standards of the LCR and NSFR20.

Outline of Pillar 3

1.7.7 A new version of the Pillar 3 standard comes into force on 1 January 2023.
According to the Basel Committee:

“The provision of meaningful information about common key risk metrics to

market participants is a fundamental tenet of a sound banking system. It
reduces information asymmetry and helps promote comparability of banks’
risk profiles within and across jurisdictions. Pillar 3 of the Basel framework
aims to promote market discipline through regulatory disclosure
requirements. These requirements enable market participants to access key
information relating to a bank’s regulatory capital and risk exposures in order
to increase transparency and confidence about a bank’s exposure to risk and
the overall adequacy of its regulatory capital” 21.

1.7.8 Banks must publish their Pillar 3 report in a standalone document that
provides a readily accessible source of information for users. The Pillar 3
report may be added to, or form a discrete section of, a bank’s normal
accounting disclosures, but needs to be easily identifiable to users22. Pillar 3
is based on five guiding principles:

disclosures should be clear23;

disclosures should be comprehensive 24;

disclosures should be meaningful to users 25;

disclosures should be consistent over time 26; and

disclosures should be comparable between banks27.

1.7.9 There are detailed prescribed disclosure requirements (with specific reporting
periods for each) in respect of key prudential metrics and RWA 28, the
comparison between modelled and standardised RWA29 (where firms use an
internal model), the composition of capital and, for global systemically
important banks TLAC30, capital distribution constraints if required by national
supervisors at a jurisdiction level31, asset encumbrance32, information related
to remuneration33, credit risk34, counterparty credit risk35, market risk36, credit
valuation adjustments 37, operational risk38, interest rate risk in the banking
book39, bank-specific counter-cyclical capital buffers 40 and liquidity41.

1.8 The Building of Basel III

1.8.1 Following consultation documents published in December 2009, the first part
of Basel III was concerned mainly with the definition of capital eligible to

 Back to contents 12
 OVERVIEW

meet capital requirements and bank liquidity standards. It was finalised in

July 2010, although subject to subsequent modification in January 2011. The
2010 publication also set out the capital conservation and countercyclical
buffers, as well as changes to address the credit valuation adjustment risk.
After that date, the Basel Committee turned its attention to other aspects of
the capital framework that were shown to be inadequate in the financial
crisis. The liquidity coverage ratio followed in 2013, the leverage ratio in
2014, the net stable funding ratio also in 2014, standards on securitisation in
2014 and 2016, and replacements for the standardised and IRB approaches to
credit risk, together with credit valuation adjustments and operational risk in
December 2017, and a revised market risk framework in 2019.

1.8.2 As mentioned above, implementation of the full Basel III framework was
delayed in 2020 from 1 January 2022 to 1 January 2023. It will therefore be
just over 13 years after publication of the first consultation document on Basel
III that the new standard fully enters into force and 15 years before it is fully
implemented in the UK and the EU. Unsurprisingly, a subsequent global
economic crisis – this time triggered by the coronavirus pandemic – will have
come (and may have gone) by the time Basel III is finally due for
implementation. What lessons may be drawn from the pandemic for banking
regulation remains to be seen, although, so far, there have been no major
banking failures, facilitated by exceptional government support provided
during the acute stages of the pandemic.

The Three Pillars of Basel at a glance

Basel II was based on three pillars which were intended to be interdependent and
mutually reinforcing. These pillars also remain in place under Basel III with
modification:

 Pillar 1: Minimum Capital Standards.

 Pillar 2: The Supervisory Review Process.

 Pillar 3: Market Discipline.

 Back to contents 13
 CONSOLIDATED SUPERVISION

2. CONSOLIDATED SUPERVISION

2.1 Supervision of Groups

2.1.1 Basel III applies on a consolidated basis to internationally active banks.

Consolidated supervision is considered to be the best means to provide
supervisors with a comprehensive view of risks within a banking group42. This
includes any holding company or companies of a bank provided that its
subsidiaries “engage predominantly in banking activities”43. This will depend
on both group structure and the activities engaged in by the holding
company’s subsidiaries. Basel III does not state how the test of
“predominantly” engaging in banking activities should be measured. Nor is
there a clear definition of banking “activities” (unlike in EU law)44. However,
it seems likely that a bare majority of activities carried on by subsidiaries
measured on a balance sheet basis would not of itself be sufficient for a
holding company to be subject to the Basel III framework. Moreover, it is clear
from other parts of Basel III that the accord applies to non-banks that engage
in activities comparable to banking (such as leasing, issuing guarantees or
credit cards, providing means of payment, and securities trading).
Supervisors are intended to have discretion in deciding on the application of
consolidated supervision to individual banking groups.

2.1.2 In addition to applying on a group-wide consolidated basis, Basel III also

applies to all internationally active banks at every tier within a banking group
on a fully consolidated basis45. This means that where a banking group subject
to Basel III contains more than one internationally active bank then Basel III
must be applied to each internationally active bank that is a part of the group,
including all of its direct and indirect subsidiaries that engage in “banking
activities”. For example, if a UK banking group has a French subsidiary and a
Japanese subsidiary (both being internationally active) then consolidation
must also be carried out at the level of the French and Japanese banks,
including all of their respective direct and indirect subsidiaries, as well as at
the level of the UK group.

 Back to contents 14
 CONSOLIDATED SUPERVISION

2.1.3 An example of this is set out below.

(A)
Diversified Financial
Group

Holding Company
(B)

Internationally
Active Bank
(C) (D)

Internationally Internationally
Active Bank Active Bank

Domestic Securities
Bank Firm

(A) represents the boundary of the predominant banking group subject to

Basel III. (B), (C) and (D) are intermediate levels at which the accord also
applies on a consolidated basis46.

2.1.4 An alternative to full consolidation set out in Basel III is to apply the
framework to banks on a stand-alone basis, provided that the full book value
of any investments in subsidiaries and significant minority-owned holdings is
deducted from the bank’s capital47.

2.1.5 Although Basel III applies only on a consolidated basis – and sets no solo or
individual capital ratios for specific banks – the Basel III text states that “it is
essential to ensure that capital recognised in capital adequacy measures is
adequately distributed amongst legal entities. Accordingly, supervisors should
test that individual banks are adequately capitalised on a stand alone basis”48.
FAQ 1 adds that the framework “does not prescribe how to measure the solo
capital requirements which is left to individual supervisory authorities”49. The
traditional approach in the UK has been to apply the Basel framework on a
stand-alone basis, as well as on a consolidated or solo consolidated basis.

2.1.6 Basel III applies to the greatest extent possible to all banking and other
relevant financial activities (regulated and unregulated) to groups containing
an internationally active bank 50. Majority-owned or controlled banks, and
securities firms, if subject to broadly similar regulation, should also be fully

 Back to contents 15
 CONSOLIDATED SUPERVISION

consolidated. Insurance activities should not, as a general matter, be

consolidated (see below)51.

2.1.7 Basel III recognises that there may be instances where consolidation is not
feasible or desirable e.g. following a debt-equity conversion previously
contracted and held on temporary basis following conversion. Other
exceptions include where non-consolidation is required by law or if the entity
is subject to different regulation52. Any majority-owned securities and
financial subsidiaries that are not consolidated are deducted from capital,
with the amount of the deduction increased to reflect any capital shortfall
that is not corrected in a timely manner, in such subsidiaries 53.

2.1.8 Significant minority investments in banking, securities and other financial

entities are either excluded from consolidated capital or (under certain
conditions) consolidated on a pro rata basis. Pro rata consolidation may be
appropriate for joint ventures, or where the supervisory authority is satisfied
that the parent is legally or de facto expected to support the subsidiary on a
proportionate basis only, and that the other significant shareholders have the
means and willingness to proportionately support it. The threshold above
which minority investments should be either deducted or consolidated on a
pro rata basis is determined by national accounting or regulatory practices 54.
The Basel approach to the deduction of “material holdings” is therefore less
prescriptive than EU requirements, and has no specified thresholds, although
the EU threshold of 20% to 50% is referred to as “an example”55.

Insurance subsidiaries

2.1.9 The Basel Committee believes that it is “in principle” correct to deduct banks’
equity and other regulatory capital investments in insurance entities, as well
as significant minority investments in such companies 56. This corresponds to
the traditional treatment of insurance subsidiaries and significant investments
in the UK since the 1980s.

2.1.10 However, “[a]lternative approaches that can be applied should, in any case,
include a group-wide perspective for determining capital adequacy and avoid
double counting of capital”57. FAQ 1 adds: “[j]urisdictions can permit or
require banks to consolidate significant investments in insurance entities as
an alternative to the deduction approach on the condition that the method of
consolidation results in a minimum capital standard that is at least as
conservative as that which would apply under the deduction approach”58. If
the method results in lower capital ratios, banks must use these. Where it
results in a capital benefit, this is disallowed. Further, majority-controlled
insurance subsidiaries which are subject to the deduction treatment must
themselves be adequately capitalised. Capital shortfalls that are not
corrected in a timely manner are deducted59.

2.1.11 Where there is a capital surplus in an insurance subsidiary, this may be

reflected in consolidated capital on a pro rata basis where there are minority
interests. If the bank does not have a majority shareholding then any surplus
capital is disregarded as the bank cannot direct the transfer of the surplus
capital60.

 Back to contents 16
 CONSOLIDATED SUPERVISION

2.1.12 Where an insurance subsidiary is not consolidated, supervisors “will ensure”

that the insurance company is itself adequately capitalised to reduce the
potential for loss to the parent banking group.

2.1.13 The treatment of insurance entities under Basel III (which is essentially the
same as under Basel II) reflects the difference in prudential requirements for
insurance companies and banks. There do not exist detailed global standards
for insurance capital comparable to those for banks formulated by the Basel
Committee. Also, insurers run different risks to banks given the nature of
their business. It follows that for bank-assurance groups there may be
significant capital differences depending on whether a bank or an insurance
company sits higher in the group structure. If it is a bank then under Basel III
the general rule is that all equity investments in insurers or reinsurers require
deduction from the banking group’s consolidated capital. If an insurance
company owns an internationally active bank, the Basel III framework will
apply to the bank while national prudential standards for insurers will
determine the treatment of the insurer’s investment in the bank.

2.2 Home/Host State Responsibilities

2.2.1 Basel II stated that the framework was not intended to change the legal
responsibilities of national supervisors for the regulation of their domestic
institutions or the arrangements for consolidated supervision. This statement
is not repeated in Basel III but remains correct. It should be noted that under
the Basel framework, the “home” supervisor is the supervisor of a parent bank
and a “host” supervisor that of a subsidiary. A different definition applies
under EU law with the “home” supervisor being the supervisor in the place of
incorporation/head office and the “host” state is one where foreign branches
are established.

 Back to contents 17
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

3. COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

3.1 Definition of capital

3.1.1 The definition of capital was the first part of Basel III to be finalised in 2010,
and only minor changes have since been made to it. Detailed transitional
provisions were originally set out, including time-limited grandfathering on a
tapering basis for non-compliant capital instruments issued prior to the new
standard. As all these provisions will have expired by 1 January 2023 no
discussion will be provided here.

3.1.2 The types of regulatory capital were originally determined by the Basel
Committee in 1988 and were left unchanged by Basel II61. However, the 2007-
2009 financial crisis demonstrated the need for reform. Under Basel III there
are only three categories of capital: common equity Tier 1, Additional Tier 1
and Tier 2 capital62. Total regulatory capital is the sum of the preceding
items63. Basel III reinforces the position of common equity (and retained
earnings), or core Tier 1 capital, as the predominant form of capital 64. Prior
supervisory approval is required for any instruments that provide for its
“dividends” to be paid in anything other than cash or shares 65. We understand
this to apply not only to ordinary and preference shares, but also to debt
securities where coupons are not (under English law) dividends but instead a
debt. Regulatory capital under Basel III focuses on high-quality capital,
predominantly in the form of shares and retained earnings that can absorb
losses. Basel III also introduced an explicit going-and gone-concern framework
by clarifying the roles of Tier 1 (going concern) and Tier 2 (gone concern)
capital.

3.2 Common Equity Tier 1

3.2.1 Common equity Tier 1 capital comprises the following items:

common shares issued by the bank satisfying the criteria set by the Basel
Committee (or equivalent for mutuals);

share premium (stock surplus) where shares are issued at a premium;

retained earnings;

accumulated other comprehensive income and other disclosed reserves;

common shares issued by consolidated subsidiaries of the banking group

held by outside third parties (i.e. minority interests) which meet the
criteria set out below; less

regulatory adjustments applied in the calculation of common equity Tier

1 capital.

3.2.2 To ensure its quality and consistency Basel III sets out a list of criteria that
common equity must satisfy. These are:

 Back to contents 18
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

it is the most subordinated claim in a liquidation;

it has an unlimited and variable claim in a liquidation after all senior

claims have been repaid (i.e. it is entitled to a claim on residual assets
proportional to its share of issued capital and not limited to a fixed
percentage as with traditional preference shares);

it is perpetual and is never repaid outside of liquidation (other than

discretionary repurchases or other reductions of capital allowed under
company law);

the bank does nothing to create an expectation at issuance that the

instrument will be bought back, redeemed or cancelled (nor does the
statutory framework);

distributions are paid out of distributable items (including retained

earnings) and are not tied or linked to the amount paid at issuance, or
subject to a cap (another feature of traditional preference shares);

there is no circumstance in which distributions are obligatory. Non-

payment is not an event of default. Nor can the bank be required to
make payments in kind (such as alternative coupon settlement
mechanisms characteristic of pre-2010 innovative Tier 1 issues);

distributions are paid only after all legal and contractual obligations have
been met and payments on more senior capital instruments have been
made. This is stated to exclude preferential distributions in respect of
other common equity Tier 166;

the issued capital takes the first and proportionally greatest share of any
losses that occur. Each instrument absorbs losses on a going concern
proportionately. This requirement is hard to understand from an English
law perspective as although losses may be expected to result in a fall in
the share price, there is no sense in which the nominal amount of share
capital is written down following a loss. In certain circumstances an
English company may reduce its capital, but this is certainly not an
automatic process and, for a public company, requires a court process.
Negative reserves do not reduce share capital;

the paid up amount is classified as equity capital under relevant

accounting standards, and not a liability for determining balance sheet
insolvency. Under English company law all preference shares are treated
as equity under the balance sheet insolvency test in section 123 of the
Insolvency Act 1986, although the accounting treatment under IFRS
depends on whether the preference share is redeemable in cash or
another financial asset, or has a mandatory coupon;

the shares are directly issued and paid-in67 and the bank did not fund the
instrument or purchase68;

the paid-in amount is neither secured or covered by a guarantee of the

issuer or a related entity, or subject to an arrangement that legally or
economically results in more senior treatment;

 Back to contents 19
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

the bank’s shareholders have consented to the issue of the instrument;

and

the shares are paid up69 either in cash or other consideration (such as
shares), and if not paid up in cash prior supervisory approval has been
obtained70; and

the shares are separately disclosed on the bank’s balance sheet71.

3.2.3 It follows that the following (amongst others) are not eligible as ordinary
shares under Basel III:

shares with a fixed claim in a winding up;

shares with a preference in either winding up or in respect to payment

of dividends;

shares with a fixed (but fully discretionary) coupon;

shares with an alternative coupon satisfaction mechanism;

shares that are accounted for as a liability; and

indirect issues of shares (e.g. through an SPV). Minority interests in

consolidated subsidiaries can count towards the consolidated common
equity requirement as described below.

3.2.4 These requirements also apply to mutuals and co-operatives (such as building
societies) taking into account their specific constitution and legal structure 72.
Where different from common shares, the instruments must preserve the
quality of the instruments by being fully equivalent to common shares in terms
of capital quality as regards loss absorption. An example of instruments that
might meet the Basel III requirements for core Tier 1 capital are profit
participating deferred shares (PPDS) issued by certain UK building societies.

3.2.5 Shares need not carry voting rights in order to meet the Basel criteria although
non-voting shares must still satisfy all of the criteria set out above.

3.3 Additional Tier 1 Capital

3.3.1 Additional Tier 1 capital consists of the sum of the following:

instruments issued by the bank that meet the criteria for inclusion in
additional Tier 1 capital;

share premium on the issue of such instruments;

instruments issued by consolidated subsidiaries to third parties that meet

the requirements for recognition of minority interests; less

regulatory adjustments applied in the calculation of total Tier 1 capital 73.

3.3.2 FAQ 1 adds that subordinated loans are eligible as additional Tier 1 capital if
the requirements set out in Basel III are met. The issuance of subordinated

 Back to contents 20
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

debt as additional Tier 1 capital has traditionally been attractive owing to the
tax deductibility of interest payments, although if the instruments are
accounted for as a liability, additional requirements in respect of loss
absorbency apply. Further, it is unclear how subordinated debt can meet the
second criterion set out below (subordination to subordinated debt holders)
unless this should (as would be logical) be interpreted as meaning that
additional Tier 1 debt instruments must be subordinated to Tier 2 debt (which
does not present any problems under English law). The problem comes from
the reference to “subordinated debt” as any additional Tier 1 debt will be
subordinated debt and an instrument cannot logically be subordinated to
itself.

3.3.3 The detailed requirements for additional Tier 1 instruments are as follows:

it is issued and paid up74;

it is subordinated to depositors, general creditors and subordinated debt

holders. For a holding company, subordination to all general creditors is
required (which, we assume, means senior creditors);

it is neither secured nor covered by a guarantee of the issuer or a related

company, or other arrangement, that economically or legally enhances
its seniority;

the instrument is perpetual75;

there are no step-ups or incentives on the issuer to redeem 76;

an issuer call may only be made after five years 77 and is subject to prior
supervisory consent;

a bank must not do anything which creates an expectation that a call will
be exercised;

a bank must not exercise a call unless either the instrument is replaced
with capital of the same or better quality, or the bank demonstrates that
its capital position is well above the minimum capital requirements
(which may at national discretion be higher than the Basel III minima)
after exercise of the call;

tax and regulatory calls 78 are permitted during the first five years after
issuance with supervisory approval provided the bank could not
anticipate such event occurring at the time of issuance;

any repurchase or redemption of the instrument is subject to prior

supervisory consent, and the bank must not assume or create a market
expectation that approval will be given;

all distributions must be fully discretionary, and a decision to cancel

payment of a coupon must not impose any restrictions on the bank except
in relation to distributions to ordinary shareholders (i.e. a dividend
stopper is, in principle, acceptable – but see the discussion below in
respect of not hindering recapitalisation). A note states that dividend

 Back to contents 21
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

pushers are prohibited (i.e. a clause providing that if a distribution is

made on a junior security or pari passu security within a previous period
of time then the coupon/distribution payable on the senior instrument
must be paid). Additionally, alternative coupon satisfaction mechanisms
(which require the issuer to replace a cancelled coupon with newly issued
ordinary shares) are prohibited, as are options allowing holders to
convert the instrument on cancellation into equity. An FAQ prohibits
optional dividends, with prior supervisory approval, equal to the
aggregate unpaid amount of prior dividends;

dividends/coupons must be paid out of distributable items. A note adds

that where (as in England) payment of coupons on debt is an expense
item this requirement means that “such distributions … should not be
allowed by the regulator if the distributable items are not adequate to
provide for them”79. Two points may be made on this. Firstly, it requires
the calculation of distributable reserves in accordance with national
company law, and the treatment of additional Tier 1 instruments, as if
they were share capital out of which distributions may be made.
Secondly, the note appears to require that the supervisor have the power
to cancel distributions. To be legally effective, such a requirement would
need to be reflected in the terms and conditions of the securities on
issue. An alternative could be to provide in the terms and conditions of
the securities that a coupon is automatically cancelled in the event that
if the securities constituted share capital on the relevant coupon
payment date the issuer would have insufficient distributable reserves to
pay a dividend on that coupon payment date (or provision to similar
effect);

dividends/coupons that are reset periodically based on the credit rating

of the issuer (i.e. a margin ratchet) are prohibited 80;

the instrument cannot contribute to liabilities exceeding assets if a

balance sheet test forms part of national insolvency law. In England, all
debt instruments are, in principle, capable of having such an effect,
although it is possible to draft the terms of an instrument so that the
amount repayable in a winding up is automatically reduced pro tanto
with the amount available to repay senior creditors. We would add that
under English insolvency law, assets exceeding liabilities constitutes
grounds for a court to make a winding up order, but it does not follow
that it will. The Supreme Court considered it “very far from an exact
test, and the burden of proof must be on the party which asserts balance-
sheet insolvency”81;

neither the bank nor a related party over which the bank exercises
control or significant influence must have purchased the instrument, or
funded its purchase82;

the instrument must not have any features which hinder recapitalisation.
An example of such an instrument given in the Basel III text is where
compensation must be paid to investors if a new instrument is issued at
a lower price during a specified timeframe. The EU considered that
dividend stoppers hindered recapitalisation when adopting the Capital
Requirements Regulation (EU) No. 575/2013. This view is not mandated

 Back to contents 22
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

under Basel III (as seen above), although the question remains open
whether the application of a dividend stopper could in practice hinder
recapitalisation, and if it did whether this is compatible with Basel III. It
seems clear that a dividend stopper could potentially hinder
recapitalisation in that it may make the issue of new common equity less
attractive in a restructuring scenario. The counterargument is that it is
unfair that new junior creditors should be able to receive distributions
when more senior creditors are unpaid, as they will have acquired their
shares after and knowing of the prior senior issue. As there is no clearly
correct solution, it is likely to be a policy choice made by national
supervisors. An FAQ published by the Basel Committee states “[d]ividend
stoppers that stop dividend payments on common shares are not
prohibited by the Basel standards. Furthermore, dividend stoppers that
stop dividend payments on other Additional Tier 1 instruments are not
prohibited. However, stoppers must not impede the full discretion that
a bank must have at all times to cancel distributions/payments on the
Additional Tier 1 instrument, nor must they act in a way that could hinder
the recapitalisation of the bank”. Examples of prohibited stoppers on
additional Tier 1 instruments in FAQ 9 are: (1) attempts to stop payment
on another instrument where payments on this other instrument were
not also fully discretionary; (2) to prevent distributions to shareholders
for a period that extends beyond the point in time that
dividends/coupons on the additional Tier 1 instrument are resumed; and
(3) to impede the normal operation of the bank or any restructuring
activity (including acquisitions or disposals). The first two examples
given are not market practice and are intended to coerce the bank into
making payments on the securities in breach of the requirement that they
be wholly discretionary. However, a restriction on acquisitions where the
bank is not paying coupons has traditionally been viewed as acceptable,
and does not immediately seem related to recapitalisation as acquisitions
usually involve a reduction in cash (unless made for non-cash
consideration). The FAQ would seem to suggest ordinary dividend
stoppers are acceptable even if they have the practical effect of
hindering recapitalisation; and

if the instrument is issued by a special purpose vehicle (“SPV”) then the

proceeds must be immediately available without limitation to a single
operating company, or the holding company, in a form which meets or
exceeds the other criteria for inclusion in addition Tier 1 capital. This
covers the situation where capital is issued by an SPV and then
transferred to the bank (or its holding company). Traditionally, prior to
2010, the intra-group transfer had been made using Tier 2 or other debt.
Basel III requires that the intra-group transaction takes the form of the
same or better quality capital, although in the latter case it will not be
recognised as higher quality capital for the bank 83 i.e. additional Tier 1
capital that is used to subscribed for ordinary shares.

Principal loss absorbency

3.3.4 Under Basel III, if an instrument is classified as a liability for accounting

purposes (which most straight debt and some preference shares are) it must
have a principal loss absorbency mechanism. Two alternatives are available:

 Back to contents 23
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

conversion to common shares at an objective pre-specified trigger point

of at least 5.125% of common equity Tier 1. This means that the
instruments must convert into common equity if the consolidated capital
ratio of the banking group, or a subsidiary bank subject to Basel III falls
below 5.125%; or

a write-down mechanism which allocates losses to the instrument at a

predefined trigger point of at least 5.125%. The write-down must have
the following effects:

reduce the claim of the instrument in the liquidation of the issuer;

reduce the amount repaid when a call is exercised; and

partially or fully reduce coupon/dividend payments on the

instrument84.

3.3.5 It is implicit in the above requirements that the write-down is permanent, and
that it is not possible for the issuer to write back up the instrument out of
future profits.

3.3.6 Non-redeemable preference shares with a discretionary coupon are equity

accounted under IAS 32 with the result that they do not need to include a
write-down feature under Basel III (preference shares with a mandatory
coupon are not eligible as Tier 1 capital and, at best, may be treated as Tier
2 capital).

3.3.7 Principal “loss absorbency” is required to be achieved through conversion or

write down. The idea is that when an instrument is written down the
reduction of the principal amount will give rise to a non- distributable reserve
that will be capable of absorbing losses. Under English company law there
are restrictions on the write down of the nominal amount of preference shares
in a public limited company (as a reduction in share capital by such a company
may only be done if sanctioned by the court). However, it may be possible to
achieve a similar outcome e.g. by conversion and re-denomination.
Conversion to equity, on the other hand, is well-established, although
resolutions may be required by the company in general meeting to approve
an increase in share capital, grant authority to directors to allot shares and
to disapply pre-emption rights.

3.3.8 Although the minimum trigger is 5.125%, current market practice is to view
such a low common equity Tier 1 ratio as equivalent to being a gone concern.
Banks may therefore choose (or be required by their supervisor, as in
Switzerland) to set a higher trigger to ensure the instruments support
recapitalisation in a crisis.

3.3.9 It should be noted that this requirement is distinct from the separate
requirement for write-down or conversion at the point of non-viability that
applies also to Tier 2 capital (common equity Tier 1 is exempt due to its
subordinated status under company law in all jurisdictions).

 Back to contents 24
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

Write-down at the point of non-viability

3.3.10 Basel III requires all non-core equity Tier 1 capital instruments (i.e. additional
Tier 1 and Tier 2 instruments) to provide for their write-down or conversion
at the point of non-viability. This reflects the observation that in the 2008-
2009 financial crisis such instruments did not bear losses as expected. Where
financial institutions were rescued by governments through equity capital
injections, insolvency was averted. However, as Tier 1 and Tier 2 capital
instruments are senior to equity they benefited directly from the new equity
provided by taxpayers as opposed to bearing losses.

3.3.11 Moreover, as such instruments invariably contained before 2010 a dividend

stopper on ordinary shares if coupons were not paid, holders often continued
to receive distributions as well as avoid losses. The existence of such stoppers
can, as has been discussed, be a hindrance to recapitalisation as private sector
capital providers may be unlikely to invest in share capital if there is no
reasonable prospect of receiving a dividend in the medium term because of
the continuing obligation to make distributions on higher ranking capital
instruments. Whether this should matter is, as we have seen, largely a
question of supervisory policy.

3.3.12 As a result, the Basel Committee announced on 13 January 2011 that:

“The terms and conditions of all non-common Tier 1 and Tier 2 instruments
issued by an internationally active bank must have a provision that requires
such instruments, at the option of the relevant authority, to either be written
off or converted into common equity”.

3.3.13 These requirements are now set out in the consolidated Basel III text. For
additional Tier 1 capital “[t]he terms and conditions must have a provision
that requires, at the option of the relevant authority, the instrument to either
be written off or converted into common equity upon the occurrence of a
trigger event”85. Any compensation must be immediately paid in the form of
common stock (or equivalent for mutuals, etc.) in the bank or its parent
company (including any successor in resolution), and must be paid before any
public sector capital contribution. According to the standard, the bank must
at all times maintain prior authorisation to immediately issue the relevant
number of shares required should the trigger event occur. Under English law,
authorisation to issue new equity share capital (and related matters) requires
consent of the members of the company in a general meeting (which, in the
usual case of a publicly listed holding company of a banking group, where the
new shares are issued by the ultimate holding company86 is not a matter of
discretion for the group but of a shareholders’ vote). Any write-down must
be permanent87.

3.3.14 The trigger is the earlier of:

a decision that write-off, without which the bank would become non-
viable, is necessary, as determined by the relevant authority; and

the decision to make a public sector injection of capital, or equivalent

support, without which the bank would have become non-viable.

 Back to contents 25
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

3.3.15 Where a bank is part of a wider banking group, and the capital instrument is
intended to be included in consolidated group capital the trigger is the earlier
of:

a decision that a write-off is necessary, as determined by the relevant

authority in the jurisdiction of incorporation of the bank (the “home
jurisdiction”); and

the decision to make a public sector injection of capital, or equivalent

support, in the jurisdiction of the consolidated supervisor.

3.3.16 This requirement is curious as it ignores the position where the group
supervisor considers that the group is non-viable but there is no public sector
injection of capital88. In this case, capital issued by the bank and its holding
company in the consolidated supervisor’s jurisdiction would be subject to
conversion/write-off, but any capital instruments issued by subsidiary banks
in other jurisdictions would not unless the subsidiary bank were itself subject
to a determination of non-viability. If only the parent bank/holding company
is non-viable then the securities issued by subsidiary banks subject to
consolidated supervision would not convert or be written off. Presumably,
this is because such institutions are still (in the view of their supervisors)
viable.

3.3.17 Moreover, the test of “non-viability” seems to be inherently subjective. Non

viability cannot be equated with insolvency on either a cash flow or a balance-
sheet basis as a bank or other entity may be solvent (as Lehman Brothers
Europe was ultimately proved to be) but regulators may still determine that
such a firm is non-viable. As this test is undefined89, it is ultimately a question
of discretion by the relevant supervisor, bounded by judicial or other legal
review. In England, such a decision could only be challenged for bad faith,
procedural impropriety or irrationality. Whether supervisors should have such
a wide discretion to resolve a bank or its group may be open to debate.

3.3.18 An exception to the requirements for write-down or conversion exists if the

governing jurisdiction of the bank has in place laws that (i) require such
instruments to be written off upon such event, or (ii) otherwise require such
instruments to fully absorb losses before taxpayers are exposed to loss.
Moreover, the requirements must be disclosed by the relevant regulator and
by the issuing bank in issuance documents issued on or after 1 January 2013 90.
An earlier requirement that a peer group review confirms that the jurisdiction
conforms to these requirements has been deleted in the consolidated Basel III
text.

3.3.19 The UK has passed legislation to facilitate the resolution of banks incorporated
in the UK under the Banking Act 2009. As amended, this Act provides very
extensive discretionary powers to effect the resolution of UK banks91 as well
as two new insolvency procedures for banks in financial difficulties. The
intention is to provide the Treasury and the Bank of England with a wide range
of tools to deal with failing banks.

3.3.20 The Act provides for the “bail-in” of regulatory capital instruments as well as
other resolution powers including:

 Back to contents 26
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

the transfer of all or part of a bank to a private sector purchaser (“PSP”);

the transfer of all or part of a bank to a “bridge bank” owned by the Bank
of England; and

the transfer of a bank, or a bank holding company, into temporary public

ownership (“TPO”).

3.3.21 The exception referred to is accordingly applicable to banks incorporated in

England.

3.4 Tier 2 Capital

3.4.1 Tier 2 capital consists of the sum of the following elements:

instruments that meet the criteria for inclusion in Tier 2 capital;

share premium (if any) on Tier 2 instruments92;

instruments issued by consolidated subsidiaries of the bank that meet the

requirements for inclusion in group consolidated capital as minority
interests;

certain loan-loss provisions (see below); less

regulatory adjustments applied in the calculation of Tier 2 capital 93.

3.4.2 The loan-loss provisions eligible for inclusion in Tier 2 capital differ depending
on whether the bank applies the standardised or IRB approach to credit risk:

under the standardised approach, provisions held against future

unidentified losses are eligible. However, provisions ascribed to
deterioration of particular assets or known liabilities are not.
Recognition of eligible reserves/provisions is restricted to 1.25% of credit
risk-weighted assets, gross of tax effects94; and

under the IRB approach, where the total expected loss amount is less
than total eligible provisions, the difference may be recognised as Tier 2
capital up to 0.6% of credit risk-weighted assets (or a lower percentage
at national discretion)95.

3.4.3 Basel III states that “[t]he objective of Tier 2 is to provide loss absorption on
a gone-concern basis”96. Where issued as debt instruments or bonds97 the
following criteria must be met:

they are issued and paid in;

they are subordinated to depositors and general creditors of the bank.

Where issued out of a holding company, subordination applies to all
general creditors98;

 Back to contents 27
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

they are neither covered by a guarantee of the issuer or a related entity,

or other arrangement that enhances their seniority, either legally or
economically;

the minimum maturity is five years. Recognition is subject to

amortisation in the five years prior to maturity on a straight line basis.
This means that the instruments’ contribution to Tier 2 capital is reduced
by 20% per year reaching 0% in the final year;

there are no step-ups or other incentives to redeem the instruments;

an issuer call is permitted only after five years. All calls are subject to
supervisory approval, and the bank must not do anything that creates an
expectation that the call will be exercised;

a bank may only exercise a call option if the capital instrument is

replaced by new capital of the same or better quality at conditions that
are sustainable for the income capacity of the issuer, or the bank
demonstrates that its capital position is well above its minimum capital
requirement;

regulatory and tax calls are permitted within the first five years;

investors have no right to accelerate future payments except in

bankruptcy or liquidation;

credit-sensitive dividend/coupon features are not allowed;

neither the bank nor a related party over which the bank exercises
control or significant influence purchases the instruments. Nor may the
bank directly or indirectly fund the instrument or its purchase;

if issued out of an SPV, the proceeds must be immediately available to a

single operating entity, or the holding company, in a form which meets
or exceeds the other criteria for Tier 2 capital; and

the terms and conditions include a provision that requires, at the option
of the relevant supervisor, the instrument to be written off or converted
into common equity, unless the law of the governing jurisdiction of the
bank enables this to be done99. We refer to the discussion on additional
Tier 1 instruments above for such principal loss absorbency, including
group level recognition. As mentioned, England has such a statutory
scheme.

Components of regulatory capital

Tier 1 Common Sum of common shares CET ≥4.5%
(going Equity (equivalent for non-joint
concern) Tier 1 stock companies) and
(CET1) stock surplus, retained
earnings, other
comprehensive income,
qualifying minority

 Back to contents 28
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

interests and regulatory

adjustments
Additional Sum of capital CET + AT1 ≥6%
Tier 1 instruments meeting the
(AT1) criteria for AT1 and
related surplus,
additional qualifying
minority interests and
regulatory adjustments
Tier 2 Sum of capital CET + AT1 +
(gone instruments meeting the Tier 2 ≥8%
concern) criteria for Tier 2 and
related surplus,
additional qualifying
minority interests,
qualifying loan loss
provisions and regulatory
adjustments

3.5 Minority Interests

3.5.1 Minority (third party) interests arising from the issue of common shares by a
fully consolidated subsidiary of the bank may receive recognition as common
equity Tier 1 if:

the instrument meets the criteria for ordinary shares; and

the issuer is a bank or other institution subject to the same minimum

prudential standards as a bank100. Historically, this was the case for
investment firms in both the UK and EU, although the PRA (and the EU)
accept that a different regime is now appropriate for all but the largest
or most significant investment firms.

3.5.2 The amount of minority interests recognised is a proportion of the amount of

genuine third party interests in the common equity of the subsidiary. The
calculation is complicated and is based on the amount of the surplus common
equity Tier 1 capital attributable to third party minority shareholders101.

3.5.3 A similar set of requirements applies to the attribution to group total Tier 1
capital and total capital provided by third party minority interests. A worked
example is provided by the Basel Committee in CAP 99.

3.5.4 As we have seen, indirect issues of capital by an SPV may be included as

additional Tier 1102 and Tier 2 capital103. This includes capital issued by a solo
consolidated SPV104.

3.6 Deductions from Capital

3.6.1 When it was adopted in 2010 Basel III introduced a radical overhaul to the
former deductions from capital that had been unchanged since 1988.

 Back to contents 29
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

Goodwill and other intangibles

3.6.2 Goodwill and other intangibles are deducted in full from common equity Tier
1 capital. The deduction is net of any associated deferred tax liability which
would be extinguished if the intangible asset was impaired or derecognised
under relevant accounting standards 105. One effect of this rule is that banking
groups that grow by acquisitions do not have a capital advantage over groups
that grow organically. Mortgage service rights are exempted from this
treatment and instead subject to the “threshold” deduction regime set out
below. With prior supervisory approval, banks using local GAAP may use the
IFRS definition of “intangible assets” to determine which assets are classified
as intangible and subject to deduction 106. The intention is to prevent
differences between local GAAP and IFRS affecting the deductions that banks
are required to make. If crypto-assets are accounted for as an intangible asset
then they must currently be deducted from capital. The Basel Committee
intends to delink the prudential treatment of crypto-asset exposures from the
accounting treatment when the new prudential standard for such assets
comes into force.

Deferred tax assets (“DTAs”) and liabilities

3.6.3 The treatment of DTAs and related liabilities is complex with three different
approaches.

3.6.4 DTAs that rely on future profitability to be realised (e.g. operating losses
carried forward, unused tax losses and unused tax credits) are deducted from
common equity107. Such DTAs can only be realised through a reduction in
future tax payments if the bank makes a profit in the future. Because of the
uncertainty of future profits the Basel Committee considers that reliance on
such assets as a reserve is not appropriate. Moreover, such DTAs provide no
protection to depositors or government insurance funds if a bank fails or
becomes insolvent.

3.6.5 Deferred tax liabilities may be netted provided that they relate to taxes levied
by the same tax authority and offsetting is permitted by that authority108.

3.6.6 Deferred tax liabilities to be netted against DTAs exclude amounts netted
against the deduction for goodwill, other intangibles and defined benefit
pension assets. Deferred tax liabilities must be allocated pro rata between
DTAs subject to deduction from common equity and those DTAs subject to
threshold deductions (see below).

3.6.7 A separate treatment applies to DTAs arising from “temporary differences”

which are subject to the threshold deductions approach referred to below.
Basel III originally gave, as an example, allowances for credit losses, although
this reference has now been removed.

3.6.8 DTAs arising from temporary differences that, under national law, are
automatically transformed into a tax credit in case a bank is not profitable,
is liquidated or placed under insolvency proceedings, and where the tax credit
is lower than the tax liability, are fully refunded attract a 100% risk weight 109.

 Back to contents 30
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

3.6.9 DTAs that are a claim on a tax authority (i.e. prepayments of tax or tax
receivables) are risk- weighted as a claim on the relevant government on the
basis that such claims represent sovereign risk. This includes over-instalments
of tax and, where so provided under local law, current year tax losses that are
a receivable from the government or tax authority110. Such DTAs are not
deducted as they are a debt owed to the bank as opposed to a contingent
right to offset losses against future profits.

Cash flow hedge reserve

3.6.10 A cash flow hedge reserve that relates to the hedging of items not fair valued
on the balance sheet (including projected cash flows) is excluded from
common equity Tier 1. Positive items are therefore deducted while negative
amounts are added back111. The reason is that the reserve reflects the fair
value of the derivative entered into but not changes in the fair value of the
hedged future cash flow and therefore reflects only one half of the picture,
thereby generating artificial volatility in common equity 112.

Shortfall in provisions

3.6.11 Unsurprisingly, shortfalls in provisions for expected losses must be deducted113.

This is a treatment allowed under the internal ratings-based (IRB) approach
followed by banks with relevant supervisory approval. Standardised banks will
recognise such losses through the profit and loss account.

Gain on sale of certain securitisation transactions

3.6.12 Any increase in equity capital resulting from securitisations (e.g. from
expected future margin income) must be deducted from capital114.

Gains and losses due to changes in own credit risk

3.6.13 All unrealised gains and losses that result from changes in the fair value of
liabilities that are due to changes in the bank’s own credit risk are excluded
from regulatory capital. This applies also to accounting adjustments arising
from changes in the bank’s own credit risk on derivative instruments.
Offsetting of valuation adjustments between a bank’s own credit risk and that
of its counterparties is disallowed 115. In each case, there is no change in the
amount of common equity capable of absorbing losses, so any notional gain or
loss is disregarded.

Pension fund assets and liabilities

3.6.14 Defined benefit pension fund liabilities are fully deducted from the
calculation of common equity. The effect is that any pension deficits are
deducted from regulatory capital. Defined benefit pension fund assets
recognised on the balance sheet are also deducted from common equity net
of any associated deferred tax liability which would be extinguished if the
asset became impaired or derecognised under applicable accounting
standards116. The reason is that such assets may not be capable of being
withdrawn and used to pay depositors and creditors, and are only of value in
reducing future payments into the fund 117. However, if the bank can satisfy
its supervisor that it has unrestricted and unfettered access to surplus assets

 Back to contents 31
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

in the fund then the bank may offset the deduction with such assets. Such
offsetting assets are risk-weighted as if they were directly owned by the
bank118.

Investment in own shares

3.6.15 All investments in own shares recognised on the balance sheet are deducted
from common equity, whether directly or indirectly held 119. This applies
regardless of whether the position is held in the trading or banking book. The
purpose of this deduction is to avoid the double counting of a bank’s capital.
It follows that if the applicable accounting regime does not recognise treasury
shares as an asset then this deduction is not necessary120. Similarly, any shares
which a bank could be contractually obliged to purchase (e.g. due to an
investor call) are also deducted121. Gross long positions may only be netted
against short positions if the short position involves no counterparty risk.
Indirect holdings held through a position in an index are also deducted. Long
and short positions may be netted, but if there is counterparty risk on the
short positions, the relevant counterparty credit risk capital charge applies 122.
Such deductions are entirely logical as a bank’s holdings of own shares
provides no protection to depositors or creditors if the bank fails.

3.6.16 Banks’ investments in their own non-core Tier 1 and Tier 2 capital is also
deducted. In accordance with the “corresponding deduction” approach (see
below), non-core Tier 1 instruments must be deducted from that tier of capital
and own holdings of Tier 2 capital from total Tier 2 capital123. If a deduction
is required to be made from a particular tier of capital, and it does not hold
enough of that tier of capital, the deduction is made at the next higher tier
of capital124.

3.6.17 G-SIBs must deduct own holdings of their TLAC instruments. If there are
insufficient TLAC instruments, the deduction is made from Tier 2 capital 125.

Reciprocal cross-holdings

3.6.18 Reciprocal cross-holdings of capital that are designed to artificially inflate the
capital position of banks are deducted. Such a holding may arise if Bank A
invests in the capital of Bank B, and Bank B roundtrips the money by making
an investment in Bank A’s capital. In such a case there is no increase in the
capital held in the banking system. However, the deduction is not confined
to banks and also applies to other financial institutions 126 and insurance
companies. A corresponding deduction approach applies, with equity holdings
being deducted from common equity. Reciprocal holdings of TLAC held by G-
SIBs are deducted from Tier 2 capital127.

Unconsolidated investments in capital instruments or TLAC issued by banks, financial

institutions and insurance companies

3.6.19 Any significant investments by a bank in the capital or TLAC liabilities of

banking, financial or insurance entities that are not consolidated are generally
deducted. Different rules apply depending on whether the bank owns less, or
more, than 10% of the issued common share capital of the entity. No
deduction applies to investments in consolidated affiliates as, by definition,

 Back to contents 32
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

such investments will be netted out on consolidation and therefore

disregarded when looking at the consolidated capital position.

3.6.20 If the bank does not own more than 10% of the ordinary shares then they must
be deducted subject to a threshold. Different thresholds apply depending on
whether the bank is a G-SIB, and whether those investments are in TLAC or
not. For holdings above 10% see the following section on material holdings.

3.6.21 If the bank is not a G-SIB then the threshold is 10% of the investing bank’s
common equity. Holdings in aggregate equal to 10% or less of the bank’s
common equity (after deductions) are ignored. It follows that if the holdings
amount to 14% then only 4% will be required to be deducted. A “corresponding
deduction” approach applies, with common equity Tier 1 instruments
deducted from common equity Tier 1, additional Tier 1 instruments from total
Tier 1 and Tier 2 instruments from total capital128. The original 2010 Basel III
standard explained as follows:

“the amount to be deducted from common equity should be calculated as the

total of all holdings which in aggregate exceed 10% of the bank’s common
equity … multiplied by the common equity holdings as a percentage of the
total capital holdings. This would result in a common equity deduction which
corresponds to the proportion of total capital holdings held in common equity.
Similarly, the amount to be deducted from Additional Tier 1 capital should be
calculated as the total of all holdings which in aggregate exceed 10% of the
bank's common equity (as per above) multiplied by the Additional Tier 1
capital holdings as a percentage of the total capital holdings”.

3.6.22 TLAC holdings are treated differently. For non-G-SIBs, TLAC holdings equal in
aggregate to less than 5% of the common equity of the investing bank (after
deductions) are ignored. Holdings between 5% and below 10% of common
equity are deducted from Tier 2 capital 129. From 10% (after deductions) the
deduction must be made from total Tier 1 or total capital130.

3.6.23 For G-SIBs, investments in TLAC liabilities may be ignored if: (1) the holding
has been designated by the bank, (2) it is in the trading book, (3) the holding
is sold within 30 business days and (4) the aggregate amount of all holdings on
a gross long basis are less than the G-SIB’s common equity131.

3.6.24 Where a holding designated under the preceding paragraph ceases to meet
the criteria listed above then it must be deducted in full from Tier 2 capital.
Any designated holdings cannot be included within the 10% threshold. The
stated reason is to ensure deep and liquid markets in TLAC instruments 132.

3.6.25 Holdings by a G-SIB not subject to the preceding two paragraphs which exceed
10% of the investing G-SIB’s common equity (after deductions) are subject to
deduction from either Tier 1 or Tier 2 capital 133 (as applicable under the
corresponding deduction approach with TLAC holdings deducted from Tier 2
capital).

3.6.26 Where a bank, subject to the above rules, does not have enough of a particular
tier of capital to make the deduction, it must be deducted from the next
highest tier of capital134.

 Back to contents 33
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

3.6.27 Amounts not deducted are risk weighted. Instruments held in the trading book
are subject to the market risk rules and investments in the banking book are
treated under the standardised or IRB approach to credit risk (as
appropriate)135.

3.6.28 In calculating whether any investment is subject to the above paragraphs as

being less than 10% of the issued common shares of the banking, financial or
insurance entity which the bank invests in, the following rules apply.
Investments include direct, indirect or synthetic holdings, as elsewhere.
Holdings in both the banking and trading books are aggregated. The
calculation is made based on the net long position (where the maturity of
short positions either matches that of the long, or has a residual maturity of
at least one year). Underwriting positions in capital instruments or TLAC held
for five working days or less are ignored. Any capital instruments in which a
bank invests that does not meet the criteria for common equity, additional
Tier 1 instruments or Tier 2 instruments are treated as common shares. This
seems to require banks to treat as common equity for the application of the
relevant limits any capital instruments (in the sense that they form part of
the capital of the issuer) that are non-compliant with the Basel III
requirements. Finally, national authorities may allow banks to disregard
investments made, with supervisory approval, in the context of resolving or
providing financial assistance to reorganise a distressed institution136.

Material holdings

3.6.29 A material holding is, basically, an investment of 10% or more in the shares,
capital or TLAC of a bank, financial institution or insurance company. The
rules on material holdings are not relevant to consolidated affiliates for the
reasons given in the preceding section. Non-consolidated affiliates are,
however, subject to these rules. An “affiliate” for these purposes is “a
company that controls, or is controlled by, or is under common control with,
the bank”. Control is defined as “(1) ownership, control, or holding a power
to vote 20% or more of a class of voting securities of the company; or (2)
consolidation of the company for financial reporting purposes”137.

3.6.30 All investments in capital instruments above 10% that are common shares are
subject to the treatment set out in respect of “threshold” deductions below138.
Curiously, the effect is to impose a less onerous treatment on individual
holdings of more than 10% in common equity than positions in additional Tier
1 or Tier 2 instruments that are less risky (because they enjoy a higher ranking
on an insolvency). This was one of the compromises reached as part of the
negotiations on Basel III.

3.6.31 All other investments in capital instruments (additional Tier 1, Tier 2) or TLAC
must be fully deducted applying a threshold deduction approach: i.e.
additional Tier 1 is deducted from additional Tier 1 and Tier 2 and TLAC
holdings from Tier 2 capital. If a bank does not have enough of a relevant tier
of capital in which it is required to make a deduction, the deduction is made
at the next higher tier of capital (so a shortfall in additional Tier 1 will result
in a deduction from common equity Tier 1)139.

3.6.32 The rules on calculating the size of holdings are basically the same as for
holdings below the threshold. Thus, direct, indirect and synthetic holdings of

 Back to contents 34
 COMPONENTS OF AND DEDUCTIONS FROM CAPITAL

capital or TLAC are included. For index securities, banks are required to look
through the index and disaggregate its component securities. Holdings in both
the banking and trading book are included, with the net long position being
relevant. Underwriting positions may be excluded if held for five working
days or less. Capital instruments that do not meet the criteria under Basel III
are treated as common equity, and national supervisors may exclude holdings
in the context of the reorganisation of a distressed institution 140.

Threshold deductions

3.6.33 Reference has been made above to “threshold” deductions. Under Basel III
the following may each receive limited recognition when calculating a bank’s
common equity:

significant investments in the common shares of unconsolidated banks,

financial institutions and insurance companies (valued based on their
balance sheet value);

mortgage servicing rights; and

DTAs arising from temporary differences.

3.6.34 The amount of all the three items that remains after application of all
regulatory adjustments must not exceed 15% of common equity calculated
after all regulatory adjustments (i.e. in both cases deductions from capital)141.
A 10% threshold applies to any individual item 142. To determine the maximum
recognition of the specified items the amount of common equity is multiplied
by 17.65% (being the ratio of 15% to 85%). Holdings of any of these items
above the 10% or 15% thresholds are deducted from capital143.

3.6.35 Any holdings not deducted under this section are subject to a 250% risk
weight144.

3.6.36 This treatment was the result of extensive debate within the Basel Committee
during the finalisation of the Basel III standard, and represents a compromise
between those members which wished to see all such assets deducted in full
and those who considered such an approach was not justified.

 Back to contents 35
 THE STANDARDISED APPROACH TO CREDIT RISK

4. THE STANDARDISED APPROACH TO CREDIT RISK

4.1 Standardised Credit Risk approach under Basel III

4.1.1 The standardised approach to credit risk applies to all banks that do not have
regulatory permission to apply an internal ratings-based (“IRB”) approach for
credit risk. Market risk is addressed in a later chapter. The standardised
approach under Basel III is ultimately derived from Basel I, although it is
considerably more sophisticated. Also, the published risk weights are only
indicative, unlike under Basel I or Basel II. However, the basic structure of
the 1988 Capital Accord is retained in the standardised approach, albeit with
many modifications and much greater risk sensitivity. Banks still determine
risk weighted assets by multiplying the size of each exposure by a
counterparty risk weight, subject to adjustment where required. The
intention is that capital charges will closely reflect the credit risk incurred.
However, unlike the internal ratings-based approach, there are a limited
number of risk buckets, and hence different capital charges.

Credit risk, the risk of loss due to a borrower being unable to repay a debt in full or
in part, accounts for the bulk of most banks’ risk-taking activities and regulatory
capital requirements. There are two broad approaches to calculating RWAs for credit
risk: the standardised approach and the internal ratings-based approach.

The Committee’s revisions to the standardised approach for credit risk under Basel III
seek to enhance the regulatory framework by:

 improving its granularity and risk sensitivity. For example, in the revised
standardised approach, mortgage risk weights depend on the loan-to-value (LTV)
ratio of the mortgage;

 reducing mechanistic reliance on credit ratings, by requiring banks to conduct

sufficient due diligence, and by developing a sufficiently granular non-ratings-
based approach for jurisdictions that cannot or do not wish to rely on external
credit ratings; and

 providing the foundation for a revised output floor to internally modelled capital
requirements (to replace the existing Basel I floor) and related disclosure to
enhance comparability across banks and restore a level playing field.

4.1.2 Risk weighted assets are calculated as the product of standardised risk weights
and the exposure amount, net of specific provisions and write-offs145. The
application of the standardised approach differs depending on whether the
home state supervisor allows use of external credit ratings, or not. This is a
result of the 2008-9 financial crisis, where banks’ reliance on credit ratings
was criticised as a result of the poor performance of certain credit ratings and
concerns about rating agency competence. External ratings came under
further challenge during the Eurozone crisis of 2010-2012, which was
interpreted, at the time, as a sovereign debt crisis. As a result, under the
Basel framework, countries are free either to permit, or not, the use of
external credit ratings, but the use of ratings is stated to be no justification

 Back to contents 36
 THE STANDARDISED APPROACH TO CREDIT RISK

for banks not undertaking their own due diligence. In retrospect, it seems
that most of the ratings that gave rise to concerns during the financial crisis
were concentrated in a limited number of sectors: securitisation, re-
securitisation and structured finance146. Sovereign debt is more complex, and
there are undoubtedly many lessons still to learn from the Eurozone crisis.
The ratings assigned to Greek sovereign debt before the crisis were too high,
but no other Eurozone sovereign has yet suffered a debt restructuring.
Perhaps the greatest problem in both crises was not the use of credit ratings
as such but the unthinking reliance on such ratings. The problems of the
Eurozone in 2010-2012 were not properly attributable to rating agency failures
but to structural deficiencies in the construction of economic and monetary
union set out in the Maastricht Treaty (1992) and not addressed in subsequent
EU treaty revisions.

4.1.3 All ratings given below are, following the Basel III text, taken from Standard
and Poor’s. Other rating agencies’ ratings are allowed, and the text states
that “[t]he ratings used throughout this document, therefore, do not express
any preferences or determinations on external assessment institutions by the
Committee”147.

4.1.4 Under Basel III “banks must perform due diligence to ensure that they have
an adequate understanding, at origination and thereafter on a regular basis
(at least annually), of the risk profile and characteristics of their
counterparties. In cases where ratings are used, due diligence is necessary to
assess the risk of the exposure for risk management purposes and whether the
risk weight applied is appropriate and prudent. The sophistication of the due
diligence should be appropriate to the size and complexity of banks’
activities”148. Where banks lend to a member of a corporate group “due
diligence should, to the extent possible, be performed at the solo level to
which there is a credit exposure”149. However, “banks are expected to take
into account the support of the group and the potential for it to be adversely
impacted by problems in the group”150. In reality, most banks reasonably rely
on express or implied support by corporate groups for their operating entities.
This is also recognised by rating agencies. Basel III also required banks to
apply “effective internal policies, processes, systems and controls to ensure
that risk weights are assigned to counterparties” and “be able to demonstrate
to their supervisors that their due diligence analyses are appropriate”151. It
remains to be seen how this will be applied in practice.

4.2 Sovereign Exposures

4.2.1 Basel III sets out two alternative methods for determining the counterparty
risk weight for sovereign exposures. The first method is based on the external
credit rating of the sovereign. Under this method, national supervisors may
adopt a lower risk weight for exposures denominated and funded in the
domestic currency (i.e. the bank has corresponding liabilities denominated in
the domestic currency)152. The counterparty risk weights for sovereigns and
central banks under this method are set out in the table below 153.

 Back to contents 37
 THE STANDARDISED APPROACH TO CREDIT RISK

External AAA to A+ to A- BBB+ to BB+ to B- Below B- Unrated

rating AA- BBB-

Risk 0% 20% 50% 100% 150% 100%

weight

4.2.2 As an alternative, national supervisors may permit banks to use country risk
scores assigned by national export credit agencies (ECAs). The ECAs must
either be those recognised by their supervisor, or the consensus risk scores of
ECAs participating in the OECD Arrangement on Officially Supported Export
Credits154. The risk weights are set out below 155.

ECA risk 0-1 2 3 4 to 6 7

score

Risk weight 0% 20% 50% 100% 150%

4.3 Public Sector Entities

4.3.1 Claims on domestic public sector entities (“PSEs”) are the subject of two
alternative treatments (the choice of which is up to the relevant supervisor
in the jurisdiction concerned)156.

Option 1

Credit AAA to A+ to BBB+ to BB+ to Below B- Unrated

assessment AA- A- BBB- B-
of sovereign

Risk weight 20% 50% 100% 100% 150% 100%

under option 1

The second option bases the risk weight directly on the credit rating of the
PSE.

Option 2

Credit AAA to A+ to BBB+ to BB+ to Below B- Unrated

assessment AA- A- BBB- B-
of PSE

Risk weight 20% 50% 50% 100% 150% 50%

under option 2

4.3.2 Subject to national discretion, PSEs may be treated as sovereign exposures.

Where this discretion is exercised, other national supervisors may allow their
banks to apply the same treatment to their banks’ PSE exposures in that
jurisdiction157. An example given by the Basel Committee of where this may
be appropriate is where PSEs have revenue-raising powers, or are guaranteed
by the central government, although the choice is up to the national
supervisor. In the UK, local authorities do have statutory revenue-raising
powers (through council tax), although they do not benefit from any central
government guarantee of their indebtedness, and traditionally have not been
treated as equivalent to sovereign exposures. The Scottish, Northern Ireland
and Welsh governments have been treated as sovereign exposures.

 Back to contents 38
 THE STANDARDISED APPROACH TO CREDIT RISK

Commercial undertakings that are publicly owned may be treated as other

corporates158, and in practice are in the UK.

4.4 International Organisations

4.4.1 Claims on the Bank for International Settlements (BIS), the International
Monetary Fund (IMF), the European Central Bank (ECB), the European Union,
the European Stability Mechanism (ESM) and the European Financial Stability
Fund (EFSF)159 may receive a 0% risk weight160. A 0% weight also applies to
claims on multi-lateral development banks (MDBs) which the Basel Committee
judges meet specified criteria, including very high-quality issuer ratings,
shareholder structure, shareholder support and lending requirements161.

4.4.2 The current list of 0% risk weighted MDBs is: the World Bank Group
(International Bank for Reconstruction and Development, the International
Finance Corporation, the Multilateral Investment Guarantee Agency and the
International Development Agency), the Asian Development Bank, the African
Development Bank, the European Bank for Reconstruction and Development,
the Inter-American Development Bank, the European Investment Bank, the
Islamic Development Bank, the Council of Europe Development Bank, the
International Finance Facility for Immunization and the Asian Infrastructure
Investment Bank.

4.4.3 Other MDBs (e.g. the Development Bank of Latin America, the Caribbean
Development Bank and the Nordic Development Bank) attract the following
“base” risk weights.

External AAA to A+ to A- BBB+ to BB+ to Below Unrated

rating AA- BBB- B- B-

“Base” 20% 30% 50% 100% 150% 50%

risk
weight

4.4.4 Where a jurisdiction does not allow the use of external ratings, a 50% risk
weighting applies162.

4.5 Bank exposures

4.5.1 Basel III sets out two methods for determining the counterparty risk weight
for exposures to banks depending on whether the use of external credit ratings
is permissible in the relevant jurisdiction 163. The option of basing risk weights
on those applicable to the sovereign of incorporation under Basel II has been
withdrawn.

Option 1

In jurisdictions that permit the use of external credit ratings the following
“base” risk weights apply164.

 Back to contents 39
 THE STANDARDISED APPROACH TO CREDIT RISK

External AAA to A+ to A- BBB+ to BB+ to B- Below B-

rating of AA- BBB-
counterparty
“Base” risk 20% 30% 50% 100% 150%
weight
Risk weight 20% 20% 20% 50% 150%
for short-term
exposures

4.5.2 Banks are expected under Basel III not to mechanistically rely on external
ratings and must perform due diligence to ensure that the external ratings
appropriately reflect the creditworthiness of the bank’s counterparties. If
the due diligence reflects higher risk characteristics than that implied by the
external rating bucket of the exposure, the bank must assign a risk weight at
least one bucket higher than the “base” risk weight. Due diligence analysis
can never result in the application of a lower risk weight than that determined
by the external rating165.

4.5.3 The category for short-term exposures applies to exposures with an original
maturity of three months or less, as well as exposures with an original
maturity of six months or less arising from the movement of goods across
national borders (including relevant off-balance sheet exposures such as self-
liquidating trade-related contingent items)166. Off-balance sheet items are
described below.

4.5.4 Unrated exposures are treated under Option 2 detailed below. This includes
ratings issued by a rating agency not recognised by a national supervisor, or
not chosen to be used by a bank when assigning risk weights. Any “implicit”
support expected to be extended by a government to a particular bank must
be excluded unless it is a public bank owned by its government. Basel III,
however, allows the continued use of external ratings that incorporate
assumptions of implicit government support for five years after the new
standard comes fully into force (which in this case will be until 1 January
2028)167.

Option 2

The second option applies both in jurisdictions that do not permit the use of
external ratings as well as for unrated exposures in other jurisdictions. Banks
must classify their bank exposures into one of three risk-weight buckets: A, B
and C168. The risk weights are given in the table below 169.

Risk weight table for bank exposures

Credit risk Grade A Grade B Grade C
assessment of
counterparty
“Base” risk 40% 75% 150%
weight
Risk weight for 20% 50% 150%
short-term
exposures

 Back to contents 40
 THE STANDARDISED APPROACH TO CREDIT RISK

4.5.5 Exposures to unrated banks may receive a 30% risk weight provided the
counterparty has a common equity Tier 1 ratio which is 14% or higher and a
Tier 1 leverage ratio which is 5% or higher 170.

4.5.6 The allocation of exposures to risk-weight buckets is as follows:

Grade A refers to exposures to banks with adequate capacity to meet

their financial commitments (including payments of principal or interest)
in a timely manner irrespective of economic cycles or business
conditions171. Grade A banks must meet or exceed their minimum capital
requirements (including capital buffers) with the exception of any
adjustments made under Pillar 2 (the supervisory review process) to
increase the minimum requirement. However, if the minimum capital
requirement (and buffers) are not made public, then the bank is assessed
as Grade B or below172. If internal due diligence results in an assessment
that the bank does not meet the definition of Grade A exposures, then
the exposures are placed into Grade B or C 173.

Grade B refers to exposures to banks where the counterparty bank is

subject to substantial credit risk, such as a repayment capability
dependent on stable or favourable economic or business conditions 174.
Grade B banks must meet their minimum regulatory capital requirement.
However, they may not satisfy their capital buffer requirements or
unpublished Pillar 2 adjustments. If this requirement is not met, or the
requirements are not publicly disclosed, the exposures are allocated to
Grade C175. Banks that fail to meet the criteria for Grade A and are not
classified as Grade C exposures fall into Grade B 176.

Grade C refers to higher credit risk exposures where the bank has
material default risks and limited margins of safety. Adverse business,
financial or economic conditions are very likely to lead to an inability to
meet financial commitments177. Specific examples of Grade C banks are
where the bank fails to meet its minimum regulatory capital
requirements (exclusive of buffers) or where the auditors express
substantial doubt whether the bank can continue as a going concern 178.
Even where these criteria are not met a bank may assess its counterparty
to be Grade C179.

4.5.7 Where Option 2 is required to be used then the risk weight of the sovereign
of incorporation of the counterparty may act as a floor to the risk weighting
set out when applying the grades described above. The floor is applicable
where: (i) the exposure is not in the local currency of incorporation or (ii) the
borrowing is booked in a branch of the bank in a foreign jurisdiction and the
exposure is in a currency other than the currency of the jurisdiction where
the branch operates. Curiously, for exposures in euro this does not reflect the
risk arising from the fact that no Eurozone member can control “its” domestic
currency, unlike Japan or the US as the ECB is constitutionally independent of
any EU member state. This floor does not apply to self-liquidating trade-
related contingent items with a maturity below one year 180.

 Back to contents 41
 THE STANDARDISED APPROACH TO CREDIT RISK

4.6 Covered Bonds

4.6.1 Exposures in the form of covered bonds attract a reduced risk weighting to
reflect their lower risk owing to the existence of collateral specifically
available to meet repayments under the bonds. To qualify, the bonds must be
subject by law to special public supervision designed to protect bondholders.
The proceeds must be invested in assets which during the whole period of
validity of the bonds are capable of meeting claims under the bonds, and
which in an insolvency are available on a priority basis to make repayments 181.

4.6.2 The asset pool must be invested in: (1) claims on, or guaranteed by,
sovereigns, central banks, PSEs or MDBs; (2) residential mortgage claims that
meet specified criteria and have a loan-to-value of 80% or lower; (3)
commercial mortgage claims that meet the same criteria and have a loan-to-
value of 60% or less; or (4) claims on, or guaranteed by, banks with a 30% risk-
weight under the standardised approach (subject to a 15% cap on covered
bond issuances)182.

4.6.3 The assets backing the covered bond must exceed the nominal outstanding
value by at least 10%. This need not be a statutory requirement, however,
provided that the bank meets it in practice and makes the requisite
disclosures183.

4.6.4 There are also specified disclosure requirements which will not be set out
here184.

4.6.5 The applicable risk weights for eligible covered bonds are as follows 185. There
is no specific treatment for covered bonds where the jurisdiction does not
recognise the use of external ratings, perhaps because it is not currently
relevant.

4.6.6 In the case of unrated bonds, the issuer’s rating is inferred from the risk
weights in the table below 186.

Risk weight table for rated covered bond exposures

Issue-specific AAA- to A+ to A- BBB+ to BB+ to B- Below B-
rating of the AA- BBB-
covered bond
“Base” risk 10% 20% 20% 50% 100%
weight

Risk weight table for unrated covered bond exposures

Risk weight 20% 30% 40% 50% 75% 100% 150%
of the issuing
bank
“Base” 10% 15% 20% 25% 35% 50% 100%
Risk weight

4.6.7 The same requirement for due diligence applies to inter-bank exposures, with
the possibility of upward revision of risk weights where justified 187.

 Back to contents 42
 THE STANDARDISED APPROACH TO CREDIT RISK

4.7 Exposures to Securities Firms and other Financial Institutions

4.7.1 Exposures to such institutions are treated as inter-bank exposures provided

that such firms are subject to prudential standards and a level of supervision
equivalent to banks (including capital and liquidity standards). The
assessment of this is up to national supervisors, and where such equivalence
is determined to exist, other supervisors may allow their banks to apply the
same treatment. In all other cases, the exposures are treated as corporate
exposures188. It is unclear whether the new prudential framework for
securities firms will meet this requirement, and, traditionally, liquidity has
not been a major feature of such regulation given the assumed liquidity of
trading positions189. A finding of equivalence may be even more challenging
for insurance companies given the different regulatory frameworks world-
wide for insurers.

4.8 Corporate Exposures

4.8.1 The risk weights for claims on corporates (including, most likely, insurance
companies190) depend on whether the exposure is a general corporate exposure
or constitutes “specialised lending” (e.g. project finance). Subordinated debt
and equity holdings are excluded from the corporate asset class and subject
to a bespoke treatment for equity exposures 191.

General corporate exposures

4.8.2 The regulatory capital treatment differs depending on whether the

jurisdiction of incorporation permits the use of external credit ratings.

4.8.3 “Base” risk weightings for jurisdictions that do allow external ratings to be
used are given below192.

External AAA to A+ to BBB+ BB+ to Below Unrated

rating of AA- A- to BBB- BB- BB-
counterparty
“Base” risk 20% 50% 75% 100% 150% 100%
weight

4.8.4 As with other exposure classes, banks must perform due diligence that may
result in the application of less favourable risk weights based on an assessment
of the exposure having higher risk193.

4.8.5 Where the national supervisor does not allow the use of external ratings, then,
as a general matter, a cross-the-board 100% risk weight applies194. A lower risk
weighting may be applied to “investment grade” exposures (as follows)195. An
exposure is treated as investment grade if the issuer has “adequate capacity
to meet its financial commitments in a timely manner and its ability to do so
is assessed to be robust against adverse changes in the economic cycle and
business conditions”196. Investment grade borrowers must have securities
(debt or equity) outstanding on a recognised securities exchange 197. Exposures
to investment grade borrowers attract a 65% risk weight198.

 Back to contents 43
 THE STANDARDISED APPROACH TO CREDIT RISK

Specialised lending

4.8.6 This asset class covers corporate borrowers that possess some or all of the
following characteristics, either in legal form or economic substance:

the exposure is not related to real estate and is either object finance,
project finance or commodities finance;

the exposure is typically to an entity (such as an SPV) specifically created

to finance and/or operate physical assets;

the borrower has few of no other material assets or activities, and

therefore little or no independent capacity to repay apart from income
received from the assets being financed; and

the lender(s) has a substantial degree of control over the assets and the
income generated199.

4.8.7 For these purposes, the following definitions apply:

Project finance is a method of funding in which the lender looks

primarily to the revenues generated by a single project, both as the
source of repayment and as security for the loan.

Object finance refers to the method of funding the acquisition of

equipment (e.g. ships, aircraft, satellites, railcars) where repayment of
the loan is dependent on the cash flows generated by the specific assets
financed and charged to the lender.

Commodities finance refers to short-term lending to finance reserves,

inventories or receivables of commodities traded on an exchange (e.g.
oil, metals or crops), where the loan will be repaid from the proceeds200.

4.8.8 The risk weighting for rated issue-specific ratings (not issuer ratings) is the
same as for general corporate loans provided that the relevant jurisdiction
accepts the use of external ratings201. If no issue-specific rating is available,
or if the relevant jurisdiction does not allow the use of external ratings, the
following risk weights apply:

object and commodities finance exposures are risk-weighted at 100%;

project finance exposures are risk-weighted at 130% during the pre-

operational phase and 100% during the operational phase; and

project finance exposures in the operational phase deemed to be “high

quality” are risk-weighted at 80%202.

4.8.9 Project finance exposures are deemed to be “high quality” if the following
criteria are met:

the borrower is able to meet its financial commitments in a timely

manner and its ability to do so is robust against adverse changes in the
economic cycle and business conditions;

 Back to contents 44
 THE STANDARDISED APPROACH TO CREDIT RISK

the borrower may not act to the detriment of the creditors e.g. by issuing
new debt without the consent of the creditors;

the borrower has sufficient reserves/other arrangements to cover

contingency funding and working capital requirements;

the revenues are availability-based or subject to rate-of-return

regulation;

the borrower’s revenue depends on one main counterparty that is a

sovereign, PSE or corporate with a risk-weight of 80% or lower;

the contractual documentation provides for a high degree of protection

for creditors in a default (e.g. through extensive security);

the main counterparty, or other counterparties, will protect the creditors

if the project is terminated;

all assets and contracts necessary to operate the project have been
charged to the creditors to the extent permitted by applicable law; and

creditors may assume control of the borrower in case of its default 203.

Corporate SME lending

4.8.10 A preferential risk weight is applied to corporate lending to small and medium
sized companies (SMEs). The definition of an SME is sales of €50 million or
less. Corporate SME lending that meets the criteria for retail SME lending is
treated as retail SME lending and risk-weighted at 75%. Rated exposures to
corporate SMEs where the national supervisor allows use of external ratings is
treated as a general corporate exposure. Unrated exposures, where external
ratings are permitted, and all exposures to corporate SMEs where use of
external ratings is not permitted, attract an 85% risk weight 204.

4.9 Subordinated Debt and other Capital Instruments

4.9.1 The treatment of subordinated debt and holdings of capital instruments is a

residuary category where another treatment prescribed by Basel III does not
apply. In practice, this means the following:

where holdings of such instruments must be deducted from capital, no

risk weight applies (to avoid double counting);

if a 250% risk weight applies under the “threshold” deductions approach

considered in the previous chapter, then that risk-weight must be
applied; and

if a 1250% risk weight for such investments applies as a significant holding

in a commercial entity, then that risk-weight applies205.

4.9.2 Any holdings of subordinated debt or other capital instruments, whether

issued by a bank, financial institution, or a corporate are subject to this
treatment. It follows that for non-banks, the reference to capital instruments

 Back to contents 45
 THE STANDARDISED APPROACH TO CREDIT RISK

means the equivalent for the relevant issuer 206. This class also includes
investments in TLAC that are not required to be deducted from regulatory
capital207.

4.9.3 Equity investments in funds are treated as other fund exposures 208.

4.10 Equity Exposures

4.10.1 The definition of “equity” under the standardised approach is based on

economic substance and not legal form, and comprises both direct and
indirect investments. The definition of equity for this purpose is as follows:

it is irredeemable in that a return of the capital investment can only be

achieved on sale or in a liquidation;

it does not embody an obligation on the part of the issuer 209; and

it conveys a residual claim on the assets or income of the issuer 210.

4.10.2 In addition to the above, the following instruments are classified as equity
exposures (regardless of their form):

instruments with the same structure as Tier 1 capital of banks; and

instruments constituting an obligation of the issuer where any of the

following conditions are met:

the issuer may indefinitely defer settlement;

the obligation requires or permits the issuer to settle the obligation

by a fixed number of shares;

the obligation requires or permits the issuer to settle the obligation

by a variable number of shares; or

the holder may require settlement in shares (e.g. convertible

debt), unless the relevant national supervisor is satisfied that the
instrument should be treated as debt 211.

4.10.3 The equity class also includes debt obligations structured with the intent of
conveying the economic substance of equity. On the other hand, instruments
that are legally considered to be equity, but are structured to convey the
economic substance of debt, or a securitisation position, are not treated as
equity212. An example might be a dated must pay preference share.

4.10.4 The risk weight for equity positions as defined above depends on whether the
instrument is considered to be a speculative unlisted equity position or not.
The definition of a speculative unlisted equity position is “equity investments
in unlisted companies that are invested for short-term resale purposes or are
considered venture capital or similar investments which are subject to price
volatility and are acquired in anticipation of significant future capital
gains”213. Long-term investments in unlisted equity of corporate clients, or
debt-equity swaps acquired as a result of a restructuring are excluded 214.

 Back to contents 46
 THE STANDARDISED APPROACH TO CREDIT RISK

4.10.5 Speculative unlisted equity positions are risk-weighted at 400%. Subject to

the next paragraph, all other equity positions attract a 250% risk weight215.

4.10.6 Basel III allows a lower – 100% – risk-weighting where the equity holding is
acquired pursuant to a national programme set out in legislation that provides
significant subsidies to equity investments and involves government oversight
and restrictions on the investments. A 100% risk-weight is applied to holdings
up to an aggregate amount of 10% of the investment bank’s total capital (Tier
1 plus Tier 2)216.

4.10.7 For such investments, the risk-weight is 150%.

4.11 Qualifying Holdings

4.11.1 A qualifying holding is defined as either a significant minority investment in a

commercial (i.e. non-banking or other financial) entity, or a majority
holding/control over such a commercial entity. It is unclear for this purpose
whether significant investments in insurance companies are treated as
financial or, as provided for in other parts of Basel III, corporate exposures.

4.11.2 Where a bank or banking group has such qualifying holdings then the following
two materiality thresholds apply:

15% of the bank’s capital for any individual investment in a single entity;
or

60% of the aggregate of the bank’s capital for all such significant
investments217.

4.11.3 It is implicit that the definition “total capital” means the sum of Tier 1 and
Tier 2 capital of the investing bank.

4.11.4 If either of the thresholds is reached then the following treatment applies.
All investments above the thresholds are subject to a risk weighting of
1250%218. With an 8% minimum capital ratio under Basel III this is the
equivalent of requiring a deduction from capital. All investments below the
thresholds set out above are treated as equity exposures described.

4.12 Retail Exposures

4.12.1 Retail exposures exclude real estate exposures (i.e. retail mortgage
lending)219. They are therefore composed of the following:

Exposures to an individual or individuals; and

SME exposures (retail or corporate) that satisfy the definition of the

“regulatory retail” portfolio (see below)220.

4.12.2 Retail exposures are, in turn, divided into three sub-categories:

The “regulatory retail” portfolio (other than exposures with

“transactors”);

Exposures within the “regulatory retail” portfolio with “transactors”; and

 Back to contents 47
 THE STANDARDISED APPROACH TO CREDIT RISK

All other retail exposures221.

The regulatory retail portfolio

4.12.3 A retail or SME exposure will only fall within the “regulatory retail” portfolio
if all of the following criteria are met:

Product criterion. The exposure takes the form of: (i) revolving credits
and lines of credit (including credit cards, charge cards and overdrafts);
(ii) personal term loans and leases (e.g. instalment loans, auto loans and
leases, student loans, personal finance); and (iii) small business facilities.
Mortgage loans, derivatives, bonds and securities are expressly excluded;

Value. The maximum aggregate exposure to one borrower cannot exceed

€1 million; and

Granularity. No individual obligor may exceed 0.2% of the overall

regulatory retail portfolio222. For this purpose, all loans to an obligor are
aggregated, whether they are “regulatory retail” exposures or not. Thus,
any mortgage loans will be aggregated to determine whether this
threshold is met. Equally, in the case of SME exposures, all lending to
affiliates (such as other group companies) must be aggregated 223.

4.12.4 Exposures to individuals and SMEs that fall within this category and are not
with “transactors” (as to which, see below) receive a risk-weight of 75%224.

Transactors

4.12.5 This category is a subset of the “regulatory retail” portfolio of exposures. All
of the requirements for an exposure to be allocated to the regulatory retail
portfolio set out above must be met. In addition, the bank must classify the
obligor as being a “transactor”. A “transactor” is an obligor of a facility such
as a credit card or charge card where the full balance is repaid at each
scheduled repayment date over the previous 12 months 225. It follows that if
there is a delay of one day in any monthly repayment, or if any part of the
balance is rolled over from one month to another, this definition will not be
met. In the case of overdrafts, a facility is treated as being with a
“transactor” if there has been no drawdown over the previous 12 months 226.
As all of these exposures are revolving, it seems clear that only revolving
exposures are eligible to be included within the specific treatment for
“transactors”.

4.12.6 Exposures to transactors within the regulatory retail portfolio attract a risk
weigh of 45%227.

Other retail exposures

4.12.7 This is a residual class encompassing all retail exposures that do not satisfy
the definition of the “regulatory retail” portfolio, and are not real estate
exposures228. As SME exposures can only be treated as retail if they fall within
the definition of “regulatory retail” exposures, it follows that this class is
confined to exposures to individuals. An example would be an unsecured loan,

 Back to contents 48
 THE STANDARDISED APPROACH TO CREDIT RISK

or a loan secured on securities or other financial instruments or commodities

that exceeds €1 million.

4.12.8 The risk weight is 100%229.

4.13 Real Estate Exposures

4.13.1 As real estate lending was at the heart of the 2007-9 financial crisis one might
have expected major changes to the capital treatment. Although the
treatment under Basel III is considerably more detailed and granular than its
predecessor in Basel II, the main changes relate to the loan-to-value of
mortgage loans. Relatively few changes have been made under the
standardised approach to address the poor quality lending that manifested
itself in the run-up to the crisis.

4.13.2 This class of exposures is sub-divided into the following three sub-portfolios:

the “regulatory real estate” portfolio;

the land acquisition, development and construction (ADC) portfolio; and

other real estate loans230.

4.13.3 A diagram at the end of this section summarises the relevant requirements
and may be of assistance in following the description of the rules in this
section.

4.13.4 Each of the three sub-portfolios will now be considered.

The “regulatory real estate” portfolio

4.13.5 This category seem inspired by the “regulatory retail” portfolio considered
above. Under Basel III it is defined as consisting of:

“regulatory residential real estate” exposures that are not materially

dependent on cash flows generated by the property;

“regulatory residential real estate” exposures that are materially

dependent on cash flows generated by the property;

“regulatory commercial real estate” exposures that are not materially

dependent on cash flows generated by the property; and

“regulatory commercial real estate” exposures that are materially

dependent on cash flows generated by the property 231.

4.13.6 It follows that in establishing the correct risk weight there are two relevant
sub-divisions: one between residential and commercial real estate lending,
and a second by whether or not the exposure is materially dependent on cash
flows generated by the property charged to the lender.

4.13.7 All exposures within this class must satisfy the following criteria to be eligible
as “regulatory” real estate lending:

 Back to contents 49
 THE STANDARDISED APPROACH TO CREDIT RISK

finished property. The property must be fully completed (unless the

loan is secured by forest or agricultural land). National supervisors may
include within this category unfinished property under construction
where: (i) the property is a one-to-four family residential housing unit
that will be the primary residence of the borrower; or (ii) the national
sovereign or a PSE have legal powers and an ability to ensure that the
property is finished232. This seems a curious category as unless all the
units in the building or development have been pre-sold how can the
lender know that the purchasers will all be owner-occupiers? It is also
unclear whether the units must all form part of the same building or can
be separate houses or developments, although the latter would seem
more logical. Why the size of the development is capped at four units is
unclear. The second limb of the exception for uncompleted property
refers to a sovereign or PSE having power to complete the development.
Whilst a sovereign could certainly take powers to complete unfinished
real estate as eminent domain, it seems unlikely that most governments
would seek to engage in such activities. As for PSEs, again it seems
unlikely that local governments would engage in such activity, although
companies such as the US government-sponsored entities Fannie Mae and
Freddie Mac might perhaps do so.

legal enforceability. The mortgage over the property must be legally

enforceable in all relevant jurisdictions, and the lender must be able to
realise its security interest within a reasonable period of time233. As land
is immovable, the reference to “relevant” jurisdictions should, in almost
all cases, mean the jurisdiction where the land is situated and where the
charge can be enforced, although in some cases foreign registration laws
may also be relevant e.g. for commercial secured property lending.

claims over the property. The lender must generally hold a first “lien”234
over the property. However, “in jurisdictions where junior liens provide
the holder with a claim for collateral that is legally enforceable and
constitute an effective credit risk mitigant, junior liens held by a
different bank than the one holding the senior lien may also be
recognised”235. To be recognised “the national frameworks governing
liens should ensure the following: (i) each bank holding a lien on a
property can initiate the sale of the property independently from other
entities holding a lien on the property; and (ii) where the sale of the
property is not carried out by means of a public auction, entities holding
a senior lien take reasonable steps to obtain a fair market value or the
best price that may be obtained in the circumstances when exercising
any power of sale on their own (i.e. it is not possible for the entity holding
the senior lien to sell the property on its own at a discounted value in
detriment of the junior lien”236. Whether these criteria can be satisfied
under English law is unclear. Second (and subsequent) mortgages are
legally enforceable provided that they are registered, and are therefore
an effective risk mitigant to the extent that there is surplus equity after
discharging any senior mortgage. However, the equitable doctrine of
tacking, and the absence of any duty of care owed by senior lenders when
exercising their power of sale to junior lenders 237 may prove problematic.
A further potential difficulty (although without much practical
relevance), is the remedy of foreclosure where the estate of the
mortgagor and any subsequent mortgagees is extinguished, although a

 Back to contents 50
 THE STANDARDISED APPROACH TO CREDIT RISK

junior mortgagee may seek to preserve its security interest though legal
process by obtaining relief from forfeiture if it would thereby recover
some or all of its loan;

ability of the mortgagee to repay. This requires that banks have

underwriting policies with respect to the granting of loans that include
assessment of the ability of the borrower’s ability to repay238. National
supervisors may provide guidance239;

prudent valuation. Specified valuation criteria (described below) must

be met. Additionally, the value of the property must not depend
materially on the performance of the borrower, which is hardly likely to
be the case in respect of retail or normal corporate borrowing owing to
the ability to sell the property; and

required documentation. All the information required at loan

origination and for monitoring purposes must be properly documented,
including on the ability of the borrower to repay 240.

4.13.8 The requirements in respect of the ability of the borrower to repay and
documentation seem quite limited. This seems to us surprising given that a
failure to take account of the former, and the absence of proper
documentation on the borrower’s ability to repay, were major problems in the
origination of sub-prime mortgages in the United States before 2007.

Some definitions

4.13.9 The “regulatory residential real estate” portfolio comprises exposures

“secured by property that has the nature of a dwelling and satisfies all
applicable laws and regulations enabling the property to be occupied for
housing purposes (ie residential property)”241. This definition seems to include
buy-to-let properties as well as owner-occupied properties as in both cases
the property will be used as a residential home.

4.13.10 The “regulatory commercial real estate” portfolio comprises exposures “that
[are] not a regulatory residential real estate exposure”242. It therefore covers
any real estate exposures falling within the regulatory real estate category
that is not a regulatory residential real estate exposure.

4.13.11 The next definition that is relevant is that for exposures materially dependent
on cash flows generated by the property. This is defined as exposures “when
the prospects for servicing the loan materially depend on the cash flows
generated by the property securing the loan rather than on the underlying
capacity of the borrower to service the debt from other sources”243. According
to the Committee “[t]he distinguishing characteristics of these exposures
compared to other regulatory real estate exposures is that both the servicing
of the loan and the prospects for recovery in the event of default depend
materially on the cash flows generated by the property” securing the
exposure244. An example set out in Basel III of where the material dependence
test is met is where more than 50% of the income of the borrower used in a
bank’s assessment of the borrower’s ability to repay is derived from cash flows
generated by the property245. Buy-to-let lending will therefore be subjected
to higher risk weights where a bank considers that more than 50% of the

 Back to contents 51
 THE STANDARDISED APPROACH TO CREDIT RISK

repayments will be dependent on rental income. For possible exceptions to

this treatment see the next paragraph.

4.13.12 Exception to the definition of material dependence set out above are made
in the following cases that are treated as not being materially dependent on
cash flows generated by the property:

exposures secured on a property that is the borrower’s primary

residence;

exposures secured by income-producing residential housing units to an

individual who has mortgaged less than a specified number of properties
or units specified by national supervisors;

exposures secured by residential real estate to associations or co-

operatives regulated under national law with the purpose of granting
members the use of a primary residence; and

exposures secured by residential real estate to public housing companies

and associations that are not run to make a profit regulated under
national law and which exist to serve social purposes 246.

Risk weights for the regulatory real estate portfolio

4.13.13 The risk weights set out in this section are intended by the Basel Committee
to be appropriate for jurisdictions where structural factors result in
sustainably low credit losses. National supervisors must determine whether
these risk weights are too low in their jurisdiction, and may increase them
where appropriate based on an assessment of the risk of such lending 247.

4.13.14 The risk weights are calculated based on the loan-to-value (LTV). This is the
amount of the loan divided by the value of the property 248. While the amount
of the loan will be amortised over the repayment term (except for interest
only mortgages), the value of the property is generally fixed at the time of
origination, preventing the LTV being reduced by increases in property values.
This requirement is clearly inspired by lending practices prior to the financial
crisis where some lenders assumed that very poor credit quality lending, or
undocumented loans, would not lead to losses as property prices would
continue to increase for the duration of the loan or earlier refinancing.

4.13.15 Three exceptions exist where the value of the property may be adjusted 249:

where national supervisors require the value of the property to be revised

downward. In this case, the value may be adjusted upwards subsequently
to an amount not more than the value at origination;

an extraordinary and idiosyncratic event occurs resulting in a permanent

reduction in the property’s value. This could include cases such as the
destruction of the unit or the creation of new transportation
infrastructure (e.g. a motorway or high-speed rail line) near the property
that materially reduces its value; and

 Back to contents 52
 THE STANDARDISED APPROACH TO CREDIT RISK

modifications made to the property that unequivocally increase its value.

An example might be an extension to a residential dwelling, or possibly
planning permission to change the use of the property that results in a
permanent increase in value.

4.13.16 A prescribed treatment applies to calculate the LTV for second and subsequent
mortgages where there is a higher ranking mortgage held by a different
lender250.

4.13.17 The value of the property must be appraised independently using prudent
valuation criteria. National supervisors are tasked with issuing guidance on
valuation251.

4.13.18 The eligibility of credit risk mitigation techniques (other than the mortgage)
to reduce the risk weight for secured loans such as mortgage insurance or
third party-provided guarantees or credit default swaps is described in the
chapter on credit risk mitigation. However, such techniques, even if
recognised, may not be used to reduce the LTV of the loan when calculating
risk weights under Basel III252.

Risk weights for regulatory residential real estate exposures not materially dependent
on cash flows from the property

4.13.19 There are two options at national supervisory discretion: a “whole loan”
approach and a “loan splitting” approach. The former takes the LTV of the
entire loan. The latter approach involves notionally splitting the loan into
different buckets with different risk weights for both segments. Supervisors
will decide the approach taken in their jurisdiction253.

Whole loan approach risk weights

LTV ≤ 50% 50% < 60% < 80% < 90% < > 100%
but ≤ but ≤ but ≤ but ≤
60% 80% 90% 100%
Risk weight 20% 25% 30% 40% 50% 70%

4.13.20 Under the loan-splitting approach the risk weight is calculated as follows.
First, the amount of the loan equal to 55% of the value of the property is
allocated a 20% risk weight. The portion of the loan (if any) above 55% of the
value of the property is risk-weighted at 75% for individuals, 85% for SME
borrowers and treated as unsecured for all other (e.g. corporate) borrowers
and risk-weighted accordingly254. Where there is a senior mortgage then the
calculations must be adjusted to take account of any higher ranking
mortgages255, as must any pari passu ranking mortgages256.

Risk weights for regulatory residential real estate exposures that are materially
dependent on cash flows generated by the property

4.13.21 The risk weights for these exposures are set out in the table below 257.

 Back to contents 53
 THE STANDARDISED APPROACH TO CREDIT RISK

Risk weights for regulatory residential real estate exposures that are
materially dependent on cash flows
LTV ≤ 50% 50% < 60% < 80% < 90% < > 100%
but ≤ but ≤ but ≤ but ≤
60% 80% 90% 100%
Risk weight 30% 35% 45% 60% 75% 105%

4.13.22 For these exposures no loan splitting approach is permitted.

Risk weights for regulatory commercial real estate exposures that are not materially
dependent on cash flows generated by the property

4.13.23 As with residential property, both a “whole loan” and a “loan splitting”
approach are recognised, although the choice will be made by the national
supervisor.

4.13.24 For the “whole loan” approach the risk weights are as follows258.

Whole loan risk weights for regulatory commercial real estate exposures
not materially dependent on cash flows generated by the property
LTV ≤ 60% > 60%
Risk weight Minimum of 60% or risk Risk weight of counterparty
weight of counterparty if
loan were not secured

4.13.25 Under the loan splitting approach the portion of the loan exposure up to 55%
of the value of the property receives the lower of 60% or the risk weight of
the counterparty if the loan were treated as being unsecured. The residual
exposure (if any) receives the risk weight of an unsecured loan 259. Similar
adjustments to those for residential loans are made to reflect any more senior
mortgages260.

Risk weights for regulatory commercial real estate exposures that are materially
dependent on cash flows generated by the property

4.13.26 Only a whole loan approach is available. The table below sets out the risk
weights261.

Risk weights for regulatory commercial real estate exposures that are
materially dependent on cash flows generated by the property
LTV ≤ 60% 60% < but ≤ 80% > 80%
Risk weight 70% 90% 110%

4.13.27 National supervisors may permit banks to apply the risk weights for regulatory
commercial real estate exposures that are not materially dependent on cash
flows generated by the property If the following two conditions are met:

overall losses stemming from commercial real estate lending up to 60%

of the LTV do not exceed 0.3%; and

 Back to contents 54
 THE STANDARDISED APPROACH TO CREDIT RISK

overall losses stemming from commercial real estate lending do not

exceed 0.5% of the outstanding loans in any given year262.

4.13.28 The remaining treatments for real estate lending are more straightforward.

ADC exposures

4.13.29 This class of lending secured by real estate is defined as “loans to companies
or SPVs financing any of the land acquisition for development and construction
purposes, or development and construction of any residential or commercial
property”263. Lending for the acquisition of forests and agricultural land where
there is no planning consent for development, and no intention to apply for
such permission, is not treated as an ADC exposure 264. The ADC risk category
is therefore intended to cover lending to finance large-scale property
development, whether residential or commercial.

4.13.30 The base risk weight for ADC exposures is 150% reflecting their higher credit
risk265. However, a preferential risk weight of 100% may be applied if the
following criteria are met:

the lender applies prudent underwriting standards; and

pre-sale or pre-lease contracts amount to a significant portion of total

contracts or substantial equity at risk 266. The levels will be the subject
of national guidance267.

Other real estate

4.13.31 This is a residual category comprised of secured lending that does not fall
within either the regulatory real estate or the ADC categories268.

4.13.32 The risk weights depend on whether the exposure is materially dependent on
the cash flows generated by the property.

4.13.33 If this is not the case the following risk weights apply:

exposures to individuals are risk weighted at 75%;

exposures to SMEs are risk weighted at 85%; and

exposures to other counterparties (including corporates) are treated as

being unsecured269.

4.13.34 Real estate exposures within this class that are materially dependent on the
cash flows generated by the property receive a cross-the-board 150% risk
weight270.

4.14 Summary

4.14.1 Given the complexity of this section we summarise the correct treatment of
all real estate exposures in the diagram below with references to the relevant
rules.

 Back to contents 55
 THE STANDARDISED APPROACH TO CREDIT RISK

Treatment of Real Estate Exposures

 Back to contents 56
 THE STANDARDISED APPROACH TO CREDIT RISK

4.15 Risk Weight Multiplier for Exposures with a Currency Mismatch

4.15.1 This paragraph applies to certain unhedged retail lending and residential real
estate loans made to individuals where there is a currency mismatch between
the lending currency and the currency of the borrower’s source of income. A
multiplication factor of 1.5 applies to the risk weight, with the result that,
for example, a 100% risk weight becomes 150%. A cap of 150% applies, so
where the relevant risk weight is above 100% the adjusted risk weight is 150%.
The classes of loans covered are: retail exposures and regulatory residential
real estate exposures271. The types of hedges recognised are natural hedges
and financial hedges. In either case, the hedge must cover at least 90% of the
loan instalment272. The reason for the multiplier would seem inspired by the
experience of foreign currency loans during the Eurozone crisis.

4.16 Exposures Giving Rise to Counterparty Credit Risk

4.16.1 Exposures giving rise to counterparty credit risk arising from over-the-counter
derivatives, long-settlement transactions and securities financing transactions
are subject to the treatment set out in the chapter on counterparty credit
risk and not to the rules described in this chapter 273. The sale of credit
derivative is described immediately below.

4.17 Credit Derivatives

4.17.1 Where a bank sells credit protection on a first-to-default or second-to-default

credit derivative the following treatment applies. For first-to-default credit
derivatives, the risk weight of the assets included in the basket is aggregated
up to a maximum of 1250% and multiplied by the nominal amount of the
protection provided by the credit derivative to obtain the risk-weighted
amount. The treatment for second-to-default credit derivatives is similar, but
the asset with the lowest risk-weighted amount is disregarded. The same
procedure is followed for nth-to-default credit derivatives with n-1 assets
being excluded274.

4.18 Defaulted Exposures

4.18.1 Under the standardised approach to credit risk an exposure is regarded as

defaulted if it is past due for more than 90 days, or is an exposure to a
defaulted borrower. A defaulted borrower is one in respect of which any of
the following has occurred:

any material credit obligation is past due for more than 90 days. An
overdraft is treated as defaulted if the customer has breached an advised
limit, or been advised of a lower limit than current outstandings;

any material credit obligation is on non-accrued status;

a write-off or account-specific provision is made as a result of a

significant perceived decline in credit quality;

any credit obligation is sold at a material credit-related economic loss;

 Back to contents 57
 THE STANDARDISED APPROACH TO CREDIT RISK

a distressed restructuring takes place (i.e. a restructuring that may result

in a diminished financial obligation caused by the material forgiveness,
or postponement, of principal, interest or fees);

the borrower is made bankrupt or a similar order has been filed 275;

the borrower has sought or been placed in bankruptcy or similar

protection where this would avoid or delay repayment; or

any other situation occurs where the bank considers that the borrower is
unlikely to pay its credit obligations in full without recourse to actions
such as realising security276.

4.18.2 The definition of default for retail exposures may be applied at the level of
each facility, rather than at the level of the borrower 277. This reflects the fact
that defaults on some facilities (such as credit cards) are much more common
than on others (e.g. mortgage loans).

4.18.3 The risk weights for defaulted exposures are set out below:

defaulted residential real estate exposures not materially dependent on

cash flows generated by the property are risk weighted at 100% net of
any specific provisions or write-offs278;

other exposures where specific provisions are less than 20% of the
outstanding amount of the loan attract a 150% risk weight; and

other exposures where specific provisions are 20% or more of the

outstanding amount of the loan attract a 100% risk weight 279. At national
discretion, this may be reduced to 50% where specific provisions are 50%
or more than the outstanding amount of the loan 280.

4.18.4 The different risk weights based on specific provisions reflect the fact that
provisioning will reduce the aggregate loss as provisions are deducted from
capital through the profit and loss account.

4.19 Other Assets

4.19.1 Other assets that are deducted from capital (including assets above the
“threshold” deduction treatment) are not risk weighted as the effect on
capital has already been recognised. A 250% risk weight applies to assets
within the thresholds of the threshold deduction regime 281.

4.19.2 Any other assets held on a bank’s balance sheet for which a specific capital
treatment has not been designated are risk-weighted at 100%, with the
following exceptions:

cash282 held at the bank or in transit is risk-weighted at 0%;

gold bullion held at the bank or in another bank on an allocated basis 283,
to the extent that such assets are backed 284 by equivalent liabilities285;
and

 Back to contents 58
 THE STANDARDISED APPROACH TO CREDIT RISK

cash items in the process of collection are ascribed a 20% risk weight286.

4.19.3 A right-of-use asset under a lease accounted for as such under applicable
accounting standards is risk-weighted at 100% provided the asset being leased
is tangible (e.g. a computer)287.

4.20 Off-balance Sheet Items

4.20.1 Basel III retains the basic framework of Basel I under which off-balance sheet
liabilities are converted into notional on-balance sheet exposures through
multiplying the gross exposure by a credit conversion factor, although with
more detail than under Basel II. The result is that the nominal off-balance
sheet exposure is multiplied first by the credit conversion factor set out in
this section and then in turn multiplied by the risk weight applicable to the
counterparty. Unless the credit conversion factor is 100% the result will be a
percentage of the risk weight for the counterparty in question. The purpose
of the credit conversion factor is to measure, in a simplified fashion, the
probability of the off-balance sheet exposure becoming an on-balance sheet
exposure before the counterparty defaults.

4.20.2 Commitments are measured as the committed but undrawn exposure amount
(drawn commitments result in an on-balance sheet loan)288. National
supervisors may exempt certain arrangements for corporates and SMEs where
those counterparties are closely monitored on an ongoing basis and the
arrangement confers on the bank full discretion whether or not to advance a
particular borrowing289. Basel III removes the 0% credit conversion factor (i.e.
a 0% risk weight) available for commitments under Basel II for commitments
that are unconditionally cancellable at any time by the bank, or are
automatically cancelled in the event of a deterioration in the counterparty’s
creditworthiness290.

4.20.3 The remaining credit conversion factors are listed from 10% to 100%.

4.20.4 A 100% credit conversion factor applies to the following:

direct credit substitutes. This includes general guarantees of

indebtedness, standby letters of credit serving as financial guarantees
for loans and securities, and acceptances (including endorsements);

sale and repurchase agreements, and asset sales with recourse;

lending of securities or posting of collateral under repo-style

transactions, including securities loans;

forward asset purchases, forward deposits and partly paid securities; and

off-balance sheet items that are credit substitutes not explicitly included
in any of the preceding categories291.

4.20.5 All of the above expose the bank to the full risk of the exposure and are
therefore treated in the same way as a direct loan.

4.20.6 A 50% credit conversion factor applies to the following:

 Back to contents 59
 THE STANDARDISED APPROACH TO CREDIT RISK

note issuance facilities and revolving underwriting facilities, regardless

of the maturity of the underlying facility; and

certain transaction-contingent items (e.g. performance bonds, bid

bonds, warranties and standby letters of credit related to particular
transactions)292.

4.20.7 A 40% credit conversion factor applies to commitments, regardless of the

maturity of the underlying facility, unless a lower percentage is applicable
under the following paragraphs293.

4.20.8 A 20% credit conversion factor applies to issuing and confirming banks of short-
term self-liquidating trade letters of credit arising from the movement of
goods with a maturity of below one year (such as documentary credits
collateralised by the underlying shipment) 294.

4.20.9 A 10% credit conversion factor applies to commitments that are

unconditionally cancellable at any time by the bank without prior notice, or
that effectively provide for an automatic cancellation due to a deterioration
in a borrower’s creditworthiness. Where the national supervisor considers the
bank’s ability to cancel to be constrained it may require a higher credit
conversion factor to be applied295. Under Basel II, this was 0%.

4.20.10 Undertakings to provide a commitment on an off-balance sheet item are

subject to the lower of the two relevant credit conversion factors 296. Thus, a
commitment to open a short-term trade letter of credit attracts a 20% credit
conversion factor and not 40%. Equally, an unconditionally cancellable
commitment to issue a direct credit substitute are subject to a 10% credit
conversion factor and not 100%297.

4.21 External Credit Assessment Institutions (Rating Agencies)

4.21.1 As has been mentioned, national supervisors may recognise or not the use of
external credit ratings to calculate capital requirements. The following
section is only relevant in those jurisdictions that permit this. Basel III
provides that national supervisors are to determine those external credit
assessment institutions (ECAIs) that are eligible to produce credit ratings that
may be used by banks298. Basel III also sets out certain criteria, derived from
Basel II, that must be met by ECAIs before they will be recognised. The
eligibility criteria are as follows:

objectivity. The methodology for assigning ratings must be rigorous,

systematic and subject to some form of validation based on historical
experience;

independence. The ECAI must be independent and should not be subject

to political or economic pressures that may influence the rating;

international access/transparency. The individual ratings, the key

elements underlying the ratings assessment, and whether the issuer
participated in the rating process must be publicly available;

 Back to contents 60
 THE STANDARDISED APPROACH TO CREDIT RISK

disclosure. An ECAI should disclose its code of conduct, compensation

arrangements, conflicts of interest, ratings’ assessment methodology,
default rates, etc.;

resources. An ECAI should have sufficient resources to carry out high-

quality ratings;

credibility. Reliance on ratings by independent parties (investors,

insurers, etc.) is evidence of credibility;

no abuse of unsolicited ratings. ECAIs may not abuse unsolicited ratings

to put pressure on unrated entities to obtain solicited ratings; and

co-operation with the supervisor. ECAIs should notify supervisors of

significant changes to methodologies and provide access to external
ratings and other relevant data299.

4.21.2 The Committee has published further details on disclosure and compensation
arrangements300. Basel III does not require regulation of rating agencies,
unlike the EU.

4.21.3 National supervisors are responsible for mapping the ratings used by ECAIs
recognised by them to ensure consistency with the risk buckets set out in the
text (which are based on Standard & Poor’s ratings)301. When conducting this
process supervisors should assess the size and scope of the pools of issuers
covered, the range and meaning of the ratings and the definition of default 302.
Guidance has been published in the Standardised Approach – Implementing
the Mapping Process (2019)303.

4.21.4 While supervisors will decide which ECAIs may be used in their jurisdiction,
banks may select which of the recognised ECAIs they elect to use subject to
the following criteria. Banks must use the chosen ECAIs and their ratings
consistently for all types of exposure, for both risk weighting and risk
management purposes. Banks are not allowed to “cherry pick” the
assessments provided by different ECAIs or to arbitrarily change the use of
ECAIs304.

Multiple assessments

4.21.5 If there is a single rating for a particular exposure then that assessment must
be used305. If there are two assessments, then the assessment that gives rise
to the higher risk weight is applied306. Where there are three or more ratings,
then the two assessments corresponding to the lowest risk weight will be used.
If they give rise to the same risk weight, then that risk weight will apply. If
they differ, then the rating giving rise to the higher risk weight of the two is
used307.

Issuer-specific and issue-specific ratings

4.21.6 Where an issue-specific rating exists it must be used. If it does not then the
following principles apply:

 Back to contents 61
 THE STANDARDISED APPROACH TO CREDIT RISK

if there is issue-specific rating for an issued debt instrument, and the

bank’s unrated exposure is either senior or pari passu with the rated issue
then the bank may use the rating. Otherwise, the exposure is treated as
unrated;

if there is an issuer rating that applies to senior unsecured exposures, the

bank can use this rating only if its exposure falls within that class. Other
unrated exposures are treated as unrated;

if there is a low-quality issue or issuer rating (i.e. one attracting a higher

risk weight than that applicable to unrated exposures), and the bank’s
unrated exposure ranks pari passu or junior to that exposure, it must be
assigned the same rating;

if there is an issue rating that is high quality (i.e. attracts a lower risk
weight), then a bank may only use the high quality rating if exposures
fall within the relevant class308.

Short term assessments

4.21.7 Detailed rules govern the use of short term ratings (exposures of less than
three months). Such ratings may not be used to assess other short-term
exposures. Nor can a short-term rating be used to support a risk weight for
unrated long-term exposures. Short-term ratings may only be used for short-
term exposures against banks and corporates. The table below “provides a
framework” for banks’ exposure to specific short-term facilities such as
commercial paper309.

Risk weight for specific short-term ratings (S&P, Moody’s)

External A-1/P-1 A-2/P-2 A-3/P-3 Others
rating
Risk 20% 50% 100% 150%
weight

4.21.8 If a short-term rated facility attracts a risk weight of 50%, an unrated short-
term exposure cannot be risk-weighted at less than 100%. If the rating of a
short-term facility is rated 150% all short-term and long-term unrated
exposures are risk-weighted at 150%, although this may be reduced by eligible
credit risk mitigation310.

Corporate groups

4.21.9 An external rating for one entity within a corporate group cannot be used to
risk-weight exposures to other group companies311.

Unsolicited ratings

4.21.10 As a general rule, banks should use solicited ratings. However, national
authorities may allow banks to use unsolicited ratings, if they are satisfied
that unsolicited ratings are not inferior in quality to solicited ratings 312.

 Back to contents 62
 THE STANDARDISED APPROACH TO CREDIT RISK

4.22 Transitional Arrangements for Equity Exposures

4.22.1 The Basel III text include a time-limited transitional provision for equity
exposures under the standardised approach.

4.22.2 Speculative unlisted equity exposures are risk-weighted at 100% in 2023

increasing by 60 percentage points per year until 2027 (i.e. over a 5 year
period).

4.22.3 All other equity exposures are risk-weighted at 100% increasing by 30

percentage points per year until 2027313.

 Back to contents 63
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5. INTERNAL RATINGS-BASED APPROACH TO CREDIT

RISK (IRB)

5.1 Introduction

5.1.1 Allowing banks to use internal ratings to determine their capital requirements
was the core of the Basel II capital accord. As will be seen, the financial crisis
persuaded the Basel Committee to restrict the availability of this approach,
which is detailed in this chapter.

5.1.2 There are generally two IRB approaches: a foundation IRB approach and an
advanced IRB approach. Originally, the advanced IRB approach was expected
to be applied only by a limited number of internationally-active banks,
although that is not what happened. The reason for the restriction on the use
of the advanced IRB approach under Basel III is the fact that many models
performed poorly during the 2007-2009 financial crisis, and there was
significant incommensurability of capital ratios between different banking
groups, as such groups legitimately applied different risk weights to otherwise
identical exposures based on their own data and loss experience.

Outline of IRB

5.1.3 Both the foundation and the advanced IRB approaches are based on six key
principles:

the classification of each exposure by broad exposure type e.g.

sovereign, corporate, retail, etc.;

the identification , within certain exposure classes, of more granular

“asset classes”;

for each exposure type, the bank determines either one or all risk
components using its own internal measurements;

the bank then calculates a continuous risk weight function (which may
vary depending on the type of exposure) and which provides risk weights
(and therefore capital charges) for given sets of these components;

the bank must comply with specified minimum requirements; and

the bank must obtain the prior consent of its supervisor to apply either
the foundation or the advanced IRB approach.

5.1.4 These elements will now be described in more detail. In this chapter it is
assumed that the loan or exposure is not collateralised, and that the bank has
not entered into a netting agreement, and does not benefit from a qualifying
guarantee or credit derivative. The treatment of collateral, on-balance sheet
netting, guarantees and credit derivatives is described in the next chapter on
Credit Risk Mitigation.

 Back to contents 64
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.1.5 The IRB approach distinguishes between exposures in five different portfolios,
and a separate approach applies to each portfolio (although, as mentioned,
individual portfolios may need to be broken down into asset classes). The
portfolios are: (a) corporate exposures, (b) sovereign exposures, (c) inter-
bank exposures, (d) retail exposures, and (e) equity exposures 314. Equity
exposures are, however, ineligible for any IRB treatment and must instead be
risk-weighted under the standardised approach315. The corporate portfolio
also includes five sub-classes of specialised lending: project finance (PF),
object finance (OF), commodities finance (CF), income-producing real estate
(IPRE) and high-volatility commercial real estate (HVCRE)316.

5.1.6 The capital charge for an exposure within an IRB portfolio depends on a set
of risk components 317. These components are intended to reflect the drivers
of credit risk, and cover features that are borrower specific as well as
transaction specific. By taking account of the characteristics of both the
borrower and the transaction, the capital charge under the IRB approach is
intended to be more closely related to the actual risk incurred by the bank,
as well as banks’ economic capital models.

5.1.7 Additionally, Basel III sets out risk-weight functions which are the means by
which risk components are transformed into risk-weighted assets, and
therefore capital requirements 318.

5.1.8 A bank must meet minimum standards in order to be able to use its own
calculations for each risk component319.

5.1.9 For some portfolios Basel III provides both a foundation and an advanced IRB
approach. The difference between the two is that under the foundation IRB
approach banks provide only one of the risk components (the probability of
default or PD) using their internal measurements. The other risk components
are calculated using supervisory estimates set out in Basel III. Under the
advanced IRB approach, banks that meet additional requirements are
permitted to apply their own internal estimates for the other risk components
(loss given default or LGD, exposure at default or EAD, and Maturity or M) 320.

5.1.10 The advanced IRB approach is not available for the following types of
exposures (unlike Basel II):

corporate exposures (other than specialised lending) where the

consolidated annual revenues of the group are greater than €500
million321, based on the consolidated group accounts calculated over a
three year period322;

exposures to banks and other financial institutions within the bank

portfolio, regardless of size323; or

equity exposures324.

5.1.11 The Basel Committee has removed these portfolios from the advanced IRB
approach as it considers that banks were unable to accurately estimate LGD
and EAD for such exposures, based on experience during the financial crisis.

 Back to contents 65
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.1.12 The IRB approach is based on measurements of unexpected losses (UL) only 325.
Expected losses are subject to their own treatment 326.

5.2 Roll Out of IRB across a Given Asset Class

5.2.1 Basel II required banks adopting an IRB approach, in principle, to roll out that
approach across all asset classes. This has been abandoned under Basel III.
Instead, the focus is now on individual asset classes and the rules on roll-out
are firmly focussed on asset classes with no expectation that a bank will roll
out a chosen IRB approach more widely.

5.2.2 An “asset class” is defined more granularily than a portfolio327. The following
asset classes are set out in Basel III:

sovereign exposures;

bank exposures;

corporate exposures (excluding specialised lending and purchased

receivables);

specialised lending;

corporate purchased receivables;

qualifying regulatory residential exposures;

retail residential mortgage exposures;

other retail exposures (excluding purchased receivables); and

retail purchased receivables328.

5.2.3 When a bank adopts an IRB approach for a particular asset class (defined
above) within a business unit it should, in principle, extend it across all
exposures falling within that asset class 329. However, the Basel Committee
recognises that for many banks it may not be practicable to implement an IRB
approach across an entire asset class, and that there may be reasons, such as
data limitations, that prevent a bank from adopting the advanced IRB
approach to all exposures in a particular class, but in different business
units330. Supervisors may therefore allow banks to adopt a phased roll-out
including: (i) adoption of IRB across the asset class within the same business
unit, (ii) adoption of IRB for the asset class across business units in the same
banking group and (iii) a move from the foundation to the advanced IRB
approach for certain risk components where an advanced approach is
permitted331. When a bank adopts an IRB approach for an asset class within a
particular business unit, it must apply that approach to all exposures within
the class in that business unit332.

5.2.4 If a bank intends to adopt an IRB approach for a given asset class, it must
produce an implementation plan specifying to what extent it intends to roll
out the IRB approach within that asset class and individual business units. This
plan should be realistic and must be agreed with the relevant national

 Back to contents 66
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

supervisor. “Cherry picking” specific exposures within a given asset class is

prohibited333. Immaterial exposures may be exempted with supervisory
approval334.

5.2.5 A bank that adopts an IRB approach for a given asset class is expected to
continue that approach for that asset class. A voluntary return to a less
sophisticated approach is permitted only in extraordinary circumstances, such
as divestiture of a large part of the bank’s credit-related business, and is
subject to prior supervisory approval335.

5.2.6 Given data limitations in respect of specialised lending, a bank may move to
the foundation or advanced IRB approach for a given class of such exposures,
without doing so for other classes of specialised lending (e.g. applying an IRB
approach for project finance but not for other forms of specialised lending).
The exception is high-velocity commercial real estate, where adoption of an
IRB approach requires adoption of the same approach to material income-
producing real estate exposures336.

5.3 Risk Components

5.3.1 The risk components for the foundation and advanced IRB approaches are as
follows337:

Probability of default (PD). PD measures the probability that a particular

borrower will default. This depends on the characteristics of the
borrower. All banks applying an IRB approach are required to determine
the probability of default 338. There is a limited exception in respect of
the specialised lending sub-classes where banks that are unable to
calculate PD owing to data limitations are able, or may be required, to
use supervisory risk weights. All banks must have a meaningful
distribution of exposures across borrower grades with no excessive
concentrations339.

Loss given default (LGD). LGD measures the extent of the loss that a
bank will suffer if a particular borrower defaults (i.e. the “recovery
rate”), measured in terms of the economic loss to the lender 340. Unlike
the probability of default, loss given default (LGD) is transaction specific
as the size of the loss will depend on the characteristics of the particular
exposure e.g. whether it is senior or subordinated, and whether the
borrower has provided any collateral, or the bank benefits from other
credit protection. Under the foundation IRB approach, Basel III specifies
the loss given default. Under the advanced IRB approach, banks are
permitted to use their own internal estimates of LGD 341.

Exposure at Default (EAD). EAD measures the size of a bank’s exposure

on the default of a borrower. For on-balance sheet transactions, the
exposure at default is normally the amount of the loan or exposure,
although EAD may be reduced if a bank has in place a recognised netting
agreement342. For off-balance sheet transactions the position is more
complex. Banks are required to estimate the extent to which the bank
is likely to be exposed if the counterparty defaults. Under the foundation
IRB approach Basel III provides standard figures for EAD based on the
credit conversion factors used in the standardised approach to convert

 Back to contents 67
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

off-balance sheet liabilities to notional on-balance sheet amounts 343.

Under the advanced IRB approach banks are permitted to determine EAD
themselves either through the committed undrawn amount multiplied by
a credit conversion factor, or through direct estimation of total facility
EAD344, provided that the facility is not subject to a 100% credit conversion
factor under the standardised approach 345. The definition of
commitments under the standardised approach applies to the foundation
and advanced IRB approaches346. Floors apply to own-estimates of EAD347.

Maturity (M). M measures the contractual maturity of a bank’s exposure

to a borrower. The relevance of maturity is that, other factors being
equal, a long term loan is more risky than a short term loan. Maturity is
also relevant in respect of collateral, on-balance sheet netting,
guarantees and credit derivatives if there is a mismatch between the
underlying exposure and the hedge. This is considered further in the
chapter on Credit Risk Mitigation.

5.3.2 Under the foundation IRB approach, transactions are assumed to have a
maturity of two and a half years (except for repo-style transactions)348.
However, national supervisors may require banks to measure the maturity of
each transaction under the foundation IRB approach349. Banks using the
advanced approach must take account of the maturity of individual exposures,
although this may be fixed (at supervisory discretion) at two and a half years
(the foundation IRB treatment) for facilities to smaller corporate borrowers350,
reflecting their lower volatility of loss rates. Such fixing must apply to all
banks using A-IRB in that jurisdiction, and cannot be applied on a bank-by-
bank basis351.

5.3.3 The maximum maturity under IRB is five years 352. There is a floor of one
year353, although this does not apply to certain short-term exposures which are
fully collateralised and are capital market driven (i.e. OTC derivatives, margin
lending, repo-style transactions with an original maturity of less than one
year)354. Nor does it apply to the following: (i) short-term self-liquidating
trade transactions; (ii) issued and confirmed short-term letters of credit355; or
(iii) other short-term exposures defined by national supervisors 356.
Instruments subject to a determined cash flow schedule have their maturity
assessed using the exposures’ effective maturity (a formula applies)357, unless
this cannot be calculated, in which case a more conservative measure must
be used, usually the contractual maturity 358.

5.4 Minimum Requirements

5.4.1 Banks that use either the foundation or the advanced IRB approach must meet
certain minimum requirements at the outset and on an ongoing basis359. These
cover: (a) the composition of minimum requirements, (b) compliance with
minimum requirements, (c) internal rating system design, (d) risk rating
system operations, (e) corporate governance, (f) use of internal ratings, (g)
validation of internal estimates, and (h) disclosure 360. The overriding
objective is that rating and risk estimation systems and processes provide for
a meaningful assessment of borrower and transaction characteristics, a
meaningful differentiation of risk, and reasonably accurate and consistent
quantitative estimates of risk361.

 Back to contents 68
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.4.2 The requirements are complex and may vary between the foundation and the
advanced IRB approach. However, all banks applying an IRB approach must
satisfy the following requirements:

The bank’s rating system must have two separate dimensions covering (i)
the risk of borrower default and (ii) transaction-specific factors362.
Separate exposures to the same borrower must generally be allocated to
the same borrower grade which measures the risk of the borrower
defaulting363. The second dimension reflects transaction-specific factors
such as collateral, seniority and product type 364. An exception is for
specialised landing, where banks may use a single rating system that
measures expected loss or EL365.

Internal estimates of PD and, where applicable, LGD and EAD, have to

incorporate all relevant, material and available data. Both internal data
and external (e.g. bought) data may be used 366. Estimates must be
grounded in historical experience and empirical evidence. Estimates
based on purely subjective considerations are not accepted 367.

Banks may use multiple rating methodologies or systems e.g. middle

market and large corporate, although where multiple systems are used
the rationale for assigning a borrower to one system must be documented
and not be driven by the desire to minimise regulatory capital
requirements368.

Internal ratings, and default and loss estimates, must play an essential
role in the credit approval, risk management, internal capital allocation
and corporate governance of IRB banks. Ratings that are developed solely
for the purpose of determining regulatory capital requirements are not
recognised369. Generally, banks must have been using an internal rating
system broadly in line with the requirements of Basel III for three years
prior to qualification370 (although longer data sets of up to seven years
apply for calculating LGD and EAD 371).

Banks must have independent credit risk control units that are
responsible for the design or selection, implementation and performance
of their rating systems 372. Ratings assessments and reviews must be
completed by a party not standing to benefit from the decision to extend
credit373. Ratings must generally be reviewed at least on an annual
basis374.

All material aspects of the rating and estimation processes must be

approved by the bank’s board of directors (or a designated committee of
directors), and by senior management. Directors and senior management
must also have a general understanding of the bank’s rating system and
detailed comprehensive understanding of management reports 375. Senior
management must also have a good understanding of the design and
operation of the system, and management must ensure that it operates
properly376.

The bank’s rating system must be reviewed annually by the bank’s

internal audit function377.

 Back to contents 69
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.4.3 The IRB approaches to corporate (other than specialised lending), sovereign,
bank and investment firm portfolios are very similar. The approach for retail
exposures differs owing to the different characteristics of retail lending.

5.5 No IRB Treatment

5.5.1 If no IRB treatment is specified for an asset class, the risk weight is 100%
(except where a 0% risk weight applies under the standardised approach) and
the resulting risk weighted assets are assumed to cover unexpected losses
only378. Equity exposures must be risk-weighted using the standardised
approach as no IRB approach may be used for equity exposures379.

5.6 Foundation IRB for Corporate Exposures

5.6.1 Corporate exposures are defined, in general, as the debt obligation of a

corporation, partnership or proprietorship380. Banks are permitted to separate
out exposures to small and medium-sized (SME) borrowers which attract a
lower risk weight (and, hence, capital charge) 381. The preferential treatment
for SME lending applies where the reported sales for the borrower’s
consolidated group are less than €50 million 382 (or, at national discretion,
where total assets are below €50 million, if total sales are not a meaningful
indicator of firm size383).

5.6.2 Under the foundation IRB approach, banks are required to specify a number
of grades for performing and non-performing loans. These grades should cover
the spectrum from loans that are virtually risk-free to those that are in
default. At a minimum, a bank is required to have seven grades for performing
loans and one grade for non-performing loans384. A “grade” means an
assessment of borrower risk on the basis of a specified and distinct set of
rating criteria, and the intention is that each grade provides for a basically
homogenous pool of exposures385. External ratings may be the primary factor
in determining an internal rating, but banks must consider other available
information386.

5.6.3 The bank then allocates each loan to a single grade. The bank determines for
each grade the long-run average probability of default (PD) of all loans within
that grade or pool over a one year time horizon 387. Although the time period
for calculating PD is one year, banks are expected to use a longer time frame
when allocating specific borrowers to individual grades 388. This is to counter
the effect of the economic cycle on the calculation of capital charges, which
would otherwise increase significantly as borrowers’ creditworthiness
deteriorated in an economic downturn389. The result is to give a single average
PD figure for each borrower grade. For very high quality grades Basel III
imposes a PD floor of 0.05% (up from 0.03% under Basel II) 390. For defaulted
assets the PD is 100%391. Three methods are permitted for the calculation of
PD estimates: (i) data on internal default experience, (ii) mapping internal
grades to the scale used by a ratings agency, and (iii) a simple average of
default-probability estimates for borrowers assigned to a given grade 392.

5.6.4 The next step involves the bank estimating the likely loss given default (LGD)
for each exposure within each borrower grade. Under the foundation IRB
approach, LGD is itself specified by Basel III and is 40% for senior claims (a
reduction from Basel II)393 and 75% for subordinated claims394. This reflects an

 Back to contents 70
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

assumption that on average a bank will recover 60% in respect of senior claims
and 25% for subordinated claims. A facility is treated as subordinated if it is
expressly subordinated to another facility, although national supervisors may
use a wider definition of subordination, including economic subordination395.

5.6.5 There is no express maturity adjustment under the foundation IRB approach.
However, as mentioned above, national supervisors may require the effect of
maturity to be taken into account under the foundation IRB approach.

5.6.6 The risk weight for on-balance sheet assets is calculated as a continuous
function of the probability of default (PD) and the loss given default (LGD).
The formula for determining the continuous risk weight function is set out in
Basel III and produces a function that sets out, for each level of PD and LGD,
a given risk weight396. As mentioned above, a downward adjustment is made
in the case of exposures to SME borrowers. The amount of the adjustment
depends on the total annual sales of the individual SME borrower 397.

5.7 Risk weighted assets

5.7.1 As with the standardised approach, the capital charge under IRB is based on
risk weighted assets. A risk weighted asset (RWA) is determined by multiplying
the risk weight of the exposure (derived from the continuous risk function
referred to above) by the bank’s exposure at default (EAD).

RWA = counterparty risk weight capital charge x 12.5 x EAD 398

5.7.2 For loans and on-balance sheet items EAD is simply the nominal amount of the
exposure. The overall capital charge is therefore equal to the sum of each risk
weighted asset.

5.7.3 For off-balance sheet items (other than derivatives) the approach is similar to
that for loans. The bank assigns the exposure to a PD grade as described
above. In the case of a commitment, this will be the PD grade of the borrower.
For a guarantee provided by a bank, the PD grade corresponds to that of the
underlying obligor. The bank then calculates the loss given default (LGD) by
applying either a 40% figure for senior commitments or a 75% figure for
subordinated commitments. The counterparty risk weight is derived in
exactly the same way as for on-balance sheet items through the continuous
risk weight function referred to above. The counterparty risk weight is then
multiplied by a credit conversion factor (CCF) to produce the capital charge.
Under the foundation IRB approach, banks are required to apply the same
credit conversion factors as under the standardised approach 399.

5.8 Advanced IRB for Corporate Exposures

5.8.1 The advanced IRB approach is similar to the foundation IRB approach. The
main difference is that banks are required, with the consent of their
supervisor, to apply their own estimates in calculating the loss given default
and the exposure at default400. The requirements in respect of banks’ own
estimates of LGD and EAD are set out in the Basel III text401. Additionally, there
is an express maturity (M) dimension so that risk weights are calculated from
a continuous function of the probability of default, the loss given default, the
exposure at default and the maturity of exposures 402. As with the foundation

 Back to contents 71
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

IRB approach, a risk weighted asset is equal to the exposure at default

multiplied by the applicable counterparty risk weight. For certain off-balance
sheet items banks may apply their own internal models in place of the
specified credit conversion factors provided the exposure is not subject to a
credit conversion factor of 100% under the foundation IRB approach (in which
case the bank must apply a 100% credit conversion factor) 403. These are
undrawn revolving commitments to extend credit, purchase assets or issue
credit substitutes. For other off-balance sheet exposures standardised credit
conversion factors must be used404. The capital charge under advanced IRB is
equal to the sum of all risk weighted assets.

5.8.2 Banks must have been estimating and employing LGDs and EADs for at least
three years prior to qualification in a manner broadly consistent with the Basel
III standard405. Data sets must meet a longer time period406.

5.8.3 As mentioned above, the A-IRB approach is not available for exposures to large
corporates.

5.9 Sovereign Exposures

5.9.1 The IRB portfolio for sovereign exposures covers exposures to sovereigns,
central banks, public sector entities that are treated as sovereigns, specified
international organisations 407, and those multi-lateral development banks that
meet the criteria for a 0% risk weight under the standardised approach 408.

5.9.2 The determination of capital charges, including the calculation of PD, LGD
and EAD, for sovereign exposures, follows a similar approach to that for
corporate exposures, although under the foundation IRB approach the
assumed LGD is 45%409. However, due to the credit risk free nature of many
sovereigns, there is no floor to the minimum probability of default (PD) 410,
which for the best credit quality sovereigns may be 0%. Exposures to
sovereigns with a 0% PD attract no capital charge.

5.9.3 Banks are able to apply both a foundation and an advanced IRB approach to
the sovereign portfolio.

5.10 Exposures to Banks and certain Securities Firms

5.10.1 The IRB portfolio for banks applies to exposures to banks, certain securities
and other financial firms and multi-lateral development banks that do not
meet the requirements for a 0% risk weight under the standardised approach.
Additionally, exposures to domestic public sector entities are treated as bank
exposures if they are not expressly treated as sovereign exposures 411. The
same applies to covered bond exposures 412. Securities and financial firms must
be subject to prudential standards and a level of supervision equivalent to
those applicable to banks (including capital and liquidity requirements) 413.
This has already been discussed in the context of the standardised approach.

5.10.2 Additionally, exposures to any of the previously listed entities which take the
form of subordinated debt or regulatory capital instruments are also included
within the IRB portfolio for banks 414 (e.g. Tier 2 capital issued by a bank). An
exception exists for: (i) exposures that are treated as equity exposures under
the IRB approach, (ii) exposures which are required to be deducted from

 Back to contents 72
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

capital, (iii) exposures risk-weighted at 250% under the threshold deductions

approach; or (iv) exposures risk-weighted 1250% under the standardised
approach as they constitute a “material holding”415. The purpose of these
qualifications is to ensure that where another, more onerous, treatment is
specified elsewhere in the Basel III framework then that treatment applies to
the exposure, and not the treatment under the IRB approach to bank
exposures.

5.10.3 The determination of capital charges for bank exposures follows the same
approach as that for corporate exposures (including the minimum 0.05% for
PD416).

5.10.4 Exposures to regulated financial institutions (including insurers – presumably

only when they are not subject to the framework for corporate exposures)
whose total assets are equal to or more than US$ 100 billion are subject to a
multiplier of 1.25 (i.e. a 25% uplift) 417. The reason is that failures of major
financial firms were more common in the financial crisis than failures amongst
diversified corporates. This reflected the greater correlation between
defaults amongst such firms and other companies. The same uplift applies to
unregulated financial institutions regardless of size418.

5.10.5 The assumed LGD under the foundation IRB approach is 45% 419. Subordinated
exposures attract an LGD of 75%420. The advanced IRB approach has been
withdrawn for exposures in this portfolio421.

5.11 Retail Exposures

5.11.1 The IRB approach for retail exposures differs from the three portfolios
referred to above because of differences in the risk characteristics of retail
lending, and the way in which banks manage such exposures. The essential
characteristic of retail lending is that it is a basically homogenous portfolio
with a large number of small value loans.

5.11.2 There is no foundation approach for retail exposures. Banks are therefore
required to calculate internal estimates for all risk components (i.e. PD, LGD
and EAD)422 including the effect of credit risk mitigation techniques in the
retail portfolio423. Banks applying, or required to apply, the foundation IRB
approach to corporate or bank exposures may apply the IRB treatment for
retail exposures.

5.11.3 For any exposure to fall within the retail portfolio it must meet criteria
relating to both the borrower and, in one case, the value of the exposure424.
The retail portfolio is comprised of the following types of transactions (the
definition has changed since Basel II):

exposures to individuals not secured on real estate regardless of size

(national supervisors may implement a size criterion to distinguish retail
from corporate lending). Examples of exposures in this asset class are
credit cards, overdrafts, retail facilities secured by financial instruments,
personal loans, car loans, student loans and other exposures with similar
characteristics425;

 Back to contents 73
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

exposures to individuals secured by residential mortgages (including

second and subsequent mortgages) regardless of size 426;

exposures to associations or co-operatives of individuals established for

the sole purpose of enabling members to acquire a primary residence,
regardless of size427; and

SME loans428 provided: (i) the loan is originated and managed as a retail
exposure and (ii) the total amount lent is less than €1 million 429. SME
loans extended through, or guaranteed by, an individual are subject to
the same treatment430. The first requirement excludes loans managed
individually in a way comparable to corporate exposures 431.

5.11.4 Supervisors have the discretion to restrict the eligibility of buy-to-let loans
included in the retail portfolio where the individual has mortgaged more than
a specified number of properties, in which case such lending must be treated
as a corporate exposure432. Where a threshold applies on the loan amount (as
is the case for SME lending, and may be required at national discretion for
other retail loans) the Basel Committee invites supervisors to provide
flexibility so that banks are not required to develop extensive new information
systems433.

5.11.5 Banks are required to divide their retail portfolio into three separate asset
classes: (a) residential mortgage lending, (b) “qualifying” revolving retail
exposures and (c) all other retail exposures434.

Residential mortgage sub-portfolio

5.11.6 The residential mortgage portfolio consists of retail mortgage lending

(including first and subsequent mortgages, term loans and revolving credits
secured on mortgage)435.

Qualifying revolving retail exposures

5.11.7 Qualifying revolving retail exposures are defined as loans that meet the
following criteria:

the exposures are revolving, unsecured and uncommitted, both

contractually and in practice;

the exposures are solely to individuals;

the maximum exposure to any individual in the sub-portfolio is €100,000

or less;

the portfolio has a low volatility of loss rates relative to average loss
rates;

data on loss rates are retained to allow analysis of loss rate volatility;
and

 Back to contents 74
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

the national supervisor agrees with the bank that treatment as a

qualifying revolving retail exposure is consistent with the underlying risk
characteristics of the included loans 436.

5.11.8 The final requirement is curious as elsewhere where supervisory consent is

required under Basel III (such as to use an IRB approach) the intention is that
the bank must demonstrate to its supervisor that it meets the relevant
quantitative and qualitative requirements. It seems difficult to believe that
national supervisors will independently decide whether particular exposures
included in this sub-portfolio are eligible as they lack the resources to make
such judgments. In practice, we expect banks will be required to show that
such loans meet the relevant requirements and, possibly, that application of
this treatment is prudent.

5.11.9 Segmentation at country level (or below) should be the general rule 437.

5.11.10 Further, exposures included within the qualifying revolving retail exposure
sub-portfolio are further divided into exposures to “transactors” and
“revolvers”. The definition of “transactor” is identical to that used under the
standardised approach438. In summary, it consists of: (i) credit/charge cards
where the balance is repaid in full at each payment date, (ii) overdrafts where
there has been no drawing over the past 12 months, and (iii) exposures with
less than 12 months of repayment history439. Any exposure to a person that is
not a transactor is an exposure to a revolver.

Other retail exposures

5.11.11 The “other retail” sub-portfolio consists of all retail exposures that do not fall
within the two previous categories. It is therefore a residuary category.

Capital requirements

5.11.12 Banks are required to identify, in accordance with their internal model,
distinct pools of retail exposures for each asset class. For each pool the bank
is required to provide quantitative measures of PD, LGD and EAD in respect of
that pool440. There is no obligation, therefore, to determine the LGD and EAD
for individual exposures within a pool. This reflects the homogenous nature
of such lending.

5.11.13 The minimum PD for retail lending is either 0.10% (qualifying revolving retail
exposures to “revolvers”) and 0.05% for all other retail exposures (including
qualifying revolving retail exposures to “transactors”)441. The minimum LGD
for residential mortgages is fixed at 5% irrespective of the collateral
provided442. The Basel III text also sets out LGD floors for different types of
retail exposure (see below)443.

5.11.14 The number of exposures in each pool must be sufficient to allow for a
meaningful differentiation of risk and provides for a grouping of sufficiently
homogenous exposures, and allows for accurate and consistent estimates of
loss characteristics at pool level444. When assigning exposures to a pool, banks
are required to consider:

borrower characteristics (borrower type, demographics);

 Back to contents 75
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

transaction characteristics (whether the bank possesses security,

seniority of the security, any guarantees, etc.); and

whether the exposure is performing or in default 445.

5.11.15 Banks must review the loss characteristics and status of each pool at least on
an annual basis. They must also review the status of individual borrowers
within each pool to ensure that the underlying borrowers continue to be
assigned to the correct pool. However, this requirement can be met through
the review of a representative sample of exposures446.

5.11.16 Once banks have identified the pools, they are required to calculate the
probability of default (PD) for each pool of loans. The bank will also
determine the LGD for each pool447.

5.11.17 The PD and LGD figures are then fed into the supervisory formulae to
determine the risk weights. There are three separate formulae 448 for: (i)
residential mortgages449, (ii) qualifying revolving retail exposures 450 and (iii) all
other retail exposures451. For LGD the following floors apply. Retail mortgages
attract a minimum 5% LGD. For qualifying revolving retail exposures, the
figure is 50%. For other retail exposures, the minimum LGD on unsecured
exposures is 30% and between 0% and 15% for secured exposures, depending
on the type of collateral provided452.

5.11.18 The bank then multiplies the capital charge for each pool of assets by the
exposure at default multiplied by 12.5 to determine the RWA for that pool 453.
For on-balance sheet exposures EAD is the current drawn amount. Off-
balance sheet items are converted into notional on-balance sheet exposures
using the bank’s own estimates of the credit conversion factors for undrawn
revolving commitments to extend credit, purchase assets or issue credit
substitutes, provided the exposure does not attract a 100% credit conversion
factor under the standardised approach. All other items (e.g. undrawn non-
revolving commitments) attract the credit conversion factors specified under
the standardised approach454. A floor also applies to the use of own estimates
of EAD455. For retail exposures with uncertain future drawdown (e.g. credit
cards), banks are required to take into account the history and/or expectation
of additional drawings prior to default. This may be done through an
adjustment to either the LGD or EAD figures456.

5.11.19 If a bank provides foreign exchange or interest rate commitments to a retail

customer, it is not permitted to use its internal model to determine the capital
charge and is, instead, required to apply the capital charge under the
standardised approach457.

5.12 Common Definition of Default

5.12.1 The risk components used under the IRB approaches (PD, LGD and EAD) are all
based on the default of the borrower. In order to ensure consistency between
banks that apply IRB, Basel III sets out a common reference definition of
default. All banks are required to use this definition in determining PD and,
for banks that use the advanced IRB approach, LGD and EAD as well. This
definition is not intended to affect a bank’s legal rights or the way in which a
loan is accounted for. However, there may be practical benefits for banks to

 Back to contents 76
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

ensure that their loan documentation is consistent with the reference

definition of default as this will govern the capital treatment.

5.12.2 Where a bank relies on collateral, guarantees or credit derivatives to reduce

the credit risk on an exposure, the circumstances in which the bank may resort
to the collateral, or draw on the guarantee or credit derivative, must be
compatible with the definition of default to attract capital relief.

5.12.3 A default is considered to have occurred when either or both of the following
events take place:

the bank considers that the obligor is unlikely to pay its credit obligations
to the banking group in full, without recourse by the bank to actions such
as realising security (if held); and/or

the obligor is past due more than 90 days on any material credit
obligation to the banking group. Overdrafts are considered to be past
due if the customer has breached an advised limit or has been advised of
a limit smaller than current outstandings 458.

5.12.4 In the case of exposures to public sector entities, and retail exposures,
national supervisors may substitute a 180 day period for the 90 day reference
period459.

5.12.5 The elements taken as indications of unlikeliness to pay include:

the bank puts the credit obligation on non-accrued status;

the bank makes a charge-off or account-specific provision resulting from

a significant perceived decline in credit quality subsequent to the bank
taking on the exposure;

the bank sells the credit obligation at a material credit-related economic

loss;

the bank consents to a distressed restructuring where this is likely to

result in a diminished financial obligation caused by the material
forgiveness, or postponement, of principal, interest or fees;

the bank has filed for the borrower’s bankruptcy, or a similar order, in
respect of the obligor’s credit obligation to the banking group; or

the borrower has sought or has been placed in bankruptcy or similar

protection where this would avoid or delay repayment of the credit
obligation to the banking group460.

5.12.6 National supervisors are required to provide appropriate guidance as to how

these elements must be implemented and monitored in each jurisdiction 461.

5.12.7 For retail exposures, the definition of default may be applied at the level of
a particular facility, rather than at the level of the obligor. As such, default
by a borrower on one obligation does not require a bank to treat all other

 Back to contents 77
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

obligations as defaulted462. This reflects the fact that default is much more
common in certain facilities (e.g. credit cards) than others (e.g. mortgages).

5.13 Corporate Specialised Lending

5.13.1 Basel III makes provision for five separate sub-classes of specialised lending 463.
These are sub-classes of the corporate portfolio, although owing to the
greater risks involved, and concerns as to the accuracy of banks’ internal
models and data, they are subject to additional requirements, or supervisory
estimates. In respect of all sub-classes banks may apply both the foundation
or advanced IRB approach if they can satisfy the requirements for the
calculation of PD and the other risk components 464. If they are unable to do
so (e.g. because of a lack of data), there exist standard supervisory risk
weights that apply to banks that otherwise use an IRB approach to corporate
exposures465. This is referred in Basel III as the “supervisory slotting” criteria
approach466, and requires banks to map their own internal risk grades to five
supervisory categories, each of which is assigned a specific risk weight467.

5.13.2 Specialised lending is considered to involve the following characteristics,

either in legal form or economic substance:

the exposure is typically to an entity (such as an SPV) that was specifically

created to finance and/or operate physical assets;

the borrowing entity has little or no other material assets or activities,

and therefore little or no independent capacity to repay the obligation,
apart from the income received;

the terms of the obligation give the lender a substantial degree of control
over the assets and the income that they generate; and

as a result of the foregoing, the primary source of repayment is the

income generated by the assets468.

5.13.3 The general restriction on using the advanced IRB approach for large
corporate exposures (i.e. group revenues of more than €500 million) does not
apply to specialised lending469.

Project finance

5.13.4 This is a method of funding in which the lender looks primarily to the revenues
generated by a single project, both as the source of repayment and as security
for the exposure. This type of financing is commonly employed for large,
complex and expensive installations including, for example, power plants,
chemical processing plants, mines, transportation, environment and
telecommunications infrastructure470.

5.13.5 Banks that are able to meet the requirements for calculating PD may use the
general foundation approach for corporates to derive the appropriate risk
weights (see above). Advanced IRB banks that are able to calculate PD, LGD
and EAD may use the advanced approach for corporate exposures.

 Back to contents 78
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.13.6 Banks that cannot meet the requirements for calculating PD are required
instead to map their internal ratings to five supervisory categories to
determine the risk weights. The supervisory categories and risk weights for
unexpected losses are as follows471:

Strong Good Satisfactory Weak Default

70% 90% 115% 250% 0%
BBB- or BB+ or BB BB- or B+ B to C- N/A
better

5.13.7 Banks are required to map their exposures to the five categories. The ratings
broadly correspond to the categories472, and as such cannot be used as a
substitute for express allocation of exposures to the categories. The mapping
process must be carried out in accordance with the Basel III text and not by
directly relying on the ratings in the table above 473.

5.13.8 A 0% figure applies to defaulted assets474 as by this time there should be no

unexpected loss and the expected loss triggers a separate capital charge.

5.13.9 At national discretion, supervisors may allow banks to assign preferential risk
weights of 50% to “strong” exposures and 70% to “good” exposures provided
they have a remaining maturity of less than two and a half years or if the
supervisor determines that the banks’ underwriting and other risk
characteristics are substantially stronger than those specified in the relevant
risk category475. It seems that this discretion must apply to all banks in the
relevant jurisdiction.

5.13.10 The capital charge for unexpected loss equals the risk weight set out above
multiplied by the exposure at default multiplied by 8%476.

5.13.11 The capital charge for expected losses under the supervisory slotting approach
is as follows477.

Strong Good Satisfactory Weak Default

5% 10% 35% 100% 625%

5.13.12 Where a national supervisor allows a preferential risk weight for exposures
that are “strong” or “good” for unexpected losses then a corresponding
preferential treatment is available for expected losses of 0% and 5%
respectively478.

Object finance

5.13.13 Object finance is a method of funding the acquisition of physical assets (e.g.
ships, aircraft, railway carriages or fleets) where repayment is dependent on
the cash flows generated by the specific assets that have been financed and
pledged or assigned to the lender. A primary source of cash flows might be
rental or lease contracts with third parties. If the loan is to a borrower whose
financial condition and debt servicing capacity enables it to repay the debt
without undue reliance on the specifically pledged assets it should be treated
as a collateralised corporate exposure479.

 Back to contents 79
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.13.14 Banks that are able to calculate PD under the corporate IRB approach may
apply the foundation IRB approach for corporate exposures. Banks that can
calculate LGD and EAD as well may apply the advanced approach.

5.13.15 Banks that are not able to determine PD must map their object finance
exposures to the five supervisory categories (see above). Requirements in
respect of the mapping process are set out in the text of Basel III 480.

Commodities finance

5.13.16 This category consists of short-term lending to finance reserves, inventories

or receivables of exchange-traded commodities (e.g. crude oil, metals, crops)
where the exposure will be repaid from the proceeds of the sale of the
commodity and the borrower has no independent capacity to repay 481. The
structured nature of this type of finance compensates for the weak credit
quality of the borrower482. Such lending can be distinguished from lending
financing the reserves, inventories or receivables of other more diversified
corporate borrowers where the value of the commodity serves as a risk
mitigant rather than as the primary source of repayment483.

5.13.17 The capital treatment is exactly the same as for project finance and object
finance (see above) with mapping to the five supervisory categories as set out
in the Basel III text484.

Income producing real estate

5.13.18 This exposure class encompasses lending in respect of real estate, offices-to-
let, retail space, multi-family residential buildings, industrial and warehouse
space and hotels where the prospects for repayment and recovery depend
primarily on cash flows generated by the asset. The primary source of
repayment will therefore be lease or rental payments, or the sale of the asset.
The borrower is generally (but not necessarily) an SPV, an operating company
focused on real estate construction or holdings, or an operating company with
other sources of income485. The distinguishing feature of this type of lending
(as compared with collateralised real estate lending) is the strong positive
correlation between the prospects for repayment of the loan and recovery in
default, with both depending primarily on the cash flows generated by the
property486.

5.13.19 The capital treatment (with mapping to the five supervisory categories for
banks that cannot calculate PD 487) is the same as for the three previous
specialised lending categories.

High volatility commercial real estate

5.13.20 The final category of specialised lending is the financing of commercial real
estate that exhibits higher loss rate volatility (i.e. higher asset correlation)
compared with other types of specialised lending. It includes:

commercial real estate exposures secured by properties categorised by

the national supervisor as sharing higher volatilities in portfolio default
rates;

 Back to contents 80
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

loans financing the acquisition, development and construction (ADC) of

properties of those types in such jurisdictions; and

loans financing ADC of other properties where the source of repayment

is either the future uncertain sale of the property or cash flows whose
source of repayment is substantially uncertain, unless the borrower has
substantial equity at risk488. Where there is such equity the loan may be
exempted from the high volatility commercial real estate category, but
does not benefit from any preferential risk weights for certain
exposures489.

5.13.21 It is for each national supervisor to determine which loans fall into the high
volatility commercial real estate category in their jurisdiction. Banks in other
jurisdictions are required to apply the same treatment to such loans that they
have in that jurisdiction490.

5.13.22 National supervisors determine whether or not to permit banks to apply the
foundation or advanced IRB approach to high volatility commercial real
estate491. Banks that are allowed to apply the foundation or advanced IRB
approach use a modified formula in calculating the risk weights, to reflect the
higher degree of risk in such lending492. If the banking supervisor does not
permit banks to apply an IRB approach, or if a bank is unable to calculate PD,
then the risk weights for unexpected loss in the high volatility commercial
real estate category are as follows493:

Strong Good Satisfactory Weak Default

95% 120% 140% 250% 0%

5.13.23 Once again, requirements for the mapping process are set out in Basel III 494.

5.13.24 As with other specialised lending sub-classes (but not for exempt high
volatility commercial real estate) national supervisors may reduce the risk
weight for “strong” and “good” exposures that meet certain criteria to 70%
and 95% respectively495.

5.13.25 The capital charge is determined in exactly the same way as for the other
categories of specialised lending using the different risk weights496.

5.13.26 The risk weights for expected losses are set out in the following table 497.

Strong Good Satisfactory Weak Default

5% 5% 35% 100% 625%

5.13.27 There is no preferential treatment available for expected losses in this class 498.

5.14 Defaulted Exposures

5.14.1 The capital requirement for a defaulted exposure is equal to the greater of
zero and the difference between its loss given default (LGD) and the bank’s
best estimate of expected loss (EL). The risk-weighted asset amount for the
defaulted exposure is the product of the capital charge multiplied by 12.5
multiplied by the exposure at default 499.

 Back to contents 81
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.15 Equity Exposures

5.15.1 The equity portfolio comprises exposures to equities based on the economic
substance of the instrument. The following criteria apply (which are identical
to the definition under the standardised approach):

the instrument is irredeemable;

it does not embody an obligation on the part of the issuer; and

it conveys a residual claim on the assets or income of the issuer 500.

5.15.2 Regardless of whether this definition is met or not, the following instruments
are automatically classified as equity exposures:

instruments with the same structure as those permitted as Tier 1 capital

(e.g. some preference shares and perpetual subordinated debt); and

instruments that embody an obligation on the part of the issuer and meet
any of the following conditions:

the issuer may indefinitely defer settlement;

the obligation permits or requires settlement by a fixed number of

ordinary shares;

the obligation permits or requires settlement by a variable number

of ordinary shares; or

the holder has the option to require settlement in ordinary shares unless
the supervisor considers this treatment not to be appropriate 501; and

securities structured with the intent of conveying the economic

substance of equity holdings502.

5.15.3 Unlike the position under Basel II, there is no IRB approach to equity
exposures. The purpose of this definition is therefore an exclusionary one as
any equity exposures must be risk-weighted under the standardised
approach503 (unless they are equity investments in funds, for which the specific
rules on funds apply). Application of the definition set out in the standardised
approach is therefore entirely logical.

5.16 Purchased Receivables

5.16.1 The rules for purchased receivables distinguish between corporate receivables
and retail receivables504. Purchased retail receivables are eligible for the top-
down approach505. For corporate receivables Basel III generally requires such
exposures to be treated as corporate exposures under the IRB approach
applicable to that portfolio. However, the top-down approach may be used
instead where it would be unduly burdensome for a bank to comply with the
general IRB corporate approach. According to Basel III the top-down approach
is mainly intended for receivables that are purchased for inclusion in asset-
backed securitisation structures, but can be used for appropriate on-balance

 Back to contents 82
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

sheet exposures with supervisory permission that share the same features 506
(presumably, such as warehousing).

Corporate receivables

5.16.2 To be eligible for the top-down treatment, corporate receivables must satisfy
the following conditions:

the receivables are purchased from unrelated third party sellers, and the
bank must not have directly or indirectly originated them;

the receivables are generated on an arm’s-length basis. Intra-group

receivables are accordingly ineligible;

the purchasing bank has a claim on all proceeds from the pool (or a pro-
rata share if it does not purchase the whole pool); and

the pool meets concentration requirements to prevent an undue

concentration of risk to a few counterparties. These limits are set by
national supervisors507.

5.16.3 Where a bank purchases a claim on a tranche of receivables, the securitisation

framework applies508.

5.16.4 For corporate exposures there is both a foundation and an advanced IRB
approach, although the advanced IRB approach can only be used for purchased
receivable from corporates in respect of which the advanced IRB approach is
permitted to be used (i.e. smaller companies and specialised lending). For
purchased retail receivables only the advanced IRB approach is available, in
line with the treatment of retail exposures under Basel III 509.

5.16.5 There are IRB capital charges for both default risk and dilution risk 510.

Default risk

5.16.6 Receivables that belong only to one asset class (as defined in the Basel III
standard) receive a risk weight for default risk based on the risk weight
appropriate for that asset class511. In other cases, different rules apply. Thus,
if a bank cannot meet the requirements for qualifying revolving retail
exposures for purchased receivables in that class, it must use the risk-weight
function for other retail exposures. Hybrid pools attract the highest capital
requirements of any of the exposures in that pool 512.

5.16.7 Generally, a bank must use the standard IRB approach for purchased corporate
receivables. However, with supervisory approval, a bank may use the
following top-down approach. The bank will estimate the one-year expected
loss (EL) for default risk in the pool. This is an estimate of the percentage of
the overall nominal amount in the pool anticipated to default over the next
year. Given the expected loss, the risk weight for the pool is determined by
using the risk weight function for corporate exposures. The precise method
of calculating the risk weight function depends on whether the bank applies
the foundation or the advanced IRB approach, and whether the bank can
decompose EL into its PD and LGD components513.

 Back to contents 83
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.16.8 Banks that are able to estimate PD will determine the risk weight in the same
way as for other corporate exposures514.

5.16.9 If the bank is unable to decompose EL into PD and LGD components then the
risk weight is determined as follows. If the pool consists solely of senior claims
an LGD of 40% is used. PD is determined by dividing EL by this LGD. EAD is
the outstanding amount less the capital charge for dilution prior to application
of any credit risk mitigants. EAD for a revolving purchase facility is the current
nominal amount plus 40% of any undrawn purchase commitments less the
dilution amount515.

5.16.10 Otherwise, where the pool consists of both senior and subordinated
receivables, PD is the bank’s estimate of EL. In this case the LGD is 100% and
EAD is the outstanding amount less the capital charge for dilution. EAD for a
revolving purchase facility is the sum of the current amount of receivables
purchased plus 40% of any undrawn commitment less the capital charge for
dilution516.

5.16.11 Under the advanced approach, banks can either estimate the pool’s default-
weighted average loss rates given default, or average PDs517. Banks may use
either an appropriate PD estimate to infer the long-run default-weighted
average loss rate or use a long-run default-weighted average loss rate to infer
PDs. Banks will then use the PD and LGD figures to calculate the risk weight
function applying the formula for corporate exposures. The EAD for purchased
corporate receivables is the nominal amount less the capital charge for
dilution (see above)518. For revolving purchased facilities EAD is calculated in
the same way as for foundation IRB banks (i.e. the nominal amount plus 40%
of any undrawn purchase commitments less the dilution amount). Banks on
the advanced approach are not permitted to apply their own estimates of
EAD519. As with other advanced IRB portfolios, the bank must then apply a
maturity adjustment520.

Retail receivables

5.16.12 The IRB treatment for retail receivables is similar to the advanced approach
to corporate receivables (although, of course, the relevant retail risk weight
functions apply and not the corporate risk weight function).

5.16.13 The bank calculates the PD and LGD (or EL) for each of the three sub-classes
of retail exposures (mortgages, qualifying revolving retail exposures and other
exposures). External and internal data may be used, and the estimates for
PD and LGD, or EL, must be calculated without any assumption of recourse
from the seller521. Specific rules apply to the calculation of M522.

Dilution risk for both corporate and retail exposures

5.16.14 Dilution is the risk that the amount received from a debtor is reduced through
cash or non-cash payments to, or set-off against, the original creditor.
Examples include goods returned to seller, disputes regarding product quality,
promotional discounts offered by the original creditor, and set-offs between
the borrower and the creditor523. Unless a bank can demonstrate to its
supervisor that the risk of dilution is immaterial, it must make an adjustment
to the value of exposures in the pool to reflect this risk524.

 Back to contents 84
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.16.15 Banks are required to calculate the one year expected loss (EL) from dilution.
This may be determined using both external and internal data 525. Either a top-
down (whole pool) or a bottom-up (individual exposures) approach may be
used. The amount is determined as follows. The expected loss is used as the
PD figure in the corporate risk weight function, and a LGD of 100% is assumed
(these will be actual losses). The amount is then subject to an appropriate
maturity adjustment. If the maturity risk is appropriately monitored and able
to be resolved within one year, a one year maturity applies526.

5.16.16 This treatment for dilution risk applies regardless of whether the receivables
are corporate or retail527.

Purchase price discounts and credit risk mitigation

5.16.17 Where receivables are purchased at a discount, and the discount serves to
provide “first loss” protection against default and dilution losses, the
purchasing bank may recognise this. Refundable purchase price discounts may
be treated as first loss protection under the Basel III framework for
securitisation. Non-refundable purchase price discounts do not affect the EL
provision or the calculation of risk weighted assets 528. Collateral and
guarantees providing first loss protection may also be recognised under the
securitisation framework529.

5.17 Treatment of Expected Losses

5.17.1 At a late stage before the final text was published, Basel II was amended to
adopt a more refined treatment of expected losses. Unexpected losses are
taken account of through the IRB capital charge for credit risk. Expected
losses, on the other hand, are measured against provisions with any additional
losses subject to a deduction from capital.

5.17.2 The rules on expected losses therefore require banks to calculate both losses
and provisions.

Expected losses

5.17.3 Banks must sum the expected loss amount for all exposures subject to the IRB
treatment. This is equal to EL multiplied by EAD. EL amounts attributable to
securitisation exposures are excluded from this calculation530.

5.17.4 For sovereign, bank, corporate and retail exposures not in default expected
losses equal PD multiplied by LGD. For exposures in default, banks must either
use supervisory LGD estimates (foundation IRB approach) or their best
estimate of expected loss (advanced IRB). Specialised lending exposures are
subject to the supervisory slotting approach 531.

Provisions

5.17.5 Total eligible provisions are all provisions (specific provisions, partial write-
offs, portfolio-specific general provisions, provisions for country risk, etc.)
attributed to exposures under the IRB approach. Discounts on defaulted
assets may also be included. However, specific provisions for securitisation
exposures must be excluded532.

 Back to contents 85
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.17.6 Where a bank applies both the standardised approach and an IRB approach to
credit risk, provisions must be separately allocated on a pro rate basis to cover
expected losses on loans subject to the standardised approach and those using
an IRB approach533. At national discretion, banks using both the standardised
and one or more IRB approaches may rely on internal methods for allocating
general provisions534. Prior supervisory approval is required.

Capital charges

5.17.7 Banks must deduct the total EL amount from the total of eligible provisions 535.

5.17.8 If the EL amount exceeds eligible provisions then this is deducted from
common equity Tier 1 capital536.

5.17.9 If the EL amount is less than total eligible provisions then banks may recognise
the excess provisions as Tier 2 capital up to a maximum of 0.6% of credit risk
weighted assets. Supervisors may set a lower limit as a national discretion
under Basel III537. Supervisors must consider whether the expected loss figure
fully reflects the conditions in the market in which banks operate before
allowing the difference to be included in Tier 2 capital. If specific provisions
exceed expected losses on defaulted assets then the same assessment must
also be made538.

Securitisation

5.17.10 Expected losses on securitisation positions are excluded from the above
treatment539. However, originator banks can offset 1250% risk-weighted
securitisation exposures by the amount of specific provisions on the underlying
securitised assets and non-refundable purchase discounts540.

5.18 Transitional Arrangements for Equity Exposures subject to an IRB treatment under
Basel II

5.18.1 Basel III contains limited transitional provisions addressing the withdrawal of
any IRB approach to equity exposures on 1 January 2023. No transitional
provisions apply to the withdrawal of the advanced IRB approach to bank,
securities firm and large corporate exposures.

5.18.2 The requirement to use the standardised approach to all equity exposures is
subject to a five year linear phase-in arrangement from 1 January 2023.
During the phase-in period the risk weight for equity exposures is the greater
of:

the risk weight calculated using the IRB approach applied to equity
exposures before 1 January 2023; and

the risk weight applicable under the transitional arrangements for the
treatment of equity exposures under the standardised approach 541. For
ease of understanding this will be repeated below.

5.18.3 Speculative unlisted equity exposures are risk-weighted starting at 100%

increasing by 60 percentage points at the end of each year until year 5 (i.e.
2028).

 Back to contents 86
 INTERNAL RATINGS-BASED APPROACH TO CREDIT RISK (IRB)

5.18.4 All other equity exposures are risk-weighted at 100% increasing by 30

percentage points until year 5542.

5.18.5 National supervisors may require banks to apply the fully phased-in
standardised treatment from 1 January 2023543.

 Back to contents 87
 CREDIT RISK MITIGATION

6. CREDIT RISK MITIGATION

6.1 Introduction

6.1.1 Credit risk mitigation refers to certain techniques which banks can use to
reduce the credit risk on their loans or other portfolios of exposures, for
example, by taking collateral, purchasing credit protection under a credit
derivative, receiving a guarantee, or entering into an on-balance sheet
netting agreement. Credit risk mitigation plays an important role in reducing
a bank’s exposure on the default of a counterparty or borrower. The rules set
out in Basel III apply to credit risk mitigation techniques used in a bank’s
banking book.

6.1.2 There are three different regimes for credit risk mitigation, and the rules
applicable to a bank depend on whether it applies the standardised approach,
the foundation IRB approach or the advanced IRB approach (where the
advanced approach is permitted).

6.1.3 All banks that use credit risk mitigation techniques to reduce their capital
requirements are required to make certain disclosures under Pillar 3 (market
disclosure). Banking supervisors are also empowered to take account of risks
inherent in credit risk mitigation (e.g. legal, operational, liquidity and market
risks) under Pillar 2. This is because although such techniques may reduce
credit risk they may simultaneously increase other risks. Banks are therefore
required to employ robust procedures and processes to address these risks,
including strategy, consideration of the underlying credit, valuation and
control of roll-off risks (where the risk mitigation technique has a shorter
maturity than the exposure)544.

6.1.4 Banks that use credit risk mitigation techniques are not required to hold more
capital against an exposure than had no such techniques been used 545. Further,
credit risk mitigation must not be double counted, meaning that where such
techniques are already taken into account in determining the risk weight, no
additional reduction in capital charges is allowed 546.

6.1.5 Where a bank has multiple credit risk mitigation techniques covering a single
exposure (e.g. collateral and a credit derivative) the bank is required to divide
the exposure into portions notionally covered by each type of technique,
calculating the risk weight on each such notional exposure. The same applies
where the techniques used have different maturities547.

6.2 Legal Certainty

6.2.1 For all types of credit risk mitigation banks must ensure that the
documentation used is binding on all parties and is enforceable in all relevant
jurisdictions. Basel III requires banks to have conducted sufficient legal
review to verify this, and to have a well-founded basis to reach this
conclusion. Banks must periodically review the legal position to ensure
continuing enforceability 548. Given that Basel III does not require an external

 Back to contents 88
 CREDIT RISK MITIGATION

legal opinion, banks may determine how to satisfy this (subject to any
additional requirements imposed by their supervisor).

6.3 Credit Risk Mitigation under the Standardised Approach

6.3.1 The standardised approach distinguishes between collateral, on-balance sheet

netting, guarantees and credit derivatives. The credit risk mitigation
technique must not bear a material positive correlation with the credit quality
of the counterparty, or the resulting residual risks 549. An example would be
taking as collateral securities issued by the debtor, or a guarantee by the
obligor of its own debt.

Collateral

6.3.2 A collateralised transaction is where a bank has a credit exposure, or potential

credit exposure, and this is hedged by collateral posted by the counterparty,
or by a third party on its behalf550.

6.3.3 Before capital relief will be granted in respect of collateral, a bank must
satisfy the following conditions:

Firstly, the bank must have a right to liquidate or take legal possession of
the collateral, in a timely manner, on the default of the counterparty
(including on its insolvency) and (where applicable) of the custodian
holding the collateral551.

Secondly, banks must take all steps necessary to obtain and maintain an
enforceable security interest under the law applicable to the bank’s
interest in the collateral (e.g. complying with any applicable registration
requirements, such as a filing in respect of security granted by an English
company)552.

Thirdly, there must not be a positive correlation between the value of

the collateral and the credit quality of the obligor553. Unless this is the
case, the collateral will provide little protection against the default of
the obligor.

Fourthly, a bank must have clear and robust procedures for the timely
liquidation of collateral to ensure that any legal conditions in respect of
a declaration of default of the borrower are observed, and the collateral
can be liquidated promptly554.

Fifthly, where a bank holds collateral through a custodian it must take

reasonable steps to ensure that the custodian segregates the collateral
from its own assets555.

6.3.4 No collateralised claim should result in a higher capital requirement than an

otherwise identical claim for which there is no collateral 556. Where the effect
of collateral has already been taken into account in determining the rating
applicable to an exposure (e.g. in a securitisation tranche or retail mortgage
lending under the standardised approach), then any collateral provided will
not further reduce the capital charge557.

 Back to contents 89
 CREDIT RISK MITIGATION

6.3.5 Basel III sets out both a simple approach and a comprehensive approach to
collateral. Banks may choose which approach to use 558, but may not use
both559. For collateralised OTC derivative transactions, exchange traded
derivatives and long settlement transactions banks may use either the
standardised approach for counterparty credit risk or an internal models
approach560.

Eligible assets

6.3.6 Under the simple approach the following assets are eligible as collateral:

cash on deposit with the lending bank (including certificates of deposit,

or comparable instruments)561;

gold;

(where external ratings are permitted to be used) securities rated BB-

and above issued by sovereigns and public sector entities treated as
sovereigns under the standardised approach;

(where external ratings are permitted to be used) securities issued by

banks, securities firms and corporates that are rated BBB- and above;

(where external ratings are permitted to be used) short-term debt

instruments rated A-3/P-3 or better;

certain listed liquid unrated senior bank debt where all rated debt is
investment grade (conditions apply);

(in jurisdictions that do not allow the use of external ratings) equivalent
securities to the last four points where issued by a sovereign, a bank
assigned to grade A under the standardised approach, debt securities
treated as “investment grade” under the standardised approach, and
securitisation exposures with a risk weight of less than 100% under the
securitisation standardised approach;

equities and convertible bonds included in a main index; and

interests in mutual funds and UCITS that are publicly quoted daily and
limited to investing in any of the foregoing 562.

6.3.7 Equities and convertible bonds not included in a main index, but traded on a
recognised exchange (and UCITS/mutual funds which include such equities),
may be used as collateral under the comprehensive approach563.

6.3.8 Re-securitisation exposures (such as collateralised debt obligations) are not

eligible financial collateral564.

The simple approach to collateral

6.3.9 The simple approach is primarily intended for banks that engage only to a
limited extent in collateralised transactions, and for whom applying the more
detailed rules under the comprehensive approach would be unduly

 Back to contents 90
 CREDIT RISK MITIGATION

burdensome. As the approach is less accurate, the capital charges for banks
applying the simple approach are generally higher than under the
comprehensive approach. The approach taken is a substitution approach so
the effect of a bank holding collateral is to substitute an exposure to the
issuer of the collateral for the original counterparty exposure 565. A corporate
loan fully secured with government securities will therefore be treated as an
exposure to the relevant government subject to the capital floor of 20% (see
below).

6.3.10 Under the simple approach, collateral must be pledged for at least the life of
the exposure, and must be marked to market and revalued at least every six
months. No capital relief will be provided in respect of collateral provided
for a period shorter than the life of the exposure. The risk weight applicable
to collateralised claims is equal to the risk duration of the collateral
instrument. This is subject to a floor of 20%, so that under the simple
approach banks are required to hold capital against one fifth of the exposure
(which is assigned the risk weight of the bank’s underlying counterparty)566.
Thus for the hypothetical corporate loan secured by government securities
the bank must hold capital equal to 20% of the value of the original loan.
Currency mismatches are disregarded under the simple approach due to the
20% floor567.

Exceptions to the 20% floor

6.3.11 There is an exception to the 20% floor for certain repo-style transactions that
are either (i) overnight or (ii) marked-to-market and re-margined on a daily
basis. These receive a counterparty risk weight of either 0% or 10%. Both the
exposure and the collateral must be cash or zero risk-weighted government
or public securities. The exposure and the collateral must be denominated in
the same currency. The transaction is required to settle across a settlement
system proven for the type of transaction and to meet various documentation
requirements. Upon a default, the bank must have an unfettered enforceable
right to seize and liquidate the collateral 568.

6.3.12 If the transaction is entered into with a “core market participant” it receives
a 0% risk weight. Otherwise the transaction is risk weighted at 10% 569. The
definition of “core market participant” is determined by the relevant
supervisor. It may include sovereigns and central banks, banks and securities
firms, other financial firms eligible for a 20% risk weight, regulated mutual
funds, regulated pension funds and certain central counterparties 570.

Specific collateralised transactions

6.3.13 OTC derivative transactions that are subject to daily marking-to-market,

which are cash collateralised and where there is no currency mismatch, are
risk-weighted at 0%. If the exposure is collateralised by government or public
securities qualifying for a 0% risk weight then the risk weight for the
transaction is 10%.

6.3.14 A 0% risk weight may also be applied to the collateralised part of transactions
where the exposure and the collateral are denominated in the same currency
and either:

 Back to contents 91
 CREDIT RISK MITIGATION

the collateral consists of cash on deposit; or

the collateral consists of government or public sector securities eligible

for a 0% counterparty risk weight and the market value of the collateral
has been discounted by 20%571.

The comprehensive approach to collateral

6.3.15 The comprehensive approach seeks to achieve a more accurate measurement

of the effect of collateral in reducing risk. The comprehensive approach is
based on the fact that where a bank holds collateral against an exposure, it
is exposed to additional risks than simply the default of the issuer of the
collateral. For example, if the collateral is only revalued periodically the
market value of the collateral, when the bank comes to sell it, may have
fallen below the level of the exposure. Equally, by that time the size of the
exposure may have increased (e.g. on a securities loan). Where the collateral
and the exposure are denominated in different currencies, the bank is
exposed to foreign exchange risk. When a bank seeks to realise its security
interest, it may also find that it lacks good title (for example, because the
collateral agreement is unenforceable) or that the collateral is economically
worthless. For these reasons a straightforward substitution approach does not
capture all the risks arising out of such transactions 572.

6.3.16 Basel III seeks to take account of these additional risks through applying
various “haircuts” (i.e. reductions) to the value of collateral to recognise the
risk that the collateral may not fully cover the exposure on a counterparty’s
default.

“Haircuts”

6.3.17 Firstly, under the comprehensive approach banks are required to apply a
“haircut” to the value of the collateral. This is to protect the bank against
the risk of changes in the value of the collateral.

6.3.18 Secondly, a “haircut” is applied to the collateral in respect of the underlying

exposure. This catches the risk that the size of the exposure may itself change
over time. No “haircut” is applied to the exposure itself.

6.3.19 Where the collateral is denominated in a currency other than that of the
exposure, a third “haircut” is applied573 to the collateral to take account of
possible foreign exchange movements 574.

6.3.20 The size of the haircut depends on the prescribed holding period i.e. the
assumed period of time over which exposure or collateral values are assumed
to move before the bank can close out the transaction 575. This varies
depending on the type of instrument, type of transaction, residual maturity
and frequency of marking-to-market and re-margining576. The exposure
amount after risk mitigation is then multiplied by the risk weight of the
counterparty to obtain the risk-weighted asset amount for the collateralised
transaction577.

6.3.21 The comprehensive approach consists of a set of different “haircuts”

depending on whether the national supervisor does 578 or does not579 allow use

 Back to contents 92
 CREDIT RISK MITIGATION

of external credit ratings. For repo-style transactions, other capital market

transactions and secured lending there are prescribed holding periods580. If
the frequency of re-margining or revaluation is longer than the minimum, the
haircut is scaled upwards581.

6.3.22 Under Basel II it was possible to use (with supervisory approval) own estimates
of “haircuts”, as well as VaR models, to calculate credit risk on certain
transactions. These options have now been removed for banks applying the
standardised approach.

6.3.23 The carve-out for certain repo-style transactions in government securities

(with no floor and no “haircuts”) referred to above in connection with the
simple approach is also available under the comprehensive approach to
collateral582.

Master netting agreements

6.3.24 Bilateral master netting agreements covering securities financing transactions

may be recognised provided that the agreements are legally enforceable in
each relevant jurisdiction upon a default. The agreement must (i) provide
the non-defaulting party the right to terminate and close-out all transactions,
(ii) provide for the netting of gains and losses and (iii) allow the prompt
liquidation or set-off of collateral following default 583. Transactions may be
netted across the banking and trading books provided all transactions are
marked-to-market daily and the collateral is recognised as eligible financial
collateral in the banking book584. A formula is used to calculate the
counterparty credit risk requirement based on the current exposure, an
amount for the idiosyncratic exposure of the securities based on the gross
exposure, and an amount for currency mismatches585.

On-balance sheet netting

6.3.25 A bank may use on-balance sheet netting agreements to reduce its exposure
to a counterparty. The effect of a netting agreement is that (subject to a
“haircut” where there is a currency mismatch) the bank is required to hold
capital against the net, as opposed to the gross, exposure to that
counterparty. Under Basel III, a bank may net loans and deposits to or from a
single counterparty586. Netting agreements relating to assets other than loans
and deposits, and multi-lateral netting agreements, are not recognised for
regulatory capital purposes as their legal enforceability is not considered to
be sufficiently well established.

6.3.26 For a netting agreement to be recognised, the following requirements must

be met:

the bank has a well-founded legal basis for concluding that the netting
or offsetting agreement is enforceable in each relevant jurisdiction
(including on insolvency);

the bank is able at any time to determine the assets and liabilities with
the same counterparty that are subject to the netting agreement;

the bank monitors and controls roll-off risks; and

 Back to contents 93
 CREDIT RISK MITIGATION

the bank monitors and controls the relevant exposures on a net basis 587.

6.3.27 On-balance sheet netting is treated as a collateralised transaction with loans

classified as exposures and deposit as liabilities 588. No haircuts are made
except where a currency mismatch exists589 with the result that capital will be
required against the net exposure only.

Guarantees and credit derivatives

6.3.28 Guarantees (which include insurance590) and credit derivatives are recognised
under Basel III as techniques that reduce a bank’s credit risk. Basel III sets
out certain requirements that are common to both guarantees and credit
derivatives, and also specified operational requirements. Only protection
provided by guarantors and credit protection sellers with a lower risk weight
than the underlying counterparty lead to capital reductions 591. The effect of
a bank purchasing eligible credit protection is to substitute the risk weight
applicable to the guarantor/protection provider for that of the counterparty
for the covered exposures592.

6.3.29 Guarantees and credit derivatives are only recognised if provided by:

sovereigns593, public sector entities, multi-lateral development banks,

banks, securities firms and other prudentially regulated financial
institutions594 (such as insurance companies) with a lower risk weight than
the counterparty;

(in jurisdictions allowing the use of external ratings) other entities with
a better external risk weighting than the counterparty. For securitisation
exposures, the credit protection provider must additionally have been
rated A- or better when the credit protection contract was provided and
still be rated BBB- or better;

(in jurisdictions that do not allow the use of external ratings) the
guarantor/credit risk provider is treated as “investment grade” and (for
corporates) has securities outstanding on a recognised securities
exchange595.

6.3.30 Sovereign (or central bank) guarantees denominated in the domestic currency
may attract a lower risk weight (at national discretion) provided that the
guaranteed exposure is denominated in that currency 596.

6.3.31 Sovereign counter-guarantees (i.e. where the sovereign guarantees payment

by a third party guarantor) may be treated as sovereign guarantees if the
counter-guarantee covers all credit risk elements of the exposure, both the
initial guarantee and the counter-guarantee meet all operational
requirements and the relevant national supervisor is satisfied that the cover
is robust and the counter-guarantee is effectively equivalent to a direct
sovereign guarantee597.

6.3.32 Parental and other group company guarantees are recognised for regulatory
capital purposes subject to the rules above. Where external ratings are not
permitted to be used, intra-group protection must not be positively correlated

 Back to contents 94
 CREDIT RISK MITIGATION

with the credit risk of the exposures for which credit protection is provided 598,
and the credit risk of the whole group must be taken into account 599.

6.3.33 The following general requirements apply to both guarantees and credit
derivatives:

the guarantee/credit derivative must represent a direct claim on the

protection provider;

the credit protection is explicitly referenced to specific exposures or a

pool of exposures so the extent of cover is clearly defined and
incontrovertible;

other than non-payment by the protection purchaser, it must be

irrevocable;

there is no clause in the contract permitting the guarantor/seller of

credit protection to unilaterally cancel the cover, change the maturity of
cover or to increase the effective cost of cover as a result of a
deterioration in the credit quality of the hedged exposure;

there is no clause in the protection contract outside of the bank’s direct

control that could prevent the protection provider being obliged to pay
in a timely manner if the underlying counterparty defaults; and

payment is unconditional600.

Guarantees

6.3.34 In respect of guarantees, Basel III sets out four specific requirements. Firstly,
on default/non-payment the bank must have the right to pursue the guarantor
in a timely way without having to bring legal action against the obligor first 601.
Secondly, the guarantee must be an explicitly documented obligation assumed
by the guarantor. Thirdly, the guarantee must cover all types of payments
that the underlying counterparty is expected to make under the
documentation governing the transaction. Finally, if the guarantee covers
principal only, interest and other uncovered payments must be treated as an
unsecured amount602.

Credit derivatives

6.3.35 Only credit default swaps and total return swaps that provide credit
protection equivalent to guarantees 603 are eligible. An exception to eligibility
exists where a bank buys credit protection through a total return swap and
records the net payments under the swap as income but does not record
offsetting deterioration in the value of the asset that is protected which are
not recognised as being eligible 604. This is because there can be no certainty
as to the value of the protection purchased.

6.3.36 Basel III also specifies minimum credit events for a credit derivative to be
eligible. These include:

 Back to contents 95
 CREDIT RISK MITIGATION

a failure to pay the amounts due (with any grace period being in line with
that of the underlying obligation);

bankruptcy, insolvency or inability to pay its debts; and

restructuring of the underlying obligation involving forgiveness or

postponement of principal, interest or fees that results in a credit loss
event605.

6.3.37 If restructuring is not covered by the credit derivative, partial recognition is

allowed. If the amount of credit protection is less than or equal to the
underlying exposure then 60% of the exposure will be treated as hedged. If
the credit protection exceeds the exposure, the hedge is capped at 60% 606.

6.3.38 The credit derivative must not terminate before the expiry of any grace period
required for default on the underlying obligation607. If the credit derivative
requires physical settlement (i.e. transfer of the reference obligation to the
protection seller for settlement) then the terms of the underlying obligation
must provide that any required consent to transfer may not be unreasonably
withheld608. If the credit derivative provides for cash settlement then there
must be a robust valuation process in place (including a specified period for
obtaining post-credit event valuations) to determine the amount of loss 609.
Further, the party responsible for determining whether a credit event has
occurred must be clearly defined, and it cannot be the sole responsibility of
the protection seller i.e. the protection buyer must have the right/ability to
inform the protection seller (or any determinations committee610) of the
occurrence of a credit event611.

6.3.39 There are also specific requirements in respect of asset mismatches (i.e. a
mismatch between the underlying obligation and the reference obligation
under the credit derivative). Generally, the reference obligation must rank
pari passu to or be junior to the underlying obligation612.

Capital treatment

6.3.40 Where an exposure is guaranteed, or is covered by a credit derivative, the

capital charge in respect of the part of the exposure covered by the credit
derivative or guarantee is assigned the risk weight of the protection provider.
Any materiality thresholds on payment are treated as first loss positions and
are assigned a risk weight of 1250% by the bank purchasing the credit
protection613 (i.e. are treated as effectively deducted from capital).

6.3.41 Where losses are shared on a pro rata basis between the bank and the
guarantor, capital relief is recognised on a proportional basis, meaning that
the protected portion receives the treatment applicable to the protection
provider, whilst the rest of the exposure is treated as unsecured 614.

6.3.42 Tranched protection (i.e. where a bank transfers a portion of the credit risk
to a protection provider and the risk retained and transferred are of different
seniority) is subject to the securitisation approach 615.

6.3.43 First-to-default and nth-to-default credit derivatives are not recognised for
regulatory purposes as a risk mitigant616. These are contracts providing credit

 Back to contents 96
 CREDIT RISK MITIGATION

protection on a basket of reference entities where the first or nth to default

leads to a payment by the protection seller and the termination of the
contract. The absence of capital relief is a reflection of the fact that such
transactions provide no determinable credit protection against individual
exposures.

6.3.44 Currency mismatches give rise to a haircut on the protection provided 617.

Maturity mismatches

6.3.45 Basel III has specific rules in respect of maturity mismatches (i.e. when
residual maturity of the credit protection is less than that of the underlying
exposure)618. In determining whether an exposure is mismatched, a bank is
required to apply a “worst case” scenario comparing the longest possible
scheduled maturity of the underlying exposure (taking into account any
applicable grace period) with the earliest possible effective maturity of the
hedge (including embedded options or calls that enable the seller to
terminate the cover, or contain a positive incentive for the bank to exercise
a call e.g. step-ups)619.

6.3.46 For financial collateral, any maturity mismatch will not be allowed if the bank
uses the simple approach to collateral 620. In other words, if the duration of
the hedge is less than that of the exposure, the exposure is treated as
uncollateralised.

6.3.47 For banks that use other approaches to financial collateral,. partial
recognition is given to maturity mismatches. The original maturity of the
hedge must originally have been one year or more, and the residual maturity
must equal or exceed three months621. If either of these conditions is not met
the exposure is regarded as unhedged. Where they exist then a simple
formula determines the amount of protection that may be recognised 622.

Pools of credit risk mitigation techniques

6.3.48 Where a bank has multiple credit risk mitigation techniques covering a single
exposure (e.g. a bank takes both collateral and a guarantee), it is required to
subdivide the exposure into notional portions covered by each type of credit
risk mitigant (e.g. the part covered by a guarantee, the part covered by
collateral). The risk weight is then calculated separately for each notional
part of the exposure so covered623.

6.3.49 If credit protection is provided by a single entity, but with different

maturities, the bank must also subdivide the exposure into separate parts 624.

6.4 Credit Risk Mitigation under IRB

6.4.1 For banks that apply the foundation IRB approach the rules on credit risk
mitigation follow closely those applicable under the standardised approach.
However, there are two significant differences. Firstly, the effect of credit
risk mitigation techniques is on the IRB risk components as opposed to on risk
weighted assets directly, as under the standardised approach. The second
difference with the standardised approach is that a wider range of credit risk
mitigation techniques is allowed. However, the range of techniques is

 Back to contents 97
 CREDIT RISK MITIGATION

narrower than that available under the advanced IRB approach (where
available).

6.4.2 Under both the foundation and the advanced IRB approaches, the effect of a
bank holding collateral is to reduce the loss given default (LGD) as the bank
is able to resort to the collateral if its counterparty defaults, thus reducing
the amount of its loss. Guarantees and credit derivatives are treated as
reducing the probability of default (PD) through reducing the probability that
a bank will suffer loss as it has a claim on the protection provider as well as
on the obligor in the event of a default or the loss given default (LGD) of the
transaction with the LGD applicable to the guarantee. On-balance sheet
netting is treated as reducing a bank’s exposure at default (EAD) as it reduces
the extent to which the bank is exposed on the default of a counterparty.

6.4.3 Under the foundation IRB approach the minimum standards that a bank must
meet are generally the same as those for the standardised approach. Thus
the bank will need to satisfy the requirements of legal certainty, low
correlation with the exposure and robust internal risk management. Banks on
the foundation IRB approach may use any of the forms of credit risk mitigation
available under the standardised approach. In addition, banks are allowed to
use additional assets as collateral.

Collateral under the foundation IRB approach

6.4.4 Banks using the foundation IRB approach are able to use both financial
collateral and eligible IRB collateral. Financial collateral is any collateral
eligible under the standardised approach (see above). Eligible IRB collateral
consists of real estate, specified receivables and other physical collateral
meeting specified minimum requirements 625.

Financial collateral

6.4.5 The treatment of financial collateral closely follows that applicable under the
comprehensive approach to collateral for banks on the standardised approach.
Banks that use an IRB approach are not able to apply the simple approach to
collateral626. Under Basel III, a bank is required to calculate the adjusted value
of the collateral through applying haircuts to the gross value of the
collateral627.

Financial receivables

6.4.6 Financial receivables are eligible as collateral provided that they have an
original maturity of one year or less and repayment occurs through
commercial or financial flows related to the underlying assets of the borrower.
Receivables associated with securitisations, sub-participations and credit
derivatives are expressly excluded628. Banks must have an enforceable
security interest over the receivables629 and the bank must have a sound
process for determining credit risk in the receivables 630.

Commercial and residential real estate

6.4.7 Commercial and residential real estate is eligible as collateral for IRB banks
provided that:

 Back to contents 98
 CREDIT RISK MITIGATION

repayment of the secured loan by the borrower is dependent not on the

performance of the underlying property or project, but rather on the
underlying capacity of the borrower to repay the debt from other
sources; and

the value of the collateral is not materially dependent on the

performance of the borrower631.

6.4.8 Income producing real estate that falls within the specialised lending sub-
class of corporate exposures is specifically excluded 632. A fortiori this applies
to high-velocity commercial real estate.

6.4.9 In exceptional circumstances, for well-developed and long-established

markets, secured commercial real estate lending may be treated as collateral
in the corporate exposure portfolio. Two tests must be fulfilled: (i) losses
stemming from commercial real estate lending up to 50% of the market value
or 60% of the LTV must not exceed 0.3% of the outstanding loans in any given
year; and (ii) overall losses stemming from commercial real estate lending
must not exceed 0.5% of the outstanding loans in any given year633.

6.4.10 Commercial and residential real estate is eligible for recognition as collateral
only if:

any collateral taken is legally enforceable in all relevant jurisdictions,

and the bank perfects its security interest in accordance with applicable
law;

the collateral must be valued at or less than the current fair value under
which the property could be sold to an arms’ length purchaser;

the bank monitors the value of the collateral on a frequent basis and at
a minimum once per year; and

(in the case of second and subsequent mortgages) there is no doubt that
the security interest is legally enforceable and constitutes an effective
credit risk mitigant634.

6.4.11 Collateral management requirements are as follows:

the types of commercial and residential real estate collateral accepted

by the bank, and its lending policies, are clearly documented;

the bank takes steps to ensure that the property is adequately insured;

the bank monitors on an ongoing basis the extent of any prior claims on
the property (e.g. tax); and

the bank appropriately monitors the risk of environmental liability, such

as presence of toxic waste635.

 Back to contents 99
 CREDIT RISK MITIGATION

Other physical collateral

6.4.12 National supervisors may allow the recognition in their jurisdictions of certain
other physical collateral if:

the bank demonstrates that there are liquid markets for disposal of the
collateral in an expeditious and economically efficient manner; and

there are well established, publicly available market prices for the
collateral636.

6.4.13 Examples could be aircraft, ships, cars, traded commodities and raw
materials.

6.4.14 The following requirements apply:

the bank has a first mortgage over the collateral;

the loan agreement includes detailed descriptions of the collateral and

the right to examine and revalue the collateral;

the types of physical collateral accepted by the bank, and policies and
practices in respect of the appropriate amount of each type of collateral,
are clearly documented;

bank credit policies address appropriate collateral requirements, the

ability to liquidate the collateral, to establish a price or market value,
and the volatility of the value of the collateral; and

in case of inventories (e.g. raw materials, work-in-process, finished

goods or dealers’ inventories of cars) and equipment, the periodic
revaluation process includes physical inspection of the collateral 637.

6.4.15 Basel III does not continue the approach under Basel II of requiring a
“meaningful” amount of collateral to be taken. Instead, a formula applies 638.
However, the distinction may be more apparent than real given the
application of haircuts to all collateral limiting the ability of such collateral
to reduce the exposure amount. Nonetheless, there is no current requirement
to hold a minimum amount of collateral.

6.4.16 For each collateralised exposure the bank determines the amount that is
considered to be fully collateralised. The LGD of a collateralised exposure is
the exposure-weighted average of the LGD applicable to the unsecured part
of an exposure under a formula 639.

6.4.17 The following table applies640.

Type of collateral LGD for the Haircut

collateralised part of
the exposure
Financial collateral 0% As determined in
accordance with the
comprehensive

 Back to contents 100

 CREDIT RISK MITIGATION

Type of collateral LGD for the Haircut

collateralised part of
the exposure
approach to collateral
under the
standardised approach
Eligible receivables 20% 40%
Eligible commercial 20% 40%
and residential real
estate
Other eligible physical 25% 40%
collateral
Ineligible collateral N/A 100% (i.e. no
recognition)

6.4.18 In cases of a currency mismatch the haircut for each currency is that
applicable under the standardised approach 641.

6.4.19 If an exposure is secured by a pool of collateral that includes both financial

collateral and other eligible IRB collateral then the exposure is notionally split
up into a part secured by financial collateral and the remainder secured by
other form(s) of eligible IRB collateral (receivables, real estate, other
collateral). The part secured by eligible IRB collateral is then subject to the
above treatment642.

6.5 Other Credit Risk Mitigation Techniques under the Foundation IRB Approach

6.5.1 On-balance sheet netting reduces a bank’s exposure at default (EAD). The
requirements for on-balance sheet netting under the foundation IRB approach
are identical to those under the standardised approach 643.

6.5.2 In respect of guarantees and credit derivatives, the foundation IRB approach
closely follows the standardised approach. The minimum conditions and
operational requirements are identical 644. Credit protection is recognised
from the same entities (sovereigns, public sector entities, banks and
corporates) except that unrated companies that are internally rated are also
eligible as credit protection providers 645. The bank then assigns to the
guaranteed exposure the PD grade of an exposure to the guarantor 646. It then
uses this PD grade to calculate the risk weight of the exposure using the risk
weight function appropriate to the guarantor (sovereign, bank or corporate as
the case may be)647.

6.5.3 As an alternative, banks may, instead of modifying the PD, use the LGD of the
guarantee (as opposed to the LGD of the underlying transaction), taking into
account the seniority of the guarantee and any collateral provided to support
the guarantee648.

6.5.4 Any uncovered portion of the exposure is assigned the risk weight appropriate
to the underlying counterparty649. Currency mismatches and partial coverage
are treated in the same way as under the standardised approach650.

 Back to contents 101

 CREDIT RISK MITIGATION

Maturity mismatches

6.5.5 The treatment of maturity mismatches is the same as that under the
standardised approach651. Mismatched hedges with a residual maturity of less
than one year are not recognised. Hedges of over one year receive
proportional recognition.

6.6 Credit Risk Mitigation under Advanced IRB

6.6.1 Under the advanced IRB approach for eligible portfolios banks directly
estimate the probability of default (PD), the loss given default (LGD) and the
exposure at default (EAD) of their exposures. In calculating PD, LGD and EAD
banks are generally permitted to use their own internal estimates of the
effect of credit risk mitigation techniques. As the advanced IRB approach
provides much greater risk sensitivity, through assessing the effect of such
arrangements on each of the drivers of credit risk, a greater range of credit
risk mitigants is recognised.

6.6.2 The effect of collateral being provided is to reduce the loss given default.
Banks must therefore be able to calculate both the LGD and EAD. Banks
unable to do so may not use the advanced IRB approach 652. Where they can
the advanced IRB approach may be used.

6.6.3 The following floors apply to banks’ internal calculation of the effect of fully
collateralised corporate exposures under Basel III653. They do not apply to the
sovereign asset class654 and banks may not use the advanced IRB approach for
bank exposures, or any IRB approach to equity exposures. A separate set of
rules is applicable to retail transactions. The effect of the floor is to prevent
a bank from applying a lower LGD to fully collateralised transactions than that
set out below655.

LGD floors for corporate exposures

Unsecured Secured
25% Financial collateral 0%
Receivables 10%
Commercial or 10%
residential real estate
Other physical 15%
collateral

6.6.4 Where an exposure is partially collateralised, then a formula determines the

floor based on a weighted average of the unsecured LGD floor and the LDG
floor for the collateralised corporate exposure 656.

6.6.5 If a bank is able to use own estimates of LGD for a given pool of exposures,
and takes collateral against one of these exposures, but is unable to model
the effects of the collateral, it may apply the foundation IRB formula provided
that the collateral is eligible under the foundation IRB approach657.

6.6.6 Subject to meeting specified minimum standards, there are no specific

restrictions on the types of collateral (including real estate) that a bank may

 Back to contents 102

 CREDIT RISK MITIGATION

use for regulatory capital purposes, or on the range of providers of guarantees

or credit derivatives. In particular, banks are not required to follow the
“haircuts” approach to financial collateral. Instead, the focus is on ensuring
that the bank’s internal figures are reliable658.

6.6.7 Where collateral is recognised, banks must establish internal requirements for
collateral management, operational procedures, legal certainty and risk
management processes that are generally consistent with those under the
foundation IRB approach659.

6.6.8 The treatment of on-balance sheet netting is the same as under the
foundation IRB approach (although banks will need directly to estimate the
effect of netting in reducing the exposure at default for each facility) 660.

6.6.9 In respect of guarantees and credit derivatives, banks must assess directly the
effect of the guarantee or credit derivative in reducing the probability of
default of the borrower through a reduction in the PD or LGD figures 661. This
requires banks to assign an internal grade to both the borrower and to the
guarantor. As stated above, there is no restriction on the identity of
guarantors, although the set of minimum requirements must be satisfied 662.
Further, banks applying the advanced IRB approach may recognise guarantees
that only cover loss remaining after the bank has first pursued the original
obligor and has completed the workout process663. Banks apply their own
estimates as to the extent of risk transfer. No recognition is given to the
“double default” effect of guarantees and credit derivatives (i.e. the fact that
a bank will only suffer loss if both the underlying obligor and the protection
provider default)664.

6.6.10 Banks are permitted under the advanced IRB approach to adjust the LGD
grade. There are two alternatives. The first is based on the foundation IRB
approach and involves the bank replacing the LGD of the underlying
transaction with the LGD applicable to the guarantee. The second option
involves the bank making an adjustment to its own LGD estimate of the
exposure to reflect the presence of the guarantee or credit derivative 665. In
this case, there is no limit on the range of eligible guarantors 666. First-to-
default, but not other nth-to-default, credit derivatives may be recognised
under the advanced IRB approach667.

6.7 Credit Risk Mitigation for Retail Exposures

6.7.1 There is no foundation IRB approach for retail exposures. Banks are therefore
required to provide their own internal estimates of the effect of credit risk
mitigation in determining both the probability of default and the loss given
default. As has been mentioned, there are floors for both PD and LGD
estimates668. The risk-reducing effect of guarantees and credit derivatives is
effected either through an adjustment to the PD or the LGD estimate 669,
although adjustments must be done in a consistent manner 670. Netting is
recognised subject to the same conditions as under the standardised
approach671.

 Back to contents 103

 CREDIT RISK MITIGATION

6.8 Purchased Receivables

6.8.1 Credit risk mitigation in the form of guarantees and credit derivatives is
recognised in the same way as under the foundation or advanced IRB
approach. In particular, guarantees provided by the seller covering either
default risk, dilution risk, or both are recognised 672. The following treatment
applies. If the guarantee covers both default risk and dilution risk then the
bank should substitute the risk weight of the exposure for that of the
guarantor. If it covers only one of these elements, then the substitution
approach applies to that capital charge, and the normal charge for the other
is added673. Proportional and tranched cover is addressed in the same way as
under the normal IRB approach674. At national discretion, unrated guarantors
internally rated and associated with a PD equivalent to A- may be recognised
under the foundation IRB approach675. This seems an anomalous survival of
the Basel II IRB approach which has not been carried forward elsewhere in the
Basel III text for guarantors and providers of credit protection.

 Back to contents 104

 SECURITISATION

7. SECURITISATION

Securitisation was at the heart of the global financial crisis and it was therefore
inevitable that it would see urgent reform in its aftermath. The process for developing
a new capital framework for securitisations was begun soon after the crisis with the so-
called Basel 2.5 package of reforms, while the Committee decided how to replace the
Basel II securitisation framework. The new framework came into force in 2018,
although further refinements apply from 1 January 2023. This reflects the iterative
process followed by the Basel Committee in first addressing, on an interim basis, certain
clear failings in the Basel II securitisation framework, whilst then developing and
refining a new standard under Basel III. This chapter accordingly describes the fully
developed securitisation framework applicable from 2023, without considering
intermediate approaches. Given that the Basel III framework bears little comparison
with the Basel II approach, only limited discussion of changes will be given.

The securitisation standard differentiates between so-called simple, transparent and

comparable (STC) securitisations, and other securitisations when applying risk weights.
Non-STC securitisations are further divided into ordinary securitisations and re-
securitisations, the latter being a securitisation of a pool of assets, at least one of which
is itself a securitisation. Securitisations of securitisations (such as the now notorious
collateralised debt obligations, or CDOs) were involved in a significant part of the losses
suffered by banks in the global financial crisis.

7.1 Introduction

7.1.1 Companies and banks may securitise assets for a variety of purposes. These
include reducing their regulatory capital requirements through transferring
assets off their regulatory balance sheet, obtaining additional funding,
improving financial ratios, managing portfolio risk and diversifying their
portfolio.

7.1.2 Banks may become involved in a securitisation at different stages of the

process. Firstly, the bank may act as an originator. In this case, the bank
transfers assets off its own balance sheet to a SPV (traditional securitisation)
or purchases eligible credit protection (synthetic securitisation). Additionally,
an originator may provide credit enhancements or liquidity facilities to the
SPV. Secondly, a bank may be an investor in the securitisation by purchasing
asset-backed securities or credit-linked notes issued by the SPV. Thirdly, a
bank may act as sponsor. In this case the bank may provide credit
enhancements and/or liquidity facilities to the SPV. The bank may also assist
in marketing the securities to investors. Fourthly, a bank may be a servicer if
it manages the underlying credit exposures of a securitisation on a day-to-day
basis through collecting principal and interest and paying it to investors.
These categories are not mutually exclusive and a bank may, for example, act
as both an originator and a servicer.

7.2 Background

7.2.1 The December 2012 consultative document published by the Basel Committee
identified the following deficiencies in the Basel II framework:

 Back to contents 105

 SECURITISATION

mechanistic reliance on external ratings;

too low risk weights for highly-rated securitisations;

too high risk weights for low-rated securitisations; and

cliff-effects in capital requirements following a deterioration in credit

quality of the underlying pool676.

7.2.2 Basel III is based on a pre-defined hierarchy of methods. At the top of the
pyramid is the securitisation internal ratings-based approach, that a bank
must use if able. If not, there is a securitisation external ratings-based
approach that must be used by banks in jurisdictions that permit the use of
external ratings. Below this is the securitisation standardised approach.

7.2.3 This chapter will firstly set out the differences between STC securitisations,
other securitisations and re-securitisations. The second category is residual
in that it encompasses any securitisation that is neither an STC securitisation
nor a re-securitisation. Then it will describe the standardised approach, the
external ratings-based approach and the internal ratings-based approach.
This is for ease of understanding. However, it must be stressed that the actual
hierarchy under Basel III is the opposite, so a bank that can use either the
internal ratings-based approach or the external ratings-based approach must
use those approaches in preference to the standardised approach 677. Any
securitisation exposure to which a bank cannot apply any of the foregoing
approaches is assigned a 1250% risk weighting (equivalent to a deduction from
capital)678.

7.3 Types of Securitisations

7.3.1 There are three types of securitisations under the Basel III framework: STC
securitisations, re-securitisations and other securitisations. Both re-
securitisations and other securitisations may take the form of either a
traditional securitisation or a synthetic securitisation (synthetic
securitisations are expressly excluded from the definition of an STC
securitisation).

7.3.2 Banks must use the securitisation framework for determining regulatory
capital requirements on exposures arising from all securitisations, as well as
other tranched exposures. As securitisations may be structured in different
ways, Basel III requires the capital treatment to be determined based on its
economic substance and not legal form. Banks that are in doubt whether a
particular transaction is a securitisation are encouraged to consult with their
national supervisors679.

7.3.3 Securitisation exposures can include asset-backed securities, mortgage-

backed securities, credit enhancements, liquidity facilities, interest rate or
currency swaps, credit derivatives and tranched cover. Reserve accounts,
such as cash collateral accounts, recorded as an asset by the originating bank
must also be treated as securitisation exposures680.

 Back to contents 106

 SECURITISATION

Traditional securitisations

7.3.4 A traditional securitisation is a structure where the cash flow from an

underlying pool of exposures is used to service at least two different stratified
risk positions or tranches reflecting different degrees of credit risk. Payments
to the investors depend upon the performance of the specified underlying
exposures. The tranched structure that characterises traditional
securitisations differs from ordinary senior/subordinated debt instruments in
that junior tranches absorb losses without interrupting contractual payments
to more senior tranches, whereas subordination is only applicable in a winding
up681.

Synthetic securitisations

7.3.5 A synthetic securitisation is a structure with at least two different stratified

risk tranches that reflect different degrees of credit risk where the credit risk
of an underlying pool of exposures is transferred, in whole or in part, through
the use of funded (e.g. credit-linked notes) or unfunded (e.g. credit default
swaps) credit derivatives or guarantees that hedge the credit risk on the
portfolio682.

Re-securitisations

7.3.6 A re-securitisation is a securitisation exposure in which the risk associated

with an underlying pool of exposures is tranched and at least one of the
underlying exposures is itself a securitisation exposure. In addition, an
exposure to one or more re-securitisation exposures is also a re-securitisation
exposure. However, the re-tranching of a securitisation exposure does not
make the exposure into a re-securitisation exposure if the bank is able to
demonstrate that the cash flows to and from the bank could be replicated in
all circumstances by an exposure to the securitisation of a pool of assets that
contains no securitisation exposures683.

7.4 Definitions

7.4.1 The following definitions apply under the Basel III framework:

An originator is a bank if it acts in one of the following two ways:

it originates, directly or indirectly, underlying exposures included

in the securitisation; or

it is the sponsor of an asset-backed commercial paper (ABCP)

conduit or similar programme that acquires exposures from third
party entities. A bank is treated as a sponsor if it manages or
advises the programme, places securities into the market, or
provides liquidity facilities or credit enhancements 684.

An ABCP programme predominantly issues commercial paper to third-

party investors with an original maturity of one year of less, and is backed
by assets or other exposures held in a bankruptcy-remote, special-
purpose entity685.

 Back to contents 107

 SECURITISATION

A credit enhancement is a contractual arrangement in which the bank

or other entity retains or assumes a securitisation exposure and, in
substance, provides some degree of added protection to other parties to
the transaction686.

A credit-enhancing interest-only strip (I/O) is an on-balance sheet asset

that represents a valuation of cash flows related to future margin income
and is subordinated687.

A liquidity facility is undefined in the Basel III framework. The Basel II

definition was, basically, an arrangement under which a bank provided
short term liquidity to an SPV to facilitate the payment of interest to
investors. This corresponds with market practice and would seem to
continue to apply under Basel III. However, it should be stressed, that
the beneficial treatment of certain liquidity facilities under Basel II has
been withdrawn.

A clean-up call is an option that permits the securitisation exposures to

be called before all of the underlying exposures, or securitisation
exposures, have been repaid. In the case of traditional securitisations,
this is generally accomplished by repurchasing the remaining exposures
once the pool balance or outstanding securities have fallen below a
specified level. For synthetic securitisations it can take the form of a
clause that extinguishes the credit protection 688.

An early amortisation provision is a mechanism that, once triggered,

accelerates the reduction of the investor’s interest in the underlying
exposures of a securitisation of revolving credit facilities, and allows
investors to be paid out prior to the original maturity of the securities
issued689.

A securitisation of revolving credit facilities is a securitisation in which

one or more underlying exposures represents, directly or indirectly,
current or future drawings on revolving credit facilities. Examples
include securitisations of credit cards, home equity lines of credit,
commercial lines of credit and other lines of credit 690.

The express spread (or future margin income) is defined as gross finance
charge collections and other income received by the SPV less interest,
servicing fees, charge-offs and other senior expenses691.

Implicit support arises when a bank provides support to a securitisation

in excess of its predetermined contractual obligation692. Common
examples include repurchasing exposures from the pool above current
market value to protect investors in the securities from losses, or
providing additional credit support after the inception of the
securitisation.

A tranche is considered to be senior if it is effectively backed or secured

by a first claim on the entire amount of the assets in the underlying
securitised pool. While this generally includes only the most senior
position within a securitisation transaction, in some instances there may
be other claims that are senior in the waterfall e.g. a claim under a swap.

 Back to contents 108

 SECURITISATION

Where several tranches have different maturities, but share pro rata loss
allocation without affecting their seniority, then they may be treated as
being of the same seniority693. While these clarifications are welcome,
the effect is that in any securitisation, only the most senior tranche
(which may be an unrated super-senior tranche) will count as being senior
with all other tranches regarded as subordinated.

7.4.2 Basel III provides the following illustrations of this:

in a traditional securitisation with all tranches above the first loss piece
being rated, only the most highly rated position qualifies as the senior
tranche694;

in a synthetic securitisation, an unrated tranche is treated as the senior

tranche provided that all the conditions for inferring a rating from a
lower tranche that meets the definition of a senior tranche are met 695;
and

a liquidity facility supporting an ABCP programme is treated as senior if

it covers all losses on the underlying receivables pool that exceed the
amount of overcollateralisation/reserves provided by the seller if it is
sized to cover all of the outstanding commercial paper and other senior
debt supported by the pool, so that no cash flows from the underlying
pool could be transferred to the other creditors until any liquidity
facilities were repaid in full696.

Exposures

7.4.3 Basel III applies capital charges to exposures. In calculating the size of an
exposure banks must determine the sum of the on-balance sheet exposure, or
carrying value, taking into account purchase discounts, write-downs, specific
provisions and the off-balance sheet exposure (if relevant)697.

7.4.4 Off-balance sheet exposures are determined as follows:

for credit risk mitigants sold or purchased by a bank the treatment is as

for credit risk mitigation (see below)698;

other facilities are subject to a 100% credit conversion factor;

in the case of servicer cash advances, the undrawn portion that is

unconditionally cancellable without prior notice receives the credit
conversion factor for unconditionally cancellable facilities under the
standardised approach699. This is 10%700.

7.4.5 An SPV (the actual term used in the Basel III text is an SPE or special purpose
entity) is a corporation, trust or other entity organised for a specific purpose,
the activities of which are limited to those appropriate to accomplish the
purpose of the SPV, and the structure of which is intended to isolate the SPV
from the credit risk of an originator or seller of exposures 701 i.e. it is
bankruptcy remote.

 Back to contents 109

 SECURITISATION

7.4.6 Maturity is the tranche’s remaining effective maturity in years. Basel III
allows banks to measure maturity in either of the following two ways, with a
floor of one year and a cap of five years702:

the weighted average703 of the contractual cash flows of the tranche

calculated as follows:

∑𝑡 + 𝐶𝐹𝑡
𝑀𝑇 = ∑𝑡 𝐶𝐹𝑡

Where Mt is the maturity of the tranche and CFt denotes the cash flows
(principal, interest and fees) payable by the borrower in period t. Where
unconditional conditional payment dates are unavailable the final legal
maturity must be used704; or

on the basis of the final legal maturity of the tranche:

𝑀𝑇 = 1 + 80% (𝑀𝐿 − 1)

where ML is the final legal maturity of the tranche 705.

7.4.7 When determining the maturity of a securitisation exposure, banks should

take into account the maximum period of time they are exposed to potential
losses from the securitised assets. Where a bank provides a commitment, the
bank should calculate the maturity of the exposure as the sum of the
contractual maturity of the commitment and the longest maturity of the
assets that might be added during the revolving period, as opposed to that of
the assets currently in the pool706 (i.e. in a worst case scenario).

7.4.8 The same treatment applies to other instruments where the risk of the
commitment/protection provider is not limited to losses realised until the
maturity of the instrument (e.g. total return swaps) 707. Credit protection only
exposed to losses occurring up to the maturity of the instrument can be
limited to the maturity of the contractual instrument without regard to the
maturity of the protected position708.

7.5 Operational requirements for the recognition of a securitisation

7.5.1 The requirements differ depending on whether the securitisation is a

traditional or a synthetic securitisation.

7.5.2 An originator (including persons treated as an originator under the

securitisation framework) must satisfy all of the following in order for the
securitisation to be recognised for regulatory capital purposes as a traditional
securitisation:

significant credit risk is transferred to third parties (the amount of which

is undefined, unlike in EU law);

the transferor does not maintain effective or indirect control over the
transferred exposures. A bank is deemed to have retained effective
control if:

 Back to contents 110

 SECURITISATION

it is able to repurchase previously transferred exposures in order

to realise their benefits; or

it is required to retain the risk of transferred exposures;

the exposures are legally isolated from the transferor in a way that puts
the exposures beyond the reach of the transferor and its creditors. Basel
III refers to both a sale of the assets and a sub-participation, although
under English law a sub-participation only transfers the economic interest
and provides no protection in the event of the transferor’s insolvency.
For this reason, securitisations are not commonly structured in England
as a sub-participation;

the bank obtains a legal opinion confirming a true sale. Under Basel III
this can be either an external legal opinion, or written advice from an in-
house lawyer;

the securities are not obligations of the transferor;

the transferee is an SPV and the holders of the beneficial interests in the
SPV have the right to pledge or exchange those interests without
restriction (save in the case of mandatory risk retention requirements).
In England, an SPV is commonly a bankruptcy-remote entity, separate
from the originating bank, but not owned by the holders of the securities
who have a proprietary interest instead in the transferred pool of
exposures (e.g. it may be owned by a charitable trust);

any clean-up calls meet specified criteria (see below);

there are no clauses in the securitisation documentation that:

require the originating bank to alter the pool of underlying

exposures to improve the credit quality of the pool;

allow for increases in any retained first loss position or credit

enhancements provided by the originating bank after the inception
of the securitisation; or

increase the yield payable to other parties (such as investors in

the securities sold, or third party providers of credit
enhancements) in response to a deterioration in the credit quality
of the pool of exposures.

(All of these constitute implicit support prohibited under Basel III); and

there are no termination options or triggers except permitted clean-up

calls, tax or regulatory calls, or permitted early amortisation
provisions709.

7.5.3 The requirements for a synthetic securitisation are the following:

the credit protection satisfies the requirements for a credit derivative or

a guarantee under the standardised approach to credit risk mitigation

 Back to contents 111

 SECURITISATION

(unfunded credit protection) or collateral (funded credit protection, such

as credit-linked notes);

eligible providers of credit protection are limited to those recognised

under the standardised approach i.e. sovereigns, public sector entities,
multi-lateral development banks, banks, securities firms, other
prudentially regulated firms, corporates meeting certain criteria, and
other group companies;

collateral is limited to that recognised under the simple approach to

collateral (including such collateral if provided by the SPV);

significant credit risk (which is undefined) is transferred to third parties;

the securitisation documentation contains no clauses that limit the

amount of credit risk transferred, such as any of the following (the list is
non-exhaustive):

clauses that materially limit the credit protection or credit risk

transferred. Basel III gives the following examples:

(A) early amortisation provisions that effectively subordinate

the bank’s interest;

(B) significant materiality thresholds below which the credit

protection cannot be triggered if a credit event occurs; or

(C) clauses that allow for the termination of the credit

protection if the credit quality of the pool of exposures
deteriorates;

clauses that require the originating bank to alter the underlying

pool of exposures to improve the pool’s credit quality;

clauses that increase the bank’s cost of credit protection in

response to a deterioration in the credit quality of the pool;

clauses that increase the yield payable to third parties, such as

investors or providers of credit enhancements, if the credit quality
of the pool deteriorates; or

clauses that provide for increases in a retained first-loss position

or credit enhancement provided by the originating bank after the
inception of the securitisation;

the bank obtains a legal opinion that confirms the enforceability of the
credit protection contract; and

any clean-up calls satisfy the requirements set out below710.

 Back to contents 112

 SECURITISATION

Early amortisation provisions

7.5.4 There are detailed rules to address risks arising out of early amortisation
provisions. This is in response to evidence obtained by the Basel Committee
that securitisations of revolving exposures with both “controlled” and
“uncontrolled” early amortisation provisions (under the Basel II framework)
typically result in very limited, if any, transfer of credit risk to investors. Basel
III therefore treats certain revolving securitisations with early amortisation
provisions as automatically on-balance sheet for regulatory purposes, as well
as setting out operational requirements for early amortisation provisions if a
securitisation is to be recognised for regulatory capital purposes.

7.5.5 A securitisation is automatically deemed to fail to meet the requirements for

traditional or synthetic securitisations (and therefore will not be recognised
for regulatory purposes) if:

the securitisation is of one or more revolving credit facilities; and

there is an early amortisation or other similar provision which, if

triggered, would:

subordinate the bank’s senior or pari passu interest in the

underlying credit facilities to the interests of other investors;

subordinate the bank’s subordinated interest to an even greater

degree relative to the interests of other parties; or

in other ways, increase the bank’s exposure to losses associated

with the underlying credit facilities711.

(All of these examples involve the provision of credit enhancements to

investors).

7.5.6 Securitisations that are not specifically excluded by the preceding paragraph,
and which meet all the other criteria for a traditional or a synthetic
securitisation, may attract regulatory capital relief if they contain an early
amortisation provision, and the securitisation falls within one of the following
categories:

replenishment structures where the exposures do not revolve, and early

amortisation ends the ability of the bank to add new exposures;

securitisation structures of revolving credit facilities containing an early

amortisation provision that mimic term structures (i.e. where the risk on
the underlying revolving credit facilities does not return to the bank)
and where the early amortisation provision does not effectively result in
subordination of the originator’s interest;

structures where a bank securitises one or more revolving facilities and

where investors remain fully exposed to future drawdowns by borrowers
after the early amortisation event occurs; or

 Back to contents 113

 SECURITISATION

the early amortisation provision is triggered solely by events not related

to the performance of the underlying assets or the selling bank e.g. a
material change in tax law 712.

7.5.7 In all of these cases the early amortisation provision does not have the
economic effect of accelerating the reduction of the investors’ interest in the
securitised pool allowing investors to be repaid prior to the original maturity
of the securities.

Clean-up calls

7.5.8 A clean-up call is a call option that permits securitisation exposures to be

called before all of the underlying exposures or securitisation exposures have
been repaid. In a traditional securitisation, this is normally done by
repurchasing the remaining assets in the pool, or the outstanding securities,
once the pool size or outstanding securities falls below a specified amount.
In the case of a synthetic securitisation, it may take the form of a clause
extinguishing the credit protection713. In both cases the intention is to
mitigate the costs of retaining a securitisation in place where this is no longer
economically viable as the size of the pool, or outstanding securities, is simply
too small.

7.5.9 Where a securitisation includes a clean-up call the following conditions apply.
If they are not met, no regulatory capital recognition will be given to the
securitisation (i.e. it is treated as unsecuritised)714. Firstly, the exercise of the
call must not be mandatory, but at the discretion of the originating bank.
Secondly, the clean-up call must not be structured to avoid allocating losses
to investors, providers of credit enhancements or otherwise improve the
position of third parties. Thirdly, the clean-up call can only be exercised when
10% or less of the original portfolio or securities remains outstanding
(traditional securitisation, or credit-linked notes) or when 10% or less of the
original reference portfolio remains (synthetic securitisation)715.

7.5.10 Clean-up calls that, when exercised, provide credit support are treated as the
provision of implicit support. This would seem to cover any case where there
has been a deterioration in the credit quality of the remaining pool since
inception where a clean-up call is actually exercised716. Whether this is
intended to be assessed fully on an ex post facto basis is unclear and national
supervisors are responsible for implementation of this provision.

7.6 Due Diligence

7.6.1 Given the role of securitisation in the global financial crisis, inevitably new
rules would be promulgated on due diligence, and the consequence of not (or
being unable to) perform such due diligence. Under Basel III if the following
requirements are not met then all exposures of a bank attract a 1250% risk
weight717, which is calibrated to equate to a deduction from capital.

7.6.2 Banks must, as a general rule, have a comprehensive understanding of the risk
characteristic of individual securitisation exposures, whether on- of off-
balance sheet, as well as of the pools underlying those exposures 718. Greater
understanding is therefore required of the bank’s own exposures than the
underlying assets that are securitised reflecting inherent data limitations on

 Back to contents 114

 SECURITISATION

pool exposures. That said, the requirements on originators imposed by Basel

III are intended to ensure that market practice develops so that non-regulated
originators will be unable to sell/transfer securitisation exposures to
regulated entities as a result of the 1250% risk weight.

7.6.3 Banks must be able to access performance information on the underlying pools
of exposures on an on-going basis in a timely manner. As appropriate, this
may include exposure type, percentage of loans 30, 60 and 90 days past due,
default rates, pre-payment rates, loans in default, type of property, etc.719.
In the case of re-securitisations (such as CDOs) the bank must have
information not only on the underlying securitisation exposures, but also on
the characteristics and performance of the asset pools underlying those
securitisations720. This reflects the fact that the performance of a CDO is
dependent both on the characteristics of the securitisation exposures as well
as the assets underlying those securitisations. Logically, exposures to a CDO
squared should require the bank to understand the characteristics, etc. of
both levels of securitisation, as well as the asset pools.

7.6.4 Banks must have a thorough understanding of all structural features of a

securitisation exposure that would materially impact the performance of the
exposures, including the contractual waterfall, waterfall-related triggers,
credit enhancements, liquidity enhancements, market value triggers and
deal-specific definitions of default 721.

7.7 Capital Requirements

7.7.1 Banks must hold capital against risks in all securitisation-related exposures.
These can include retained tranches, investments in asset-backed securities,
provision of credit enhancements or of a liquidity facility. Where a bank
purchases securitisation exposures securitised by it, it must treat such
exposures as retained securitisation exposures 722 as there is no risk transfer.

7.7.2 Risk-weighted assets on a securitisation exposure are calculated by

multiplying the exposure amount by the appropriate risk weight determined
by the hierarchy of approaches (IRB, ERBA and SA), applying applicable caps
on exposures (see below)723.

Caps on securitisation exposure capital charges

7.7.3 This section sets out capital adequacy caps applicable to securitisation
exposures. The intention is to avoid any unduly onerous capital treatment as
a result of the application of the securitisation framework in specific
situations. It does not apply to re-securitisations owing to the greater risks in
such securities.

7.7.4 The stated purpose of these caps is to continue the Basel II treatment that a
bank should not have to hold more capital against a securitisation exposure
than it would have had to do had the exposures not been securitised. As the
December 2013 consultation document Revisions to the Securitisation
Framework put it:

“An originator can view securitisation as similar to credit risk mitigation, as

at least some of the risk of the underlying exposure(s) is transferred to

 Back to contents 115

 SECURITISATION

another party. On this view, it would be inappropriate for a bank to be

compelled to hold more capital after securitisation than before. Supporting
this concept is the operational requirement that significant risk be transferred
in order to recognise any benefits from a securitisation for originators and
sponsors.

Following this rationale, the Committee initially proposed to apply an overall

cap to capital requirements not only to IRB banks …, but also to all originators
and sponsors, even if they cannot calculate IRB inputs for the underlying pool.

The Committee proposes that the overall cap be applicable regardless of the
approach that is applied: Internal Ratings-Based Approach, External Ratings-
Based Approach or Standardised Approach.

In the case mixed pools it would be inappropriate for a bank to be compelled

to hold more capital after securitisation than before, as its risk should be
reduced through the process of securitisation, subject to the significant risk
transfer requirement. The implication is that the overall cap should be
calculated by adding up the capital before securitisation; that is, by adding
up the capital required under the general credit risk framework for the IRB
and SA part of the pool”724.

7.7.5 In the case of “senior” securitisation exposures (which, basically, means only
the most senior tranche in the waterfall) banks are allowed to apply a “look
through” approach under which the securitisation exposure receives a capital
charge applicable to exposure weighted-average risk weight applicable to the
underlying exposures, provided the bank has knowledge of the composition of
the underlying exposures at all times. An example would be an originating
bank that retains a super-senior tranche ranking higher than any sold and
rated tranches.

7.7.6 In this case the risk weight under the IRB approach must be calculated after
taking into account expected losses. If a bank uses exclusively the
standardised approach or the IRB approach then the cap for senior exposures
is the exposure-weighted average risk weight applicable to exposures under
the standardised or IRB framework, as applicable. Where there are “mixed
pools”, the IRB portion of the pool receives the IRB risk weight and the
standardised pool the SA risk weight725. The Basel III text does not mention
either the ERBA or SEC-IAA approaches, although it would be logical to apply
these methods, where available, and relevant to determine the capital
charge.

7.7.7 If the risk weight derived from applying the cap is less than that from the 15%
floor (see below) then the risk weight deriving from the cap applies 726.

7.7.8 If a bank uses the IRB approach under the hierarchy of methodologies then
the bank is able to cap its capital charge for its exposures to a securitisation
(including expected losses) at the IRB capital charge had the exposures not
been securitised727.

7.7.9 If a bank uses the ERBA or SA approach to a securitisation exposure, then the
maximum capital charge is equal to that applicable to the underlying
exposures had they not been securitised.

 Back to contents 116

 SECURITISATION

7.7.10 For “mixed pools”, the same approach applies. A bank is not required to apply
a capital charge for a securitisation exposure that is higher than the applicable
capital charge before securitisation calculated under the standardised and IRB
approaches. The IRB portion must include expected losses 728. In respect of
mixed pools applying both the ERBA and SA approaches the Basel III text caps
the capital charge at that calculated under the IRB and SA approaches 729.

7.7.11 The maximum aggregated capital charge for a bank’s securitisation exposures
in the same transaction is equal to KP x P. For this purpose KP equals the
capital charge for an underlying pool of exposures, and P is the largest
proportion of interest that a bank holds for each tranche of any given pool.

7.7.12 The following clarifications apply:

For a bank that has one or more securitisation exposure(s) that are in a
single tranche of a given pool, P is equal to the proportion (as a
percentage) that the bank holds in that tranche calculated as the total
nominal amount of the bank’s exposures in the tranche, divided by the
nominal amount of the tranche730.

For a bank that has securitisation exposures that are in different tranches
of a given securitisation, P equals the maximum proportion of interests
across tranches731.

The capital charge for the underlying pool (KP) is determined as follows.
For an IRB pool, it is the capital charge had the IRB pool not been
securitised (KIRB). For a pool under the standardised approach it is the
capital charge under the standardised approach (KSA)732. The ERBA and
SEC-IAA approaches are not mentioned.

The treatment of mixed pools is based on an exposure-weighted average

of the capital charge under the ERBA and SA approaches 733. It is unclear
whether the exclusion of the SEC-IAA approach is intended or accidental.

In applying the capital charge cap, the entire amount of any gain on sale
and credit enhancing interest-only strips is deducted734.

7.8 Hierarchy of Methods

7.8.1 Unlike Basel II, Basel III now has a prescriptive hierarchy. A bank must use the
IRB approach to securitisation exposures of an IRB pool unless instructed
otherwise by its supervisor735. If a bank is unable to use the IRB approach (e.g.
if it does not have permission to use an IRB approach), it must use the
securitisation external ratings-based approach, if its national supervisor
allows the ERBA approach to be used, and if there is an external rating that
meets the operational requirements 736. If the national supervisor allows the
use of external ratings, an internal assessment approach (SEC-IAA) is also
available for unrated exposures to a pool within a rated ABCP programme.
This may be relevant for unrated credit enhancements and liquidity facilities.
A precondition of using the SEC-IAA is that the bank has supervisory approval
to use an IRB approach for non-securitisation exposures737.

 Back to contents 117

 SECURITISATION

7.8.2 Banks that cannot use the IRB, ERBA or SEC-IAA must use the standardised
approach (SEC-SA)738. If they cannot, then a 1250% risk weight applies 739.

7.8.3 If a bank cannot apply the IRB approach to 95% of all the underlying exposures,
then the normal (non-IRB) hierarchy applies (ERBA/SEC-IAA, SA or 1250% risk
weight)740.

Overlapping exposures

7.8.4 A specific treatment applies to so-called “overlapping” exposures. Exposures

A and B overlap if and only if the performance by the bank of its obligations
in respect of exposure A will preclude in all circumstances a loss on exposure
B. An example is a bank providing credit support to a tranche of notes and
holding some of the notes. In this case, if the bank performs its obligations
there will be no loss on its own holding of the notes. In this case only exposure
A is risk-weighted and exposure B is disregarded741.

7.8.5 In identifying overlaps, banks are allowed to split or expand exposures to

notionally derive an overlap 742. Splitting means dividing up the actual
exposure to two notional ones, one that is treated as overlapping, in the sense
referred to above, and the other as not overlapping. An exposure may be
“expanded” by assuming the exposure is larger than it contractually is, such
as by expanding notionally the trigger events or the size of the credit
enhancements or liquidity facility provided, when determining the capital
treatment743.

7.8.6 An example of this is given in Basel III. A liquidity facility may not
contractually cover defaulted assets, or may not fund an ABCP programme in
certain circumstances. For regulatory capital purposes this would not
constitute an overlap with notes held by the bank issued by the ABCP conduit.
However, a bank may calculate the capital charge for the liquidity facility as
if it were expanded to cover defaulted assets, to preclude losses on the notes,
and then only calculate a capital charge on the facility (and not the notes
held)744.

7.8.7 A bank may recognise overlaps between the trading and banking books,
provided the bank can calculate the relevant capital charges and compare
them745.

7.9 Re-securitisation Exposures

7.9.1 The Basel Committee determined that as a result of the financial crisis neither
external ratings nor an IRB approach could be used for re-securitisation
exposures as neither could accurately address the risks in such complex
transactions. As a result, only the standardised approach may be used,
although adjustments apply to take account of the higher risks on such
exposures746.

7.10 Non-Performing Securitisations

7.10.1 A bespoke capital treatment applies to securitisations of non-performing loans

(see below).

 Back to contents 118

 SECURITISATION

7.11 Implicit Support

7.11.1 Implicit support arises when a bank provides support to a securitisation

exposure beyond its contractual obligations after the inception of the
securitisation. Basel III requires capital to be held against all of the underlying
exposures associated with the securitisation as if they had not been
securitised747.

7.12 Recognition of Credit Risk Mitigation

7.12.1 Under all approaches credit risk mitigation may be recognised.

7.12.2 If the bank applies the IRB approach then collateral recognition limited to
that available under the foundation IRB approach 748. Guarantees and credit
derivatives must comply with the standardised approach to credit risk
mitigation749. Advanced IRB collateral, or IRB-recognised sellers of credit
protection may not be recognised. A partial exception to this allows collateral
pledged by SPVs to be recognised750.

7.12.3 For banks applying the standardised approach, understandably, only collateral
and credit derivatives/guarantees recognised under the standardised
approach are eligible751.

7.12.4 Where a bank provides full, or pro rata, credit protection to a securitisation
exposure, it is treated as directly holding the securitisation exposure for
which it provides credit protection for regulatory capital purposes 752.

7.12.5 Where a bank purchases credit protection then it may recognise this
protection under the credit risk mitigation framework (i.e. the standardised
and foundation IRB approaches)753.

7.12.6 Tranched credit protection is notionally split into protected and unprotected
sub-tranches. The protection provider calculates its capital requirement on
the basis of a direct exposure to the particular sub-tranche in accordance with
the hierarchy set out in the next paragraph. Credit protection buyers
determine their capital requirement in accordance with the credit risk
mitigation framework for the protected sub-tranche, and as normal under the
specified hierarchy of approaches for unprotected exposures 754.

7.12.7 The hierarchy of approaches is as follows. If the bank is required to use the
IRB approach, or the standardised approach, then that approach must be
used755. Where the ERBA approach is used the sub-tranche of highest seniority
attracts the risk weight applicable to the original securitisation exposure.
Lower-ranking tranches receive the risk-weight applicable from an inferred
rating if this is possible. Otherwise, the standardised approach applies with
modified parameters756. A lower ranking sub-tranche is always regarded non-
senior even if the original tranche was senior 757.

7.13 Simple, Transparent and Comparable (STC) Securitisations

7.13.1 This category of securitisations benefits from lower capital charges as

securitisations falling within the definition of STC securitisations are
considered to present less risk. There are therefore significant potential

 Back to contents 119

 SECURITISATION

advantages of structuring a securitisation as being simple, transparent and

comparable (STC), where possible. However, some structures (such as
synthetic securitisations) can never be a STC securitisation, and in other cases
there may be commercial considerations that preclude the securitisation from
being an STC securitisation. Apart from conferring regulatory capital benefits,
all the other rules applicable to securitisations continue to apply in full.

7.13.2 STC securitisations include both non-ABCP traditional securitisations that

meet specified criteria and ABCP conduits and/or transactions financed by
such conduits that meet certain criteria 758. The criteria differ, although there
is a degree of overlap.

7.13.3 Originators or sponsors of a STC securitisation must reveal to investors “all

necessary information” to enable investors to determine whether or not the
securitisation is STC compliant. Investor banks are then required to make
their own assessment of whether the securitisation is compliant 759. This is
important as the STC requirements apply to all securitisations, regardless of
whether the originator/sponsor is a bank, corporate or other entity 760.
However, where a bank acts as the originator and retains certain tranches of
the securitisation, then the originator will determine whether or not the
securitisation is an STC securitisation761.

7.13.4 In principle, the STC criteria must be met at all times, although the nature of
the criteria is such that some will only need to be checked at origination.
Investors are expected to take into account developments that might
invalidate a compliance assessment 762, such as failure to provide investor
reports, or changes to the underlying documentation that make the
securitisation non-STC compliant763.

7.13.5 It will be seen that many of the Basel III STC criteria are either qualitative or
open-textured and incapable of precise definition. It follows that in some
cases it may prove difficult to assess whether a given securitisation is or is not
STC compliant. In this case it may fall to national supervisors when providing
guidance to assess whether a particular securitisation is STC compliant. A
further aspect of the criteria requiring experience with securitisations may
exclude new entrants for a period of time from being able to apply the STC
risk weights regardless of whether or not the securitisation would otherwise
have been STC compliant.

Criteria for traditional securitisations

7.13.6 Homogeneity. The assets underlying the securitisation should be credit

claims or receivables that are homogenous. In making this assessment,
consideration should be given to asset type, jurisdiction, legal system and
currency. The assets should have contractually identified periodic payment
streams, and interest payments should reference commonly encountered
market interest rates764.

7.13.7 In assessing homogeneity, the following principles should be applied:

the asset pool should be such that investors do not need to analyse and
assess materially different legal and/or credit risk factors;

 Back to contents 120

 SECURITISATION

the assessment of homogeneity should be based on common risk drivers;

claims or receivables should have standard obligations that result in a

periodic and well-defined stream of payments to investors; and

the repayment of securities should rely mainly on the proceeds of the

securitised assets, as opposed to refinancing or resale of assets in the
pool765.

7.13.8 Asset performance history. Verifiable loss performance data, such as

defaults, should be available for assets 766 with substantially similar risk
characteristics to those being securitised for a sufficient period of time to
permit meaningful evaluation. The minimum performance history required is
five years for retail securitisations and seven years for all other
securitisations767. The sources of, and access to, data should be made
available to all market participants 768. Investors should also consider the
performance history of the parties involved with a securitisation e.g.
originator, sponsor, etc.769. The originator/sponsor, as well as the original
lender (if different) should have sufficient experience in originating the type
of assets being securitised770.

7.13.9 Payment status. None of the assets in the pool may be delinquent or in
default at the inception of the securitisation771 (subsequent defaults do not
render the securitisation non-STC compliant as defaults are inevitable). The
following detailed requirements apply:

the underlying obligors have not been subject to insolvency proceedings

or debt restructuring within three years of exposures being included in
the securitisation;

none of the underlying obligors has an adverse credit history in any public
credit registry;

the current credit rating or credit score of the obligors in the pool is not
associated with a significant risk of default; and

there is no dispute between the obligors and the original lender in

respect of the loan772.

7.13.10 In addition, at the time of inclusion in the pool all obligors must have made
one payment unless the terms of the loan requires discharge in a single
instalment at maturity of the facility 773.

7.13.11 Many of these criteria seem vague. In particular, it is not obvious what an
“adverse credit history” or a “significant risk of default” really means. In the
absence of guidance from their supervisor, it seems likely that different banks
or groups will take different approaches.

7.13.12 The assessment of compliance is made 45 days or less from the closing date 774.

7.13.13 Underwriting standards. Originators must demonstrate to investors that any

exposures in the pool was originated in the ordinary course of business in
accordance with non-deteriorating underwriting standards. Changes to

 Back to contents 121

 SECURITISATION

underwriting standards must be disclosed, and the standards applied to

securitised exposures must be as robust as in respect of assets retained on the
originator’s balance sheet. The originator must ensure before granting a loan
that the borrowers have the ability and will to make timely payments under
the loan775. All credit claims and receivables must be originated in accordance
with sound and prudent underwriting criteria 776. Where claims are acquired
from a third party vendor for securitisation, the purchaser must review the
underwriting standards applied by the original lender before securitising
them, including an assessment of the willingness and ability of the borrowers
to repay777. These requirements seem inspired by certain poor origination
practices prior to the financial crisis, where some lenders failed to follow
prudent practices when making loans as the loans would be securitised and
therefore removed from their balance sheet.

7.13.14 Asset selection. The performance of the securitisation should not rely on the
selection of assets through active management on a discretionary basis.
Obligations forming part of the pool of assets should be transferred on the
basis of clearly defined eligibility criteria. Any replenishments of the pool
must be made based on of objective criteria and not “cherry picked”778.

7.13.15 Credit quality. At the portfolio cut-off date all exposures must have the
following risk weights or lower (after applying any available credit risk
mitigation) under the standardised approach to credit risk and credit risk
mitigation:

40% for exposures in the regulatory residential real estate sub-portfolio;

50% for exposures in the regulatory commercial real estate sub-portfolio;

50% for exposures in the other real estate sub-portfolio;

50% for exposures in the ADC sub-portfolio;

75% for exposures in the regulatory retail sub-portfolio; and

100% for other exposures779.

7.13.16 Granularity. At the portfolio cut-off date no single exposure may exceed 1%
of the exposure value of all exposures in the pool780. In jurisdictions with
structurally concentrated corporate loan markets this figure may be increased
to 2%, with prior supervisory consent, provided that the originator or sponsor
retains subordinated tranches which cover at least 10% of losses. These
retained tranches are ineligible for STC treatment 781.

7.13.17 Transfer. The transfer of exposures must be by way of a true sale. The
following requirements must be met before a true sale will be recognised:

the claims/receivables must be enforceable against the underlying

obligors782, and the securitisation documentation must include
representations and warranties to this effect;

 Back to contents 122

 SECURITISATION

the claims/receivables are beyond the reach of the seller, its creditors or
any liquidator, and are not subject to material re-characterisation risk or
clawback783;

the securitisation is not effected by derivatives or guarantees (i.e. it is a

traditional securitisation);

there is effective recourse to the underlying obligors;

the transaction is not a re-securitisation; and

a third party provides an independent legal opinion on the foregoing 784.

7.13.18 It is not possible for external lawyers to ascertain the enforceability of each
obligation in the pool. Nor can an external legal opinion address questions of
fact, such as whether there are re-securitisation exposures in the securitised
pool, as opposed to advising on compliance of the legal documentation with
the aforementioned rules. We assume that external lawyers will not be
expected to go beyond normal practice in issuing legal opinions.

7.13.19 In some jurisdictions, it may be possible to transfer credit claims other than
through a true sale, where there are material obstacles to achieving a true
sale785. Examples given in the Basel III text are equitable assignments and a
perfected contingent transfer786.

7.13.20 Initial and on-going data. Investors must be provided before pricing of a
securitisation with sufficient loan-level data or (in the case of granular pools)
summary stratification data on the risk characteristics of the pool. The
originator must publish quarterly reports on loan-level data or (in the case of
granular pools) stratification data to facilitate the trading of securities in the
secondary market. Prior to the inception of the securitisation the conformity
of the initial pool of exposures with the contractual eligibility requirements
must be reviewed by an independent third party e.g. an accountant or the
calculation agent787. This may take the form of a review of a representative
sample. The report need not be disclosed, but its results must be disclosed
in the initial offering documentation788.

7.13.21 No reliance on sale or refinancing of exposures. The securitisation structure

should not rely on the sale or refinancing of underlying credit claims or
receivables in order to repay the securities issued. The sale or refinancing of
underlying exposures may be permitted if the pool is sufficiently granular and
has a sufficiently differentiated repayment profile 789.

7.13.22 Interest rate and currency mismatches. Interest rate and currency
mismatches must be appropriately mitigated. This does not require a
matching hedge. Any swaps should be documented in industry-standard
master agreements (e.g. ISDA master agreements). Derivatives are only
allowed under the STC framework if entered into for genuine hedging
purposes790.

7.13.23 The waterfall. The priority of payments for all liabilities must be clearly
defined at the inception of the securitisation. This must be verified by a legal
opinion confirming the enforceability of the contractual waterfall791.

 Back to contents 123

 SECURITISATION

Presumably, this is directed at preferred creditors under applicable insolvency

laws as well as general contractual enforceability. Basel III prohibits junior
liabilities having payment preference over senior liabilities, or so-called
“reverse” cash flow waterfalls. All triggers affecting the cash flow waterfall,
payment profile or priority of payments must be fully disclosed in offering
documents and investor reports. Investor reports must also allow investors to
monitor over time the evolution of indicators subject to triggers 792. A cash
flow model, or information on the waterfall allowing its modelling, must be
made available to investors both before pricing and on an on-going basis793.
Any debt forgiveness, forbearance, payment holidays, restructuring and asset
performance remedies must be clearly identified on an on-going basis in the
periodic reports794.

7.13.24 Securitisations with a replenishment period must include either early

amortisation provisions and/or triggers terminating the replenishment period
including in the event of:

a deterioration in the credit quality of the underlying exposures;

a failure to acquire new underlying exposures of similar (or better) credit

quality; and

the insolvency of the originator or servicer 795.

7.13.25 Following the occurrence of an event of default, an acceleration event or a

trigger, the securitisation positions must be repaid in order of seniority, and
there must not be provisions in the documentation requiring the immediate
realisation of all the underlying assets at market value796.

7.13.26 Voting and enforcement rights. If the originator or sponsor becomes subject
to insolvency proceedings, then all voting and enforcement rights must be
transferred to the securitisation (presumably, the SPV). These rights must be
clearly defined797.

7.13.27 Documentation and legal review. Sufficient initial offering (e.g. draft
offering circular, draft offering memorandum or red herring) and draft
underlying documentation (e.g. asset sale agreement, servicing and back-up
servicing agreement, administration and cash management agreement, trust
deed, security deed, agency agreement, bank account agreements, inter-
creditor agreement, master trust, swap documentation, liquidity facilities
and legal opinions) must be made available to investors, and potential
investors, on a continuous basis either prior to pricing or when legally
permissible to provide full disclosure of legal information, commercial
information and risk factors. Final offering documents must be available from
closing, and all documentation shortly thereafter. An independent legal
practice (law firm) must review the documentation, e.g. lawyers instructed
by the arranger or trustee798.

7.13.28 Alignment of interest. Originators and sponsors must retain a material net
economic exposure (which is undefined) to the pool of securitised assets, and
demonstrate a financial incentive in the performance of the assets after
securitisation799. This is to try to ensure that the interests of the
originator/sponsor are aligned with those of investors, in order to prevent

 Back to contents 124

 SECURITISATION

securitisation of poor quality assets, such as occurred in the run-up to the

financial crisis.

7.13.29 Servicers. Servicers must have expertise in the servicing of the underlying
claims or receivables, supported by a management team with extensive
industry experience. A servicer must act in accordance with reasonable and
prudent standards. Policies, procedures and risk management controls must
be well documented, and follow good market practice. There must be strong
systems and reporting capabilities800.

7.13.30 Trustees/fiduciaries. Such parties must act on a timely basis in the best
interests of the noteholders, and the initial offering documentation and all
underlying documentation (see above) must contain provisions facilitating the
resolution of conflicts between different classes of noteholders by the
trustee801. It should be noted that under English law a trustee’s duty of care
is not regarded as a fiduciary duty. Fiduciary duties are confined to those
duties specific to fiduciaries and trustees802 e.g. the “no conflict” rule (a
fiduciary must not allow a conflict to exist between his own interest and that
of his client, or between two clients) 803 and the “no profit” rule (a fiduciary
must not make a profit at the expense of his client) 804. Basel III adds that the
trustee/fiduciary must demonstrate sufficient skills and resources to comply
with its duty of care in performing its role in the securitisation. Remuneration
must be sufficient to incentivise these parties to meet their responsibilities in
full and in a timely way805. This sits slightly incongruously with English law as
a trustee’s duty of care is not fiduciary (see above) and a fiduciary is only
normally entitled to remuneration if expressly agreed by the beneficiaries.
The rationale is, however, clear: to ensure the trustee acts carefully in
discharging its responsibilities under the securitisation and has an economic
incentive to do so.

7.13.31 Transparency. The contractual obligations, duties and responsibilities of all

key parties must be clearly defined in the initial offering and all underlying
documentation. There must be provisions addressing the replacement of
servicers, bank account providers, swap counterparties and liquidity providers
in the event of non-performance, insolvency or a deterioration in their
creditworthiness (e.g. a ratings downgrade below a specified threshold) 806.
Reports to investors must disclose income and disbursements (including
deferment, forbearance and repurchases 807), such as principal, interest
payments, restructured obligations and credit events in the underlying
portfolio808.

Criteria for ABCP short-term securitisations

7.13.32 The following definitions apply to ABCP STC securitisations:

An “ABCP conduit” is an SPV which issues commercial paper.

An “ABCP programme” is a programme of commercial paper issued by an

ABCP conduit.

“Assets” or an “asset pool” are the credit claims or receivables

underlying a transaction in which a conduit holds a beneficial interest.

 Back to contents 125

 SECURITISATION

“Investors” are holders of the commercial paper issued under an ABCP

programme, or any type of exposure to the conduit representing a
liability of the conduit e.g. a loan.

An “obligor” is the borrower underlying a credit claim or receivable that

forms part of an asset pool.

A “seller” is the party that: (i) originated the credit claims or receivables,
or (ii) purchased the claims from the original lender, and (iii) (in either
case) transfers those credit claims or receivables to the ABCP conduit.

A “sponsor” is the sponsor of an ABCP conduit.

A “transaction” is the transaction in which the ABCP conduit holds a

beneficial interest809.

7.13.33 Compliance with the STC criteria depends on the level of the exposure. If the
exposure is at the level of the transaction, then the transaction will qualify
for STC treatment if the criteria are complied with at the level of the
transaction810. If, however, the exposure is at the conduit level then the STC
criteria must be complied with both at the transaction level and at the conduit
level811.

7.13.34 Nature of assets. The assets underlying the transaction must be homogenous
in terms of asset type with contractually identified payment streams.
Securitisation exposures are expressly excluded 812 (although programme-wide
credit enhancements do not make a conduit ineligible) 813. The sponsor must
make representations and give warranties to the investors in the securitisation
documentation.814 Provided that each individual underlying exposure is
homogenous in terms of asset type, a conduit can be used to finance
transactions of different asset types815.

7.13.35 The following criteria apply when determining if a pool is homogenous:

there must be no need to analyse and assess materially different legal or

credit risk factors when carrying out due diligence;

the pool must have common risk drivers, including similar risk factors and
profiles;

the obligors must have standard features which result in a periodic

stream of payments to investors; and

repayment of securities issued must not rely on refinancing or resale of

the assets in the pool, although partial reliance is permitted providing
refinancing or resale is sufficiently distributed within the pool and the
residual values on which the transaction relied are sufficiently low 816.

7.13.36 Asset performance history. The sponsor must make available to investors
sufficient stratified loss performance data, such as delinquencies/defaults for
a time period long enough to permit meaningful evaluation817. A sponsor must
obtain from the original lender sufficient data to permit meaningful
assessment818.

 Back to contents 126

 SECURITISATION

7.13.37 The sponsor must have sufficient experience in risk analysis with exposures
similar to those being securitised. It must also have well documented
procedures and policies regarding the underwriting of transactions and the
on-going monitoring of the performance of securitised exposures. The original
lender must also have sufficient experience in underwriting the underlying
loans. Responsibility for this is placed on the sponsor, and the management
of the originator must also have such experience. Investors will request
confirmation from the sponsor that the performance history of the originator
has at least three years’ experience for retail claims, and five years for other
exposures819.

7.13.38 Asset quality. The sponsor is required to make representations and

warranties to investors that each individual transaction is not, at the time of
acquisition, in default or delinquent, or subject to a material risk of increase
in expected losses. This is required based on representations made by the
seller(s)820. Further, the original seller or sponsor must verify that in respect
of each obligation in the pool: (a) the obligor has not been subject to
insolvency or debt restructuring in the last three years, (b) the obligor is not
recorded on a public credit registry of persons with an adverse credit history,
(c) the obligor does not have an external credit rating or credit score
indicating a significant risk of default, and (d) the credit claim is not subject
to a dispute821. This must be assessed no earlier than 45 days before the
acquisition by the conduit822. The qualitative nature of such criteria has
already been remarked on.

7.13.39 Underwriting standards. The conduit must represent to investors that:

(save as set out below) it has taken steps to verify that the transactions
in the conduit, and any underlying credit claims and receivables have
been originated subject to consistent underwriting standards;

explain how this is the case; and

disclose any material changes to underwriting standards, as well as the

timing and purpose of such changes823.

7.13.40 Sponsors must also inform investors of the material selection criteria used in
selecting sellers824. Sponsors are also required to ensure that the seller(s) (as
original lender):

originated any credit claims or receivables in the ordinary course of

business subject to non-deteriorating underwriting standards; and

assessed the obligors as being able and willing to repay the obligation 825.

7.13.41 Presumably, this is to be effected through representations and warranties,

together with adequate due diligence, by the sponsor or by an independent
third party. Basel III states that the sponsor must “review the underwriting
standards (i.e. to check their existence and assess their quality) … and to
ascertain that [originators] have assessed the obligors’ ‘ability and volition to
make timely payments’ on their obligations’”826. Although it is not stated, we
assume this will be possible to satisfy based on an analysis of a representative

 Back to contents 127

 SECURITISATION

sample, as it would be impractical to carry out due diligence on a transaction-

by-transaction basis.

7.13.42 At conduit level, the sponsor must:

give representations and warranties about the checks it has carried out
on the enforceability of the underlying assets; and

disclose to investors the representations and warranties received by it

from the seller(s) that the underlying obligations in the pool(s) are not
subject to any condition that would adversely affect their
enforceability827. This clearly places an obligation on the sponsor to
obtain such representations and warranties from the seller.

7.13.43 Asset selection and transfer. Sponsors must also ensure that any credit
claims or receivables satisfy clearly defined eligibility criteria and are not
actively selected after the closing date, actively managed, or otherwise
subject to discretionary cherry-picking828. The last requirement is not
breached by replenishment of the asset pool provided it is effected on a non-
discretionary basis829.

7.13.44 The transactions with the conduit must be a true sale. Basel III specifically
requires that the underlying assets must:

be enforceable against the obligor;

be placed beyond the reach of the seller, its creditors or insolvency

official, and must not be subject to material re-characterisation or claw-
back risks;

not be effected through credit derivatives or guarantees (i.e. a synthetic

securitisation), but by a transfer of the credit claims; and

have recourse to the obligor830.

7.13.45 This is a somewhat rag-bag list. Only the second point is relevant to a true
sale (which requires a legally enforceable transfer of the obligations and is a
sine qua non of a traditional securitisation). Enforceability of underlying
obligations is clearly important for the obligations being valuable as
collateral, but can only ever be assessed on a portfolio-basis as an assessment
of compliance of lending with applicable laws on a transaction basis (including
consumer protection laws) is likely to be impractical in many cases as it is fact
specific. The exclusion of securitisation and re-securitisation positions, or
credit derivatives, follows from the definition of an STC securitisation. As
with other STC securitisations, conduits may use “other means” where a true
sale faces material obstacles831.

7.13.46 Basel III seems to require an in-house or external legal opinion, that: (i) the
underlying credit claims or receivables are enforceable against the obligor,
and (ii) are beyond the reach of the seller 832. Whilst the latter should not
present any real difficulty (subject to customary assumptions, reservations
and exclusions), it seems hard to see how lawyers could opine on the
enforceability of the underlying obligations given the fact-sensitive nature of

 Back to contents 128

 SECURITISATION

the inquiry (legal opinions never address matters of fact as they cannot be
known by the lawyer giving the opinion). Admittedly, it would be possible
(but very difficult) to review all of the underlying documentation, but it is
hard to see what practical benefit would be obtained, as defaults on retail
and some corporate exposures are only loosely related to defects in legal
enforceability i.e. many/most obligors that can pay will pay even if the
underlying credit exposure is unenforceable under applicable laws due to
technical deficiencies in compliance.

7.13.47 It is not clear, however, whether Basel III requires this, as the requirement in
CRE 40.122 which cross-refers to CRE 40.118(1) refers to it being met at the
level of the “true sale and the transfer of assets”, whereas CRE 40.118(1)
refers to the underlying credit claims or receivables being enforceable against
the obligor. If Basel III merely requires enforceability of the transaction with
the conduit, as opposed to the underlying obligors, then this requirement
should present no real difficulty. Hopefully, national supervisors, in
implementing the Basel standards, will make clear that only transaction
documentation enforceability is required. Where there is an external rating,
the ratings agency is likely to carry out some due diligence on the underlying
obligors as part of its internal processes in assigning a rating.

7.13.48 Data and disclosure requirements. Sponsors are required to disclose certain
data to investors, both at the outset and on an on-going basis833. Standardised
investor reports with specified information must also be provided 834.

7.13.49 Liquidity and credit facilities. Sponsors are required to provide liquidity
facilities and any credit enhancements for an ABCP programme issued by a
conduit835. Such support must be provided either at the level of the
programme as a whole, or at the level of transactions within an overall
programme836. Such facilities must enable investors to rely on the sponsor to
receive timely and full repayment of commercial paper issued by the
conduit837. Investors should also be fully protected against credit, liquidity,
and dilution risks838, as well as other risks 839. Further “[t]he full support
provided should be able to irrevocably and unconditionally pay the ABCP
liabilities in full and on time”840. Disclosure requirements apply841.

7.13.50 However, these requirements cannot be taken to mean that investors in

commercial paper cannot suffer an unexpected credit-related loss if the pool
of assets falls in value, or becomes worthless, as if this were the case there
would be no transfer of credit risk to investors.

7.13.51 Where there is more than one sponsor, then the majority of support (in terms
of coverage) must be provided by a single sponsor, the main sponsor 842. There
are specific requirements where the sponsor suffers a ratings downgrade
(increasing the credit risk to investors in the commercial paper issued by the
conduit), or where a funding commitment is not renewed. In this case, the
sponsor is required to collateralise its commitments in cash 843.

7.13.52 Redemption cash flow. Unless the underlying pool is sufficiently granular
with sufficiently distributed repayment profiles at transaction level,
repayment must depend on the general ability and willingness of the obligor
to repay, as opposed to refinancing or selling the collateral, or drawing on
liquidity facilities provided by the sponsor(s) 844. It follows that a sponsor

 Back to contents 129

 SECURITISATION

cannot satisfy this requirement by means of liquidity facilities provided by

itself845.

7.13.53 Interest rate and currency risk. Interest rate and currency risks at
transaction-level or conduit-level must be appropriately mitigated. This can
be done through appropriate hedging agreements such as swaps documented
under standard ISDA documentation846. Perfect hedges, however, are not
required. Non-derivative hedges are also acceptable if fully-funded (i.e.
collateralised)847.

7.13.54 The waterfall and commercial paper. The sponsor must ensure that
priorities of payment (i.e. the waterfall) are clearly defined at the time of
acquisition of interests in transactions by the conduit. “Appropriate legal
comfort” concerning the waterfall’s enforceability must be provided 848. In
most securitisations this will be through a legal opinion confirming that the
priority of payments is valid and enforceable. The sponsor must also ensure
that all triggers affecting the cash flow waterfall are fully disclosed in the
transaction documentation and reports, with information in the reports that
clearly identify breach status, the ability for a breach to be remedied and the
consequences of the breach849. Any triggers breached between payment dates
must also be disclosed on a timely basis in accordance with the transaction
documents850. Where the asset pool in which the conduit holds an interest is
itself tranched (i.e. a securitisation position) then the position in the structure
held by the conduit must be the most senior, and subordinated positions
cannot enjoy payment preference851. This suggests that ABCP securitisations
may be eligible for STC treatment where the underlying asset pool is itself a
securitisation, provided that the conduit holds the most senior tranche.

7.13.55 In the case of securitisations with a replenishment period, there must be

either an early amortisation provision, or a trigger terminating the
replenishment period, in the case of a deterioration in the credit quality of
the underlying exposures, an inability to replenish with exposures of similar
credit quality, or the occurrence of an insolvency event in respect of the
seller852.

7.13.56 Policies with regard to delinquency, default, dilution or restructuring of the

underlying obligors in the pool must be clearly identified so that debt
forgiveness, forbearance, payment holidays, restructuring and the like can be
identified on an on-going basis853.

7.13.57 Sponsors must obtain, before the conduit acquires a beneficial interest in a
transaction, liability cash flow analysis or information on the cash flow
provisions to enable proper analysis of the cash flow waterfall854.

7.13.58 A sponsor is required to make representations and warranties to investors that

all of the foregoing requirements are met, and, in particular, that it can
analyse the cash flow waterfall for each transaction. The sponsor must also
make available to investors a summary of the waterfall and credit
enhancements at both programme and transaction level 855.

7.13.59 Commercial paper issued under the programme must not include extension
options, or other features, which could extend the final maturity of the
paper856. Curiously, the text adds “where the right of trigger does not belong

 Back to contents 130

 SECURITISATION

exclusively to investors”857. As an extension to the term of the commercial

paper increases the risk to investors, it may seem logical to permit investor-
extensions. However, in a declining interest rate environment, investors may
extend the paper if a roll-over would lock-in a higher coupon. In any case, to
constitute “commercial paper”, as opposed to a term loan, the tenor must be
short (in the United States, up to 270 days, but on average 30 days 858, and in
England, a tenor of up to one year). Also, regulatory rules generally ensure
that commercial paper is short term to prevent the issuance of such paper
being regulated as a banking business859.

7.13.60 Insolvency of parties. Investors must be provided by the sponsor with

information on their enforcement rights against the underlying credit claims
should the sponsor become insolvent 860. In the event that the seller becomes
insolvent all voting and enforcement rights must be transferred to the
conduit. In any event voting and enforcement rights must be clearly
defined861.

7.13.61 Documentation and legal review. The initial offering documentation for the
ABCP programme must be provided to investors (and be available to potential
investors) within a reasonable period of time before issuance. Investors must
be provided with comprehensive risk factors in a readily understandable
format862. The sponsor must also ensure that the terms and documentation of
conduits is reviewed and verified by external lawyers prior to publication, and
also for material changes. Lawyers can certainly review documents but it is
not clear what is meant by verification, as all external lawyers can confirm is
if the documentation is legally enforceable and, in some cases only, whether
it conforms to existing market practice863.

7.13.62 The sponsor must receive sufficient initial offering documentation for each
transaction and be provided within a reasonable period of time prior to
inclusion in the conduit with full disclosure of the information and risk factors
required to provide liquidity facilities and/or credit enhancements.

7.13.63 Skin in the game. Either the original seller, or the sponsor, must retain a
material net economic exposure to the securitised exposures, and must have
a financial interest in the performance of the exposures 864. This is another
lesson of the financial crisis, where under the originate-to-distribute model
originators could have no economic interest in the performance of loans that
they originated once they had successfully been securitised. This contributed
to a decline in underwriting standards, and the origination of loans that had
a high risk of default (including, but not limited to, sub-prime and alt-A
mortgages). Unlike under EU law there is no prescribed threshold.

7.13.64 Maturity transformation. Maturity transformation (the making of longer-

term loans to support short-term borrowing) by ABCP conduits must be
limited. This is to reduce the risk of the conduit being unable to repay the
commercial paper at maturity, or being unable to refinance it through a new
issue of such paper. Sponsors must ensure that the weighted average maturity
of all the transactions is three years or less. This figure is calculated as the
higher of:

the exposure-weighted average residual maturity of the conduit’s

beneficial interests held, or the assets purchased by the conduit; and

 Back to contents 131

 SECURITISATION

the exposure-weighted average maturity of the underlying assets

financed by the conduit using a formula865.

7.13.65 Status of sponsor. The sponsor must be a bank subject to appropriate

prudential standards and levels of supervision 866. It need not be a bank subject
to the Basel III framework, or an equivalent framework, if it is not
internationally active, and its national supervisor applies a different
framework. This may (but need not) include liquidity, leverage and other
requirements. Risk-based capital requirements applicable under national law
must, however, be adhered to867.

7.13.66 Representations and warranties. The sponsor of the conduit must provide
investors with representations and warranties at conduit level that the
following criteria are satisfied at the transaction level that the seller and
other parties responsible for origination and servicing:

have well-documented procedures and policies to ensure appropriate

servicing;

have expertise in the origination of the same or similar assets;

have extensive servicing and work-out expertise, and a proven track

record in loss mitigation;

have expertise in the servicing of the underlying claims; and

are supported by a managing team with extensive industry experience 868.

7.13.67 The sponsor must provide representations and warranties to investors that the
above criteria are met, and explain how they are met. Equally,
representations and warranties must be given that the seller’s policies,
procedures and risk management controls are well-documented, adhere to
good market practices and are compliant with applicable regulatory
requirements869. Sponsors must also have expertise in providing liquidity
facilities and credit enhancements in the context of ABCP conduits 870.

7.13.68 At transaction level the sponsor must obtain from the original seller(s) the
representations referred to above871.

7.13.69 Transparency and documentation. The duties of all key parties must be
clearly defined in all documentation of the conduit and the ABCP programme
(but not in respect of underlying transactions) 872. The sponsor must represent
and warrant that this is the case at transaction level 873. The documentation
must also provide for the replacement of key parties in the event of a breach
of contract, their insolvency or a decline in their credit-worthiness874.
Investors must also be provided with information about liquidity facilities and
credit enhancements875.

7.13.70 At transaction level, the sponsor is required to undertake due diligence. The
duties and responsibilities of all key parties must be clearly defined 876, and
provisions to replace such parties must be documented877. There are specified
requirements in respect of performance reports 878.

 Back to contents 132

 SECURITISATION

7.13.71 Credit quality of underlying exposures. In order to avoid the creation of

unduly risky securitisations under the STC framework, Basel III specifies
certain minimum requirements in respect of credit quality. All underlying
exposures included in the pool must meet the following minimum risk weights
under the standardised approach (or presumably lower), after taking into
account any available credit risk mitigation:

40% on a value-weighted average exposure basis for the portfolio in the

case of regulatory residential real estate exposures;

50% on an individual exposure basis where the exposures are regulatory

commercial real estate, other real estate or ADC exposures;

75% on an individual exposure basis in respect of regulatory retail

exposures; and

100% on an individual exposure basis in respect of any other exposures 879.

7.13.72 Granularity. At the date of acquisition of the assets in the pool, no exposure
to a single obligor may exceed 2% of the aggregate outstanding exposure value
of the pool880. This figure may increase to 3% for conduits backed by corporate
exposures in jurisdictions with structurally concentrated corporate loan
markets subject to two conditions. Firstly, banks must obtain prior
supervisory approval. Secondly, the seller or sponsor must retain subordinated
tranche(s) providing credit enhancements which cover at least the first 10%
of losses. Such first loss tranches are ineligible for STC treatment 881.

7.14 The Standardised Approach to Securitisation Exposures

7.14.1 Unlike the position under Basel II, this approach is at the bottom of the
hierarchy and can only be used if no other approach is available for
securitisation exposures. It is therefore a default treatment.

7.14.2 The new standardised approach is a revision of the simplified supervisory

formula approach that was consulted on in December 2012. Capital
requirements under the standardised approach are calculated using the
weighted-average standardised approach capital charge for the underlying
exposures in the pool (KSA) and a factor – referred to as “w” – which is the
ratio of the sum of the amount of all the underlying pool of exposures that
are delinquent to the total amount of underlying exposures. The “w” factor
represents an uplift to the capital charge for the underlying pool of exposures
to reflect a deterioration in the credit quality of the pool, and is intended by
the Basel Committee to enhance the risk sensitivity of the standardised
approach.

7.14.3 The calibration of the standardised approach is intended to generate capital

requirements that, overall, are slightly higher than that under the IRB
approach, and roughly comparable to the external ratings-based approach.
Re-securitisations attract a 50% capital surcharge over other securitisations
under the standardised approach. Other approaches are not available for re-
securitisations.

 Back to contents 133

 SECURITISATION

7.14.4 Under the standardised approach (SEC-SA) a bank is required to use the
following inputs:

the capital charge determined under the standardised approach for the
underlying exposures had they not been securitised (K SA);

the ratio of delinquent underlying exposures to total underlying

exposures in the securitisation pool (w);

the tranche attachment point (A); and

the tranche detachment point (D)882.

7.14.5 For this purpose, the “attachment point” is the minimum amount of pool-
level losses at which any given tranche begins first to suffer losses. The
“detachment point” is the amount of pool losses that completely wipes out
the tranche. For this reason, in a securitisation, the detachment point of a
subordinated tranche is the attachment point for the next immediately senior
tranche. Given the tranched nature of all securitisation positions, it is
necessary to use both A and D to obtain an accurate measurement of the level
of risk in any given securitisation tranche. Where the only difference between
exposures to a transaction is related to maturity, then A and D will be the
same883.

7.14.6 KSA is defined as the weighted-average capital charge of the entire portfolio
of underlying exposures, calculated under the standardised approach to credit
risk (CRE 20), in relation to the sum of the exposure amounts of underlying
exposures, multiplied by 8%884. Any eligible credit risk mitigation technique
applicable either to individual exposures, or to the pool as a whole, should be
taken into account, and thereby benefit the whole pool from a capital
perspective. KSA is a decimal between 0 and 1, so a weighted-average risk
weight of 100% translates into a figure for K SA of 0.08885. A risk-weight of 1250%
means that KSA = 1.

SPV structures

7.14.7 Where a traditional securitisation involves an SPV, all of the SPV’s exposures
related to the pool are treated as exposures in the pool. This includes, but is
not limited to, reserve accounts, cash collateral accounts, and claims under
interest rate or currency swaps. For swaps (other than credit derivatives) KSA
must include in the numerator the positive current market value multiplied
by the risk-weight of the swap provider multiplied by 8%. However, the swap
is ignored in calculating the denominator as it does not provide any credit
enhancement886.

7.14.8 A bank may exclude SPV exposures from the pool for regulatory capital
purposes with supervisory consent provided that the risk either does not affect
its securitisation exposure or the risk is immaterial 887. Examples of techniques
that may either significantly reduce or even eliminate the potential risk of
default of a swap provider include cash collateralisation together with an
obligation on the swap provider to post variation margin in the event of an
increase in the market value of the swap, or the automatic substitution of the

 Back to contents 134

 SECURITISATION

swap provider with another swap provider at no cost to the SPV if the initial
swap provider suffers a ratings downgrade 888.

7.14.9 For funded synthetic securitisations, any proceeds from the issue of credit-
linked notes, or other funded obligations of the SPV, that serve as collateral
for repayment must, in principle, be included in the calculation of K SA if the
default risk of the collateral is tranched. This is not, however, necessary if
the bank obtains supervisory consent to disregard the risk on the basis that it
is immaterial889. Unfunded synthetic securitisations will not have an SPV as
the credit risk is transferred by credit derivatives or equivalents to the
protection provider.

Specific provisions and non-refundable purchase price discounts

7.14.10 If a bank sets aside a specific provision, or has a non-refundable purchase

price discount, on an exposure or exposures in the pool K SA is calculated
disregarding such amounts and the bank must calculate KSA based on the gross
amount890. The reason is that such provisions or discounts are not fungible and
do not reduce overall losses on the pool.

Calculating “w”

7.14.11 “w” was introduced above. It is the ratio of the sum of the notional amount
of delinquent underlying exposures to the nominal amount of the underlying
exposures891. The definition of “delinquent exposures” is underlying exposures
that are: (i) 90 days or more past due, (ii) subject to bankruptcy or insolvency
proceedings, (iii) in the process of foreclosure 892 on a mortgage, (iv) held as
real estate owned, or (v) where the debt is in default (as defined in the
securitisation documents)893.

The capital calculation

7.14.12 The main elements of the capital calculation under SEC-SA have already been
introduced. We will now refer to KA which is the capital requirement relevant
to the attachment point. KA is calculated as follows:

𝐾𝐴 = (1 − 𝑤) 𝑥 𝐾𝑆𝐴 + 0.5𝑤 894

In circumstances where a bank cannot calculate “w” because it does not know
the delinquency status of the entire pool, but this lack of knowledge applies
to no more that 5% of all underlying exposures in the pool, then a modified
definition of KA applies:

𝐾𝐴 =
𝐸𝐴𝐷𝑠𝑢𝑏𝑝𝑜𝑜𝑙1 𝑤ℎ𝑒𝑟𝑒 𝑤 𝑘𝑛𝑜𝑤𝑛 𝐸𝐴𝐷𝑠𝑢𝑏𝑝𝑜𝑜𝑙 2 𝑤ℎ𝑒𝑟𝑒 𝑤 𝑢𝑛𝑘𝑛𝑜𝑤𝑛
( 𝑥𝐾𝐴 𝑠𝑢𝑏𝑝𝑜𝑜𝑙 1 𝑤ℎ𝑒𝑟𝑒 𝑤 𝑘𝑛𝑜𝑤𝑛 ) + 895
𝐸𝐴𝐷 𝑇𝑜𝑡𝑎𝑙 𝐸𝐴𝐷 𝑇𝑜𝑡𝑎𝑙

7.14.13 If a bank cannot calculate “w” for 95% or more of the underlying exposures
then the capital charge is 1250%896 which, as we have seen, generates a KSA
decimal of 1 equivalent to a deduction from capital. Banks are therefore
incentivised to obtain data to calculate “w”.

 Back to contents 135

 SECURITISATION

7.14.14 The actual capital calculation under the standardised approach is given below.
In addition to the components we have already seen, the following factors
must be determined:

a = − (1/(𝑝 𝑥 𝐾𝐴 ))

u = 𝐷 − 𝐾𝐴

l = 𝑚𝑎𝑥 (𝐴 − 𝐾𝐴 ; 0)

“p” is a supervisory parameter which represents the relative capital surcharge

for all securitisation exposures compared to the capital requirement for the
underlying pool determined as follows:

(𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑚𝑒𝑛𝑡 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑠𝑒𝑐𝑢𝑟𝑖𝑡𝑖𝑠𝑎𝑡𝑖𝑜𝑛 𝑒𝑥𝑝𝑜𝑠𝑢𝑟𝑒𝑠 𝑢𝑛𝑑𝑒𝑟 𝑡ℎ𝑒 𝑔𝑖𝑣𝑒𝑛 𝑎𝑝𝑝𝑟𝑜𝑎𝑐ℎ)−(𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑟𝑒𝑞𝑢𝑖𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑢𝑛𝑑𝑒𝑟𝑙𝑦𝑖𝑛𝑔 𝑒𝑥𝑝𝑜𝑠𝑢𝑟𝑒𝑠 𝑖𝑓 ℎ𝑒𝑙𝑑 𝑑𝑖𝑟𝑒𝑐𝑡𝑙𝑦 𝑏𝑦 𝑎 𝑏𝑎𝑛𝑘)
𝑃=
(𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑢𝑛𝑑𝑒𝑟𝑙𝑦𝑖𝑛𝑔 𝑒𝑥𝑝𝑜𝑠𝑢𝑟𝑒𝑠 𝑖𝑓 ℎ𝑒𝑙𝑑 𝑑𝑖𝑟𝑒𝑐𝑡𝑙𝑦 𝑏𝑦 𝑎 𝑏𝑎𝑛𝑘)

p is equal to 1 for any securitisation exposure that is not a re-securitisation

exposure . For re-securitisation exposures p = 1.5 , reflecting a 50% uplift
897 898

in capital charges for such exposures.

7.14.15 Taking all of the relevant factors into account, the capital charge determined
under the standardised approach as follows:

𝑒 𝑎𝑢 −𝑒 𝑎𝑙
𝐾𝑆𝑆𝐹𝐴 (𝐾𝐴 ) = 899
𝑎(𝑢−𝑙)

7.14.16 The actual risk weights are as follows900:

if D is less than or equal to KA the exposure receives a 1250% risk weight.

This reflects the view that first loss positions must effectively be
deducted from capital.

If A is equal or greater that KA than the risk weight equals KSSFA(KA)

multiplied by 12.5. The most senior tranche is accorded a risk weight
equal to the formula above.

If A is less than KA and D is greater than KA then the capital charge is a

weighted average of both 1250% and 12.5 times K SSFA(KA) applying the
following formula:

𝐾𝐴 −𝐴 𝐷 −𝐾𝐴
𝑅𝑊 = (12.5 𝑥 ) + ( 12.5 𝑥 𝐾𝑆𝑆𝐹𝐴 (𝐾𝐴 )) 𝑥
𝐷−𝐴 𝐷−𝐴

7.14.17 The risk weight for market hedges, such as currency or interest rate swaps,
must be inferred from a securitisation exposure that is pari passu to the swap
(if it exists) and, if not, from the immediately junior tranche 901.

7.14.18 Any risk weight derived from the above formula is subject to a floor of 15% 902.

7.14.19 Where a bank is required to apply the SEC-SA approach to an unrated junior
exposure in a transaction where the more senior tranches are rated, and
therefore no rating can be inferred for the junior piece, the risk weight under
the SEC-SA is subject to a floor equal to the risk weight of the immediately
senior rated exposure903.

 Back to contents 136

 SECURITISATION

STC securitisations

7.14.20 The capital charge for STC securitisations is calculated in exactly the same
way as is set out above with the following two modifications:

p = 0.5 as opposed to 1 904 (representing a 50% reduction in capital

charges); and

the risk-weight floor is 10% for senior exposures and 15% for subordinated
exposures905.

Re-securitisation exposures

7.14.21 Exactly the same approach is applied to re-securitisation exposures. The

following modifications apply:

w = 0 (delinquencies) for any exposure to a securitisation tranche in the

underlying pool; and

p = 1.5 to reflect the 50% uplift due to the greater potential for loss on
re-securitisations906.

7.14.22 Where only some of the exposures in the pool are securitisation exposures,
and others are not, the pool is notionally sub-divided into two sub-pools with
KA calculated separately for each sub-pool using different “w” factors907. The
resulting risk weight is subject to a 100% floor908, and the caps for
securitisation exposures set out above do not apply 909.

7.15 External Ratings-based Approach

7.15.1 This section describes the external ratings-based approach for those
jurisdictions where use of external ratings is permitted. Ultimately, it derives
from the Basel II securitisation standard, although with many changes. It is
intended for those banks in jurisdictions where use of external ratings is
permitted as a more risk sensitive approach than the standardised approach
(supra).

7.15.2 Securitisation exposures for which an express or inferred rating can be

obtained are risk-weighted at the exposure amount multiplied by the
designated risk weight. Operational criteria apply 910.

7.15.3 For the purposes of this section a rating may be “inferred” if the exposure is
senior to a rated exposure for which there exists a credit rating provided by
a recognised ratings agency.

7.15.4 Basel III applies primacy to short-term ratings. Where a short-term rating is
available then it must be used (including an inferred rating based on a short-
term rating). The following risk weights apply 911.

 Back to contents 137

 SECURITISATION

External ratings-based approach risk weights for short-term ratings

External A-1/P-1 A-2/P-2 A-3/P-3 All other
rating ratings
Risk weight 15% 50% 100% 1250%

7.15.5 In cases where a short-term rating does not exist, and cannot be inferred,
then the bank may use long-term ratings to determine its capital charge. In
this case, the capital charge depends on the external rating (or inferred
rating, if possible), the seniority of the position, the tranche maturity and the
thickness (for non-senior tranches)912. The risk weights under the external
ratings-based approach are set out in the following table913:

External ratings-based approach risk weights for long-term ratings

Senior tranche Non-senior tranche
Rating Tranche maturity (MT) Tranche maturity (MT)
1 year 5 years 1 year 5 years
AAA 15% 20% 15% 70%
AA+ 15% 30% 15% 90%
AA 25% 40% 30% 120%
AA- 30% 45% 40% 140%
A+ 40% 50% 60% 160%
A 50% 65% 80% 180%
A- 60% 70% 120% 210%
BBB+ 75% 90% 170% 260%
BBB 90% 105% 220% 310%
BBB- 120% 140% 330% 420%
BB+ 140% 160% 470% 580%
BB 160% 180% 620% 760%
BB- 200% 225% 750% 860%
B+ 250% 280% 900% 950%
B 310% 340% 1050% 1050%
B- 380% 420% 1130% 1130%
CCC+/CCC/CCC- 460% 505% 1250% 1250%
Below CCC- 1250% 1250% 1250% 1250%

7.15.6 Readers will note the much higher capital charges for non-senior tranches due
to the higher losses on thin securitisation tranches in the financial crisis.

7.15.7 The risk weight assigned to a securitisation exposure is calculated as follows:

to account for tranche maturity, banks must use linear interpolation

between the risk weight for one and five years. This is a method of curve

 Back to contents 138

 SECURITISATION

fitting using linear polynomals to determine the maturity of a

securitisation position between 1 and 5 years; and

to account for tranche thickness, banks must calculate the risk-weight

for non-senior tranches as follows: thickness (T) is measured as D - A i.e.
the difference between the attachment and detachment points as
discussed above914.

7.15.8 The risk weight is calculated as follows:

Risk weight = (risk weight from the table set out above after adjusting for
maturity) x (1 – min (T, 50%))915

7.15.9 In the case of market risk hedges (e.g. currency or interest rate swaps) the
risk weight is inferred from a securitisation exposure pari passu or junior to
the hedge916. Under the external ratings-based approach there is a capital
floor of 15% and any inferred risk weight can never be lower than the risk
weight corresponding to a senior tranche of the same securitisation with the
same rating and maturity917.

Operational requirements

7.15.10 In order to use the external ratings-based approach, the following operational
criteria must also be satisfied. If they are not then the bank must use the
fall-back SEC-SA approach (supra). The criteria may be summarised as
follows:

the external rating must take into account and reflect the entire amount
of credit risk exposure (e.g. both principal and interest);

the ratings agency (ECAI) must be recognised by the bank’s national

supervisor, and the key elements underlying the assessments must
generally be publicly available on a non-selective basis free of charge 918,
and published in an accessible form. Ratings made available only to the
parties to the transaction are not eligible;

ratings agencies must have demonstrated expertise in assessing

securitisations, which may be evidenced by strong market acceptance;

the standardised approach rules for multiple ratings apply (see the
earlier chapter on the Standardised Approach to Credit Risk);

credit risk mitigation must not be double-counted. Therefore, if the

mitigant is already taken into account in the external rating then it may
not, obviously, be recognised again. If the provider of credit protection
is not rated, the exposure is treated as unrated;

where a credit risk mitigant solely protects a specific exposure within a

given structure (e.g. an asset-backed security tranche) and this is
reflected in the external assessment, the bank must treat the exposure
as if it is unrated and then apply the credit risk mitigation treatment
under either the standardised approach or the foundation IRB approach
to the hedge; and

 Back to contents 139

 SECURITISATION

banks may not recognise external ratings where the assessment is

partially based on unfunded support provided by the bank. By way of
example, if a bank buys ABCP and provides an unfunded securitisation
exposure to the programme (e.g. a liquidity facility), and that facility is
material in determining the credit rating of the ABCP, the bank must treat
the commercial paper as unrated919. Whilst this may be logical, it is
unclear how the bank will know unless it has detailed knowledge of the
rating methodology and the factors taken into account when determining
the rating.

Inferred ratings

7.15.11 We have earlier discussed inferred ratings, which play a significant role under
the external ratings-based approach. Where a rating can be inferred, it must
be used and the SEC-SA is treated as a fall-back only where a rating cannot be
inferred. To infer a rating, the securitisation position from which the rating
is inferred must rank either pari passu or be subordinated in all respects to
the unrated securitisation exposure920. Any credit enhancements must be
taken into account in assessing the relative subordination of the unrated
exposure and the rated exposure. For example, if the rated exposure benefits
from credit risk mitigation, but the unrated exposure does not, then it is not
permitted to infer a rating from the rated tranche that does so benefit 921.
Secondly, the maturity of the rated exposure must be equal to, or longer than,
the unrated exposure for which the rating is being inferred 922. Thirdly, on an
on-going basis, any inferred rating must be updated continuously to reflect
any subordination of the unrated position, or changes in the external rating
of the rated securitisation exposure923. Finally, the external rating must satisfy
the securitisation operational criteria for recognition of ratings924. All of this
seems common sense, although the requirements may present some
challenges to banks, especially in respect of on-going monitoring of
compliance of an external rating not obtained by the bank with the reference
criteria. It is not immediately obvious how a bank could do this. Equally,
whilst it is hard to quibble with the requirement that the reference
securitisation obligation rank pari passu or junior to the unrated exposure, or
have an equal or longer maturity, these requirements may present practical
difficulties when it comes to assessing compliance where a bank has limited
information.

STC securitisations

7.15.12 We have already described the lower risk weights for STC securitisations under
the standardised approach. We now proceed to consider the capital charges
under the external ratings-based approach. The following rules apply 925.

7.15.13 For exposures with short-term ratings (or where a short-term rating can be
inferred - supra) the following risk weights apply 926 :

STC risk weights for short-term ratings

External A-1/P-1 A-2/P-2 A-3/P-3 All other
rating ratings
Risk weight 10% 30% 60% 1250%

 Back to contents 140

 SECURITISATION

7.15.14 It follows that any rating below A-3/P-3 will result in a capital treatment
equivalent to a deduction from capital.

7.15.15 For long-term ratings the following table applies to STC securitisations927:

STC risk weights for long-term ratings

Senior tranche Non-senior (thin) tranche
Rating Tranche maturity (MT) Tranche maturity (MT)
1 year 5 years 1 year 5 years
AAA 10% 10% 15% 40%
AA+ 10% 15% 15% 55%
AA 15% 20% 15% 70%
AA- 15% 25% 25% 80%
A+ 20% 30% 35% 95%
A 30% 40% 60% 135%
A- 35% 40% 95% 170%
BBB+ 45% 55% 150% 225%
BBB 55% 65% 180% 255%
BBB- 70% 85% 270% 345%
BB+ 120% 135% 405% 500%
BB 135% 155% 535% 655%
BB- 170% 195% 645% 740%
B+ 225% 250% 810% 855%
B 280% 305% 945% 945%
B- 340% 380% 1015% 1015%
CCC+/CCC/CCC- 415% 455% 1250% 1250%
Below CCC- 1250% 1250% 1250% 1250%

7.15.16 The risk weights set out above are subject to the following floors. A 10% floor
applies to senior tranches, and a 15% floor for all other tranches928.

7.16 The Internal Ratings-based Approach

7.16.1 The internal ratings-based approach is at the summit of the hierarchy for
securitisation exposures where a bank is able to apply it. To do so the bank
must be able to calculate the IRB capital charge for the underlying pool of
exposures (KIRB)929.

7.16.2 The following criteria are relevant to the IRB approach to securitisation
exposures:

KIRB this is the capital charge under the relevant IRB approach (foundation or
advanced) applicable to the securitised pool of exposures and is essential to
calculating the IRB capital charge for securitised exposures.

 Back to contents 141

 SECURITISATION

A is the tranche attachment point.

D is the detachment point.

Where the only difference relates to maturity, then A and D are the same 930.

Definition of KIRB

7.16.3 KIRB is the ratio of the following, expressed in decimal form (so a capital charge
of 15% of the pool is 0.15)931:

the IRB capital requirement for the underlying pool (including expected
loss, and dilution risk, where applicable); to

the exposure amount of the pool (e.g. the sum of drawn amounts related
to securitised exposures plus the exposure-at-default associated with any
undrawn commitments932.

7.16.4 In calculating KIRB for mixed pools, KIRB must be calculated as if the exposures
in the pool were held directly by the bank, reflecting any applicable credit
risk mitigation933.

SPV structures

7.16.5 Where there is an SPV in the securitisation structure (e.g. traditional

securitisations) all of the SPV’s exposures must be treated as exposures in the
pool. This may include reserve accounts, cash collateral accounts, swap
accounts, etc. However, this is subject to the following exception. If the
bank can demonstrate that the risk of the SPV’s exposures is either
immaterial, or does not affect the bank’s securitisation exposure, then
national supervisors may permit banks to disregard such exposures 934.

7.16.6 For funded synthetic securitisations, any proceeds of the issue of credit-linked
notes or other funded obligations of an SPV that serve as collateral for the
repayment of the securitisation exposure in question, are included in the
calculation of the capital charge for the exposures had they not been
securitised (KIRB) unless the exposure is immaterial and the bank’s national
supervisor consents935.

Purchased receivables

7.16.7 The standard IRB treatment of purchased receivables applies to securitisation

exposures:

which are retail exposures where the bank is unable to rely primarily on
internal data; and

for all other exposures, if it would be an undue burden on the bank to

assess the default risk of obligors individually 936.

7.16.8 Where the top-down approach to purchased receivables is used under the IRB
approach to securitisation exposures then certain modifications are made.
Supervisors may exclude the application of the top-down approach

 Back to contents 142

 SECURITISATION

“depending on the bank’s compliance with minimum requirements”937. This is

unclear. If it means that the use of the top-down approach is dependent on
compliance with the relevant requirements then this is unobjectionable. If,
however, there is a general discretionary power to deny use of the top-down
approach in respect of securitised exposures where the relevant criteria are
satisfied then it is not obvious why this should be the case.

7.16.9 The following specific changes apply to the top-down approach when applying
it to securitisation exposures:

the requirement for a bank to have a claim on all proceeds of the pool,
or a pro-rata share, does not apply. Instead, the bank must have a claim
on all proceeds from the pool of securitised exposures that have been
allocated to the bank’s exposure in accordance with the terms of the
securitisation documentation;

the purchasing bank means the bank calculating KIRB;

references to the “bank” mean “the bank estimating probability of

default, loss-given-default (LGD) or expected loss for the securitised
exposures”; and

if the bank calculating KIRB cannot meet the specified IRB minimum
requirements, then it may apply the IRB approach if another party to the
securitisation does so who is acting “for and in the interests of” the
investors938.

7.16.10 These all seem logical changes to make the IRB approach for purchased
receivables operational in a securitisation context.

Specific provisions, etc.

7.16.11 If a bank sets aside a specific provision, or has a non-refundable purchase

price discount on an exposure in the pool, KIRB must be calculated using the
gross exposure939. As has been seen in the context of the standardised
approach, such an adjustment is also logical.

Dilution risk

7.16.12 Banks must determine a capital charge for dilution risk (e.g. the risk of
charge-backs, customer disputes, etc.) unless it is immaterial and the bank’s
supervisor gives consent to disregard it 940.

7.16.13 If default risk and dilution risk are treated in an aggregate manner (e.g. an
identical reserve or overcollateralisation covers both risks) then the bank is
required to separately calculate KIRB for default risk and KIRB for dilution risk
and then combine both into a single KIRB when applying the IRB frameworks to
securitisation exposures941. An example of such a calculation is given in the
Basel III consolidated text942.

7.16.14 Where there are separate waterfalls for credit risk and dilution risk a bank
must consult with its national supervisor to determine how the capital
calculation must be performed943.

 Back to contents 143

 SECURITISATION

Definitions of regulatory parameters for calculating the IRB capital charge

7.16.15 The IRB approach to securitisation exposures is based on certain parameters,

in addition to KIRB and other standard IRB inputs. These are as follows: the
attachment point (A), the detachment point (D) and the supervisory
parameter p. These were referred to in our discussion of the standardised
approach and the reader is referred to that section for explanations of A and
D. “p” is simply a regulatory multiplier to scale up the capital charge for
securitisation positions.

7.16.16 The attachment point (A) is defined, under the IRB approach, as “the
threshold at which losses within the underlying pool would first be allocated
to the securitisation exposure”944. A is a decimal between 0 and 1 and is
calculated as the greater of the following:

zero; and

the ratio of: (a) the outstanding balance of all underlying assets in the
securitisation minus the outstanding balance of all tranches that rank
senior or pari passu to the tranche that contains the securitisation
exposure held by the bank to (b) the outstanding balance of all underlying
assets in the securitisation945.

As seen above, the purpose is to identify the point at which the tranche
to which the bank’s exposure corresponds first begins to suffer losses.

7.16.17 The detachment point (D) “represents the threshold at which losses within
the underlying pool result in a total loss of principal for the tranche in which
a securitisation exposure resides”946. It is a figure between 0 and 1, and is the
greater of the following:

zero; and

the ratio of (a) the outstanding balance of all underlying assets in the
securitisation minus the outstanding balance of all tranches that rank
senior to the tranche that contains the securitisation exposure of the
bank to (b) the outstanding balance of all underlying assets in the
securitisation947.

7.16.18 When calculating A and D overcollateralisation and funded reserve accounts

must be recognised as tranches in their own right. However, unfunded reserve
accounts (e.g. unrealised excess spread) and assets that do not provide credit
enhancements (e.g. pure liquidity facilities or swaps) are not taken into
account in calculating A and D948.

7.16.19 The following parameters are used in calculating the IRB capital charges:

the “p” parameter floor is 0.3;

N is the effective number of loans in the underlying pool (see below);

KIRB is the capital charge for the underlying pool calculated under the IRB
approach;

 Back to contents 144

 SECURITISATION

LGD is the exposure-weighted average loss-given-default of the

underlying pool; and

MT is the maturity of a tranche949.

7.16.20 The supervisory parameters A, B, C, D and E are determined in accordance

with the following table950.

Table for supervisory parameters A, B, C, D and E

A B C D E
Wholesale Senior, 0 3.56 -1.85 0.55 0.07
granular
(N≥/25)
Senior, 0.11 2.61 -2.91 0.68 0.07
granular
(N<25)
Non- 0.16 2.87 -1.03 0.21 0.07
senior,
granular
(N≥/25)
Non- 0.22 2.35 -2.46 0.48 0.07
senior,
non
granular
(N<25)
Retail Senior 0 0 -7.48 0.71 0.24
Non- 0 0 -5.78 0.55 0.27
senior

7.16.21 Based on the supervisory parameters set out in the above table “p” is
calculated as follows:

𝐵
𝑝 = max[0.3, (𝐴 + + (𝐶 𝑥 𝐾𝐼𝑅𝐵 ) + (𝐷 𝑥 𝐿𝐺𝐷) + (𝐸 𝑥 𝑀𝑇 ))]951
𝑁

7.16.22 If an IRB pool contains both retail and non-retail exposures then the pool must
be separated into two notional pools, one retail sub-pool and one wholesale
sub-pool, and N, KIRB, and LGD are determined for each notional sub-pool.
Subsequently, a weighted average p-parameter for the transaction is
calculated on the basis of the two sub-pools and the nominal size of exposures
in each sub-pool952.

7.16.23 In the case of so-called “mixed pools” where a bank can calculate KIRB on a
percentage of exposures in the pool, the determination of “p” must be based
solely on the IRB underlying assets, with assets to which the standardised
approach is applied disregarded953.

Effective number of exposures

7.16.24 The effective number of exposures (N) is calculated as follows:

(Σ𝑖 𝐸𝐴𝐷𝑖 )2
𝑁=
Σ𝑖 𝐸𝐴𝐷𝑖 2

 Back to contents 145

 SECURITISATION

where EADi represents the exposure-at-default associated with the ith

exposure in the pool954.

7.16.25 The exposure-weighted average LGD is calculated as follows:

LGDi is the average LGD associated with all exposures to the i th obligor955.

7.16.26 The LGD is calculated as follows:

Σ𝑖 𝐿𝐺𝐷𝑖 𝑥 𝐸𝐴𝐷𝑖
𝐿𝐺𝐷 = 956
Σ𝑖 𝐸𝐴𝐷𝑖

7.16.27 If default and dilution risks for purchased receivables are treated in an
aggregate manner (e.g. a single reserve or over-collateralisation account
covers both) the LGD is a weighted average of the LGD for default risk and
the 100% LGD for dilution risk957.

7.16.28 A simplified method is available for calculating the effective number of

exposures and the exposure-weighted average LGD. It is a prerequisite for
applying this approach that the portfolio share associated with the largest
exposure (C1) is no more than 3% of the underlying pool. In the formula below
Cm is the share of the pool corresponding to the sum of the largest m exposures
determined by the bank using its internal calculations.

7.16.29 In this case banks may determine:

(𝐶𝑚 − 𝐶1 ) 𝑥 max (1−𝑚 𝑥 𝐶1 ,0) −1

𝑁 = ((𝐶1 𝑥 𝐶𝑚 ) + )
𝑚−1

1
7.16.30 If only C1 is known to the bank then the 𝐿𝐺𝐷 = 0.5 and 𝑁 = .
958
𝐶1

Risk weight calculations

7.16.31 Using the above factors, and the normal IRB credit risk factors, the capital
charge under the IRB approach is determined as follows:

KIRB

KSSFA(KIRB) is the capital charge per unit of securitisation exposure based on the
following parameters: e, a, u and l. For these purposes:

e = 2.71828;

a = - (1/p x KIRB))

u = D – KIRB;

l = max (0, A - KIRB)959; and

𝑒 𝑎𝑢 − 𝑒 𝑎𝑙
𝐾𝑆𝑆𝐹𝐴 (𝐾𝐼𝑅𝐵) = 960
𝑎 (𝑢−𝑙)

7.16.32 Taking all of this into account the risk weights for securitisation exposures
under the IRB approach are as follows:

 Back to contents 146

 SECURITISATION

if D is less than or equal to KIRB then the risk weight is 1250% (equivalent
to a deduction from capital);

if A is equal to or greater than KIRB the risk weight (expressed as a

percentage) equals KSSFA(KIRB) multiplied by 12.5; and

for all other positions (i.e. A < K IRB and D > KIRB) the risk weight is
determined in accordance with the following formula:

12.5 𝑥( KIRB − 𝐴) 12.5 𝑥 𝐾𝑆𝑆𝐹𝐴 (KIRB ) 𝑥 (𝐷−KIRB )

𝑅𝑊 = + 961
𝐷−𝐴 𝐷−𝐴

7.16.33 The risk weight for currency and interest rate swaps must be inferred from a
securitisation exposure pari passu or junior to it962.

7.16.34 There is a floor on all risk weights determined in accordance with the above
of 15%963 (7% under Basel II).

STC securitisations

7.16.35 As mentioned before, a preferential risk weight applies to STC securitisations,

although the calculation is the same (with revised parameters). The following
rules apply:

p is subject to a floor of 0.3;

N must be calculated as described above (although a different table for

the supervisory parameters A, B, C, D and E applies when calculating N).
The table is set out below;

KIRB is calculated in exactly the same way (i.e. it is the IRB capital charge
for the underlying pool, assuming it had not been securitised);

LGD is the exposure-weighted average loss-given-default of the

underlying pool (not individual exposures) calculated as set out above;
and

maturity (MT) is the maturity of the tranche964.

7.16.36 The table for calculating supervisory parameters for determining N is as

follows for STC securitisations is set out below.

Supervisory parameters A, B, C, D and E

A B C D E
Wholesale Senior, granular (N≥ 0 3.56 -1.85 0.55 0.07
25)
Senior, non granular 0.11 2.61 -2.91 0.68 0.07
(N < 25)
Non-senior granular 0.16 2.87 -1.03 0.21 0.07
(N≥ 25
Non-senior, non- 0.22 2.35 -2.46 0.48 0.07
granular (N < 25)

 Back to contents 147

 SECURITISATION

Supervisory parameters A, B, C, D and E

A B C D E
Retail Senior 0 0 -7.48 0.71 0.24
Non-senior 0 0 -5.78 0.55 0.27

7.16.37 There is a floor for STC exposures of 10% for senior tranches and 15% for
subordinated tranches965.

7.17 The Internal-assessment Approach to ABCP programmes (SEC-IAA)

7.17.1 The internal assessment approach is available, subject to prior supervisory

approval, for securitisation exposures to ABCP programmes (e.g. liquidity
facilities and credit enhancements). It is a prerequisite for using the SEC-IAA
approach to such exposures that the bank has an IRB approved model,
although this need not relate to the securitised exposures. For example, if a
bank has an approved IRB model for retail exposures, it may apply the SEC-
IAA to corporate exposures to a conduit. The operational requirements must,
of course, be satisfied. Under SEC-IAA internal assessments of exposures to
ABCP programmes must be mapped to the equivalent ratings of a selected
ratings agency (ECAI). Those ratings are then used to determine the capital
charge966. SEC-IAA is therefore only ever relevant to unrated exposures where
the internal ratings-based approach cannot be used.

7.17.2 Basel III sets out the following operational requirements for using the SEC-IAA
approach:

the ABCP programme paper must have an external rating. If it does not
SEC-IAA cannot be used;

any internal assessment of the credit quality of a securitisation tranche

must be based on ratings agency criteria for the asset type covered. Such
assets must be (at purchase) at least investment grade. Also, the internal
assessment must be used in the bank’s internal risk management
processes, including management information and economic capital
systems. All of the relevant requirements for the IRB framework must be
met, even though the bank need not have an IRB permission to apply the
IRB approach to the relevant asset class;

the relevant supervisor must be satisfied that the chosen ratings

agency(ies) meets the criteria for the issuance of ratings under the
standardised approach. Further, the bank’s national supervisor must be
satisfied with the ratings methodology used in the process. This reflects
the hybrid standardised/IRB approach used under SEC-IAA;

banks are required to satisfy their relevant supervisor that internal

estimates correspond to the chosen ratings agency standards (i.e. prior
supervisory consent is required). For example, the Basel III text states
that supervisors may “if warranted, disallow on a full or partial basis any
seller-provided recourse guarantees or excess spread, or any other first-
loss credit enhancements that provide limited protection to the bank”967;

 Back to contents 148

 SECURITISATION

the bank’s internal assessment process must identify grades of risk, and
must correspond to rating agency grades, so supervisors can assess the
correspondence of internal assessments to rating agency grades;

the bank’s internal assessment process, including, in particular, the stress

factors for credit enhancements, must be as conservative as that applied
by major ratings agencies rating the ABCP programme’s commercial
paper. Specific rules apply to the use of individual ratings agency ratings;

regular reviews of the internal ratings process must be carried out either
by internal or external auditors, an external ratings agency, or the bank’s
own credit review or risk management function;

the bank must track the performance of its internal assessments over
time, and make adjustments if necessary;

the ABCP programme must have credit and investment guidelines,

including underwriting standards;

a credit analysis of the seller’s risk profile is required to be performed,

and must consider past and expected future financial performance, the
current market position, expected future competitiveness, leverage,
cash flow and interest coverage amongst other factors;

the ACBP programme’s underwriting policy must have minimum asset

eligibility criteria including: (a) excluding the purchase of assets
significantly past due, (b) limiting excess concentration to individual
obligors or geographical areas, and (c) limiting the maturity of the assets
purchased;

the ABCP programme must have collection processes that consider the
operational capability and credit quality of the servicer. The programme
should mitigate, to the extent possible, seller/servicer risks;

the estimate of loss on an asset pool must take into account all sources
of potential risk, including credit and dilution risk; and

the ABCP programme incorporates structural features into the purchase

of assets968.

Risk weights under the SEC-IAA approach

7.17.3 The risk weights (and therefore capital charges) are assigned based on the
inferred rating applicable to the external ratings-based approach969. If a bank
cannot use the SEC-IAA approach then it must use the standardised approach
as if it could use the IRB approach this would be required under the
securitisation hierarchy.

Supervisory judgment

7.17.4 If the relevant national supervisor considers a bank’s internal assessments not
to be adequate, then it may preclude reliance on the SEC-IAA approach to
ABCP exposures, both existing and newly originated, until any deficiencies are

 Back to contents 149

 SECURITISATION

remediated970. This potentially penal capital consequence may act as a

significant deterrent against breaches of the SEC-IAA requirements, although
as there are many uncertainties in the practical application of the
requirements, this may be a source of risk to banks.

7.18 Securitisations of Non-performing Loans

7.18.1 Specific rules apply to securitisations of non-performing loans (NPLs). An NPL

securitisation is a securitisation where the underlying pool’s “w” (see above)
is equal to 90% or higher (i.e. the ratio of non-performing loans is at least 90%)
at the origination cut-off date, and any subsequent date at which the pool is
replenished, or assets are removed from the pool. NPL securitisations exclude
pools containing securitisation exposures (i.e. re-securitisations)971.

7.18.2 National supervisors may specify a stricter definition of NPL securitisations (as
elsewhere in the Basel III standard which represents only minimum
harmonisation). Examples include a higher level of “w” than 90% or a
requirement that non-delinquent exposures meet specified criteria 972.

Risk weights

7.18.3 National supervisors must scrutinise NPL securitisations to avoid regulatory

arbitrage, and, in particular, transactions entered into with the main purpose
of reducing capital charges below the 100% capital requirement for securitised
NPL exposures973.

7.18.4 The risk weight for NPL securitisations is floored at 100%974, although under
the IRB, or look-through approach (for purchased receivables), the actual risk
weight may be higher.

7.18.5 Where (under the relevant hierarchy of treatments) a bank must use either
the standardised approach or the IRB approach, then a bank may apply a 100%
risk weight to the senior tranche of an NPL securitisation under the following
conditions:

the securitisation is a traditional securitisation; and

the sum of non-refundable purchase price discounts (NRPPD) is equal or

higher than 50% of the outstanding balance of the pool of exposures 975.

7.18.6 NRPPD is equal to the difference between the outstanding balance of

exposures in the pool and the price at which these exposures are sold 976. The
maximum capital cap applies to originators and sponsors of NPL securitisations
provided the originator uses the IRB approach to any exposure to the NPL
securitisation977.

 Back to contents 150

 UNSETTLED TRANSACTIONS AND FAILED TRADES

8. UNSETTLED TRANSACTIONS AND FAILED TRADES

Both unsettled transactions and failed trades give rise to counterparty credit risk.
However, they are not treated under the general CCR framework. Instead a bespoke
treatment applies which will be described in this very short chapter.

8.1 Failed Trades

8.1.1 A failed trade is a transaction where both legs of the transaction (i.e. delivery
and payment) are required to take place at the same time (delivery versus
payment or DvP), but where one leg remains outstanding (i.e. there has been
a contractual default). The requirement also applies to payment-versus-
payment transactions (e.g. the sale of pounds sterling for euro). The Basel
framework does not treat cryptocurrencies as a payment.

8.1.2 A default by a counterparty to settle a trade is not treated as such under the
standardised or IRB frameworks for credit risk978. Such transitory defaults are
in practice very common, and do not necessarily indicate poor
creditworthiness, due to issues like system delays, an unavailability of
securities for purchase in the market or a failure of a counterparty to make
delivery to the bank. However, if a default persists for five business days or
more then the following treatment applies 979.

8.1.3 The rules apply to all transactions in securities, FX and commodities that give
rise to a risk of delayed settlement or delivery. The rules cover transactions
entered into through clearing houses and CCPs. However, transactions
otherwise subject to the CCR framework (i.e. OTC derivatives, exchange-
traded derivatives, long-settlement transactions and securities financing
transactions) are excluded980. The relationship between long-settlement
transactions under the CCR framework and free deliveries, referred to below,
is not clear. Logically, a long-settlement position would be a transaction
where both legs are required to settle beyond the normal settlement period
for the instrument in question, whereas a free delivery occurs in case of a
contractual mismatch in the two legs of a transaction.

8.1.4 If both legs of a DvP transaction do not appear on the balance sheet at the
same time (i.e. settlement day accounting) then a 100% CCF applies to the
unaccounted for mismatch where a trade falls 981. This is to avoid the
accounting treatment driving the regulatory treatment.

System-wide failures of settlement

8.1.5 Where there is a system-wide failure of a settlement, clearing system or CCP

then a national supervisor may waive capital requirements until the situation
is rectified982. This presumably only refers to the default and not all capital
requirements.

 Back to contents 151

 UNSETTLED TRANSACTIONS AND FAILED TRADES

Capital charge for failed trades

8.1.6 As has been seen neither the standardised nor IRB frameworks for credit risk
apply in their normal way. Nor does the CCR framework apply. Instead the
normal capital charges generated under the standardised or IRB framework
are scaled up in accordance with the following table 983.

Number of business days after the Risk multiplier

agreed settlement date
From 5 to 15 8%
From 16 to 30 50%
From 31 to 45 75%
From 46 100%

8.1.7 If a bank recognises a loss by making a specific provision in respect of such

transactions then the provision will be deducted from the profit and loss
account. In accordance with the general approach to deductions from capital,
transactions that are written off through making a specific provision should
not be subject, pro tanto, to a capital charge.

8.2 Free Deliveries

8.2.1 A free delivery occurs where a bank transfers title to securities without
receiving the payment leg at the same time i.e. there is a delay between
delivery of the securities and payment, so the bank takes credit risk on its
counterparty defaulting. It also occurs where a bank makes a payment today
for delivery of securities at a later date. In this case the bank is exposed to
the risk of its counterparty not delivering the securities at the contractual
performance date due to a default.

8.2.2 All transactions where the price or the securities are not received on the same
business day as the securities are delivered or the price is paid are treated as
free deliveries984. For cross-time zone payments this means that the payments
are on the same day based on the relevant local time where the second leg of
the transaction is performed985. Non-DVP transactions where settlement is
nevertheless required on the same business day are not free deliveries.

8.2.3 If the bank uses the standardised approach to credit risk then it uses the
standardised risk weights to determine its capital charge for the relevant
counterparty986. Banks using the IRB approach unsurprisingly use IRB risk
weights. However, if the bank has no other banking book exposures to the
particular counterparty on which to generate an IRB risk weight it may instead
use an external rating to generate a PD figure 987.

8.2.4 An alternative to the above treatment, where exposures arising out of free
deliveries are not material, is to apply a 100% cross-the-board risk weight to
avoid the burden of performing a full credit assessment 988.

 Back to contents 152

 UNSETTLED TRANSACTIONS AND FAILED TRADES

Default on the second delivery or payment leg

8.2.5 A default in making a payment or delivery is not treated as a default under

the normal standardised or IRB capital framework for credit risk989. However,
if five business days or more have passed since the contractual delivery or
payment date then Basel III mandates a 1250% risk weight, equivalent to a
deduction from capital. This is because of the very high risk of the bank
suffering a credit risk loss. This applies until delivery or payment are actually
made990.

 Back to contents 153

 EQUITY INVESTMENTS IN FUNDS

9. EQUITY INVESTMENTS IN FUNDS

This chapter describes the capital treatment for a bank’s equity investments in funds.
This is not a separate capital framework but a menu of approaches as to how the
standardised and IRB approaches to credit risk must be applied to such exposures, with
a penal fall-back approach where they cannot be applied. There is a prescribed
hierarchy to the approaches. Banks that can apply the look-through approach must do
so. If they cannot then the mandate-based approach applies. If a bank cannot apply
either the look-through or mandate-based approaches then the fall-back approach must
be used991. Exposures to funds that are deducted from capital are excluded from the
treatment described in this chapter992.

9.1 Exclusions

9.1.1 The following equity exposures in funds are disregarded:

equity holdings in entities qualifying for a 0% risk weight under the

standardised approach to credit risk. This included PSEs that qualify for
a 0% risk weight993; and

at supervisory discretion, exposures to entities established under

legislated programmes that provide significant subsidies, and are subject
to some kind of governmental oversight and restrictions on the
investments they can make. There is a cap in such exempt investments
of 10% of the bank’s total regulatory capital994.

9.2 The Look-through Approach (LTA)

9.2.1 This is the most risk sensitive framework which is why it must be used where
possible. Under the LTA banks break down all their fund exposures as notional
exposures to all of the equities in the fund and then apply the applicable
approach (standardised or IRB) to those exposures. This is described in more
detail below.

9.2.2 Basel III imposes the following two requirements to use the LTA:

the bank must have sufficient and frequent information to identify the
underlying exposures of the fund; and

the information is verified by an independent third party 995.

9.2.3 The bank is not required itself to confirm the underlying exposures as in most
cases it would be unable to do this.

9.2.4 The first requirement set out above means that the frequency of the financial
reporting of the fund must be the same as, or more frequent than, that of the
bank’s. Additionally, the granularity of the reporting must be sufficient for
the bank to be able to apply the LTA to the fund’s exposures996. Basically, this
means that the bank must have full disclosure of the fund’s actual investments
on an ongoing basis.

 Back to contents 154

 EQUITY INVESTMENTS IN FUNDS

9.2.5 The second requirement will be satisfied if the underlying exposures are
verified by an independent third party. Examples given include the
depositary, the custodian bank or (where relevant) the management
company997. An external audit is not required998.

9.2.6 Under the LTA the bank calculates risk weights for all underlying exposures of
the fund as if it directly held those exposures. This includes any exposures
the fund has to derivatives used by the fund 999 e.g. interest rate or currency
swaps. A scaling factor of 1.5 applies to credit value adjustments (CVAs)
calculated under the market risk framework, so the capita charge for such
exposures only is subject to a 50% uplift from that applicable if the bank
directly held those exposures on its balance sheet 1000. The CVA framework is
described in chapter 13.

9.2.7 If a bank is unable to calculate the risk weights itself under the LTA it may rely
on third party calculations. In this case the risk weights are subject to a 20%
uplift1001 as the bank is relying on a third party to set its capital requirements.
This means that if the normal risk weight is 20% then the risk weight is
increased to 24%1002.

9.2.8 A leverage adjustment applies (see below).

9.3 The Mandate-based Approach (MBA)

9.3.1 If a bank cannot use the LTA then it must, if it can, use the MBA. The
underlying idea is that if the bank doesn’t know what exposures are included
in the fund then it can use the fund’s investment mandate as a proxy on the
assumption that the fund invests to the greatest degree allowed under the
fund documentation in riskier assets. Banks are not expected to assume that
the fund will include unauthorised investments. Where the investment
mandate is set out in legislation or regulatory rules then these may likewise
be used.

9.3.2 If the fund publishes some, but incomplete, information, then this may be
taken into account as well1003.

9.3.3 The actual capital calculation under the MBA is the sum of the following three
elements determined as follows:

balance sheet exposures (i.e. investments by the fund) are risk weighted
assuming the fund invests to the maximum extent possible in those assets
attracting the highest capital requirement. Where there is more than
one possible risk weight the highest is used 1004;

derivatives receive a risk weight based on the notional amount of the

derivative1005; and

CCR exposure on derivative positions is calculated under the CCR

standardised approach (see chapter 10). If the PFE cannot be known then
it is set at 15% of the value of the notional. A 1.5 scaling factor applies
to CVA risk as is also the case under the LTA1006.

9.3.4 A leverage adjustment also applies.

 Back to contents 155

 EQUITY INVESTMENTS IN FUNDS

9.4 Capital Calculation under LTA or MBA

9.4.1 Under both the LTA and MBA the bank calculates the appropriate risk weighted
assets of the fund. Then the total assets of the fund are divided by the total
risk-weighted assets to give the average risk weight of the fund (Avg
RWfund)1007.

9.4.2 The capital calculation is then the product of the average risk weight for the
fund, the size of the equity investment by the bank and a metric based on the
fund’s leverage. The leverage adjustment is a measure of the underlying
riskiness of the portfolio1008.

9.4.3 Formally, the capital calculation can be set out as follows:

RWAinvestment = Avg RWfund * leverage * equity investment 1009

9.5 Fall-back Approach

9.5.1 Where neither of the preceding approaches can be used (e.g. due to
information limitations) then there is a fall-back approach (FBA). This is a
cross-the-board 1250% risk weight equivalent to a deduction from capital for
the bank’s equity investment in the fund. This is clearly likely to incentivise
banks to obtain the required information, and therefore for funds to provide
it where they seek to secure bank investments.

9.5.2 There is no leverage adjustment as a 1250% risk weight cannot be increased 1010.

9.6 Funds-of-Funds

9.6.1 Funds may invest in other funds. This necessitates a capital approach to
investments in funds-of-funds made by banks.

9.6.2 If a bank invests in one fund (fund A) and that fund invests in one or more
funds (fund B), the bank can use either the LTA or MBA provided the bank
meets the necessary requirements (as to which see above). For three or more
tiers of funds then only the LTA may be used. In any case where the
requirements for the LTA and MBA (if available) cannot be met the fall-back
approach with its penal capital charge applies1011.

9.7 Partial Use of an Approach

9.7.1 A bank may combine the three approaches. For example, it may use the LTA
for those exposures in respect of which it can meet the relevant standards,
and the MBA for others, only using the FBA where it is unable to use either of
the other approaches1012. This is sensible as otherwise if a bank could not use
one approach for all of its equity exposures to funds, then the totality of all
such exposures would be subject to a 1250% risk weight.

 Back to contents 156

 COUNTERPARTY CREDIT RISK

10. COUNTERPARTY CREDIT RISK

10.1 Introduction

10.1.1 Counterparty credit risk (CCR) is the risk that the counterparty to a
transaction could default before the final settlement of the transaction’s cash
flows. An economic loss would occur if the transactions or portfolio of
transactions with the counterparty has positive economic value at the time of
default. Unlike a firm’s exposure to credit risk through a loan, where the
exposure to credit risk is unilateral and only the landing bank faces the risk
of loss, CCR creates a bilateral risk of loss as the market value of a transaction
can be positive or negative on either side of the transaction, and may change
over time with the movement of underlying market factors 1013. Bilateral risk
is explained further below through examples.

10.1.2 CCR may arise on both trading book and banking book transactions. For
example, derivative contracts are more likely to be encountered in the
trading book than the banking book, but need not be eligible for inclusion in
the trading book as a currency swap or an option that is entered into to hedge
an exposure arising in the banking book must be booked in the banking book.
The same applies to long-settlement positions or repo-style transactions not
entered into for trading purposes.

10.1.3 Bilateral risk will now be explained further through the following examples:

If a bank makes a loan, then the only party that can suffer a loss is the
lending bank, as even if the banks subsequently fails, the borrower will
not be affected (usually, it will be required to repay the insolvency
official as per the contractual repayment schedule). However, if a bank
makes a collateralised loan then the bank is exposure to the risk that the
borrower defaults and the value of the collateral is insufficient to cover
the loss on the loan. The borrower is exposed to the risk that the bank
defaults and does not return the collateral. Even if the borrower has a
legal right to offset the amount it owes against the value of the
collateral, the borrower will still suffer a loss if the value of the collateral
is higher than the value of the loan when the bank defaults 1014.

If a bank borrows cash from a counterparty and posts collateral (or enters
into an economically equivalent transaction, such as a repo) then the
bank is exposed to the risk that the counterparty defaults and does not
return the collateral. The risk for the counterparty is that the bank
defaults and the proceeds from realising the collateral are insufficient to
cover the loss on the loan1015.

A bank may borrow securities from a counterparty and post cash as

collateral. In this case the bank is exposed to the risk that its
counterparty defaults and does not return the cash posted as collateral.
The risk for the counterparty is that the bank defaults and the collateral
provided is insufficient to cover the loss on the securities borrowed 1016.

 Back to contents 157

 COUNTERPARTY CREDIT RISK

If a bank enters into a derivatives transaction with a counterparty (e.g.

a swap or purchases an option) then both parties are exposed to risk.
The bank incurs the risk of the counterparty defaulting in circumstances
where the derivative has a positive value for the bank. The counterparty
is exposed to the risk that the bank defaults when the derivative has a
positive value for the counterparty 1017.

10.1.4 The CCR rules address the risk of loss to banks from counterparty default. The
risk of gains or losses on the changing market value of derivative transactions
falls is addressed by the market risk framework described in chapter 11.

10.1.5 Not all transactions that give rise to bilateral risk result in a CCR capital
charge. Basel III only requires a capital charge to be calculated for CCR on
the following four categories of transaction:

OTC derivatives;

exchange-traded derivatives;

long-settlement positions; and

securities financing transactions1018.

10.1.6 The four classes of transactions generally exhibit the following

characteristics:

the transactions generate a current exposure or market value;

the transactions have an associated random future market value based

on market variables;

the transactions generate an exchange of payments or an exchange of a

financial instrument or commodity against payment; and

the transactions are undertaken with an identified counterparty against

which a unique probability of default can be determined 1019.

10.1.7 Other common characteristics are:

collateral may be used to mitigate risks and may be inherent in certain

transactions (e.g. repos);

short-term financing may be a primary objective in that the transactions

mostly consist of an exchange of one asset for another asset (cash or
securities) for a relatively short period of time, usually for the purpose
of financing;

netting may mitigate the risks;

positions are frequently valued; and

re-margining may be used1020.

 Back to contents 158

 COUNTERPARTY CREDIT RISK

10.1.8 Although the latter techniques are common for transactions subject to the
CCR rules, it is not necessary that they are used in any transaction in respect
of which a bank must calculate a CCR capital charge.

10.1.9 The CCR rules differ depending on whether or not the transaction in question
is cleared through a central counterparty (CCP). It follows that banks must
divide up all their transactions subject to a CCR capital charge into two
categories: those not centrally cleared, and those that are centrally cleared.
A bank will always know whether its transactions are centrally cleared.
However, it is not the role of the Basel III framework to determine which
transactions are required to be centrally cleared in any given jurisdiction.
Other legislation, like the Dodd-Frank Wall Street Reform and Consumer
Protection Act in the United States, and the European Market Infrastructure
Regulation (EMIR) in the EU do this. At present, the UK continues to adhere
to a modified version of the EMIR framework as retained EU law 1021.

10.1.10 For non-centrally cleared transactions the following approaches are allowed
to calculate the capital charge for CCR:

a standardised approach (SA-CCR). The standardised approach must be

used if a bank does not have permission to use the internal models
method (IMM-CCR);

a value-at-risk (VaR) models approach to calculate the capital charge for

banks applying an IRB approach to credit risk; and

the IMM-CRR approach, subject to prior supervisory approval 1022.

10.1.11 For centrally-cleared exposures only a single approach is available, which

generates the capital charges for: (i) exposures to a central counterparty
(CC); (ii) exposures to clients if the bank is a clearing member of the CCP and
(iii) exposures to a clearing member if the bank is not a clearing member of
the CCP1023. Exposures to CCPs arising from the settlement of cash transactions
(equities, fixed income securities, spot foreign exchange and spot
commodities) are subject to a bespoke capital regime 1024 described in
chapter 8 on capital charges for Unsettled Transactions and Failed Trades.

10.1.12 The Basel III standard exempts the following transactions from the CCR capital
charge:

credit protection purchased by a bank under a derivative against either

a banking book transaction, or against a counterparty credit risk
exposure; and

sold credit default swaps held in the banking book treated as a guarantee
and risk-weighted at the full amount1025.

10.1.13 These are highly technical exemptions. In the former case, there is no CCR
capital charge as the risk to the bank is addressed under the credit risk
mitigation framework (either standardised or IRB). In the latter case, as the
risk has already given rise to a capital charge, it would be illogical to impose
a second charge under the CCR framework.

 Back to contents 159

 COUNTERPARTY CREDIT RISK

10.2 Definitions relevant to the CCR Framework

10.2.1 We set out below the definitions used by the CCR framework, before
describing the different methodologies for calculating CCR capital charges set
out above.

A long settlement transaction is a transaction where a counterparty

undertakes to deliver a security, a commodity, or a foreign exchange
amount against cash, other financial instruments, or commodities, or the
opposite, at a settlement or delivery date specified in the contract that
is longer than the market standard delivery date for that currency,
instrument or commodity. Any settlement date that is more than five
business days after entry into the transaction is automatically deemed to
be a long settlement position1026. It follows that five days acts as a cap,
and only market practices requiring settlement before then will be
recognised for capital purposes.

A securities financing transaction is a transaction where “the value of

the transactions depends on market valuations and the transactions are
often subject to margin agreements”1027. This is a singularly unhelpful
definition as the value of a great many transactions depends on market
valuations. For example, almost all derivatives satisfy this definition.
However, it is clear from the examples that the range of transactions
caught is narrower. These are: repurchase agreements, reverse
repurchase agreements, securities lending and borrowing, and margin
lending transactions1028. Only repo style transactions are intended to be
caught.

A margin lending transaction is a transaction in which a bank extends

credit in connection with the purchase, sale, carrying or trading of
securities. Margin lending expressly excludes other loans that are
secured by securities collateral. Generally, in margin lending the loan is
collateralised by securities with a greater value than the amount of the
loan1029.

10.2.2 The next set of definitions relates to netting sets and similar terms:

A netting set is a group of transactions with a single counterparty subject

to a legally enforceable bilateral netting agreement for which netting is
recognised for regulatory capital purposes applicable to the group of
transactions. Each transaction that is not subject to an enforceable
bilateral netting arrangement must be interpreted as its own netting
set1030. Netting sets are relevant to many different areas of Basel III which
permit bilateral netting to be recognised for capital purposes. Multi-
lateral netting is not recognised as in many jurisdictions it is not
enforceable in an insolvency1031.

A hedging set is a set of transactions within a single netting set within

which full or partial offsetting is recognised when calculating the
potential future exposure add-ons under the standardised approach to
CCR1032.

 Back to contents 160

 COUNTERPARTY CREDIT RISK

A margin agreement is a contractual agreement, or part of a wider

agreement, under which one counterparty must supply variation margin
to another counterparty when the exposure or exposures to that
counterparty exceeds a specified level1033. In practice, the provision of
variation margin may depend on other factors than simply exposure size
or value.

The margin threshold is the largest amount of margin that remains

outstanding until one party has the right to call for variation margin 1034.
Normally, the trigger for a margin call will be some metric being
exceeded, such as outstanding exposure value. This will be defined
either in the relevant agreement with the counterparty, or by the CCP
where transactions are centrally cleared.

The margin period of risk is defined as the period of time from the last
exchange of collateral covering a netting set of transactions with a
defaulting counterparty until that counterparty is closed out and the
resulting market risk is re-hedged1035. This is different from the general
definition in Basel III. Also, margin can take other forms than collateral,
for example, a letter of credit.

The effective maturity under the Internal Models Method for a netting
set with a maturity of greater than one year is the ratio of the sum of
expected exposure over the life of the transactions in a netting set
discounted at the risk-free rate of return divided by the sum of expected
exposure over one year in a netting set discounted at the risk-free rate1036.
This is calculated using a formula 1037.

Cross-product netting refers to the inclusion of transactions in different

product categories within the same netting set1038.

10.3 Calculation of Capital Charges

10.3.1 The capital charge is based on the amount of the exposure at default. The
CCR framework interchangeably refers to an “exposure” or the “EAD”. This
is potentially confusing, as EAD is a risk input under the IRB framework, and
does not apply to banks using the standardised approach to credit risk. The
reason given is that “the amounts calculated under the counterparty credit
risk rules must typically be used as either the ‘exposure’ within the
standardised approach to credit risk, or the EAD within the internal ratings-
based (IRB) approach to credit risk”1039. Whilst this is correct, consistent use
of a single term would enhance clarity. We refer solely to “exposures” save
when discussing the IRB capital charge.

10.3.2 The exposure amount for a given counterparty is equal to the sum of the
exposure amounts for each separate netting set with that counterparty 1040.
Where there is only a single netting set this may be one, but may be more,
for example if the bank has different exposure types with that counterparty
(OTC derivatives, repos) and no cross-product netting agreement.

10.3.3 A bespoke treatment applies to certain credit valuation adjustments (CVAs)

on OTC derivative transactions with a given counterparty. The CVA measures
the risk of a bank suffering a market risk loss as a result of the deterioration

 Back to contents 161

 COUNTERPARTY CREDIT RISK

of a counterparty’s creditworthiness, and is discussed in the chapter 12. The

exposure value for any given counterparty is the greater of zero and the sum
of all exposures across all netting sets with that counterparty less any CVA
losses that have been deducted from capital as actual losses 1041. The purpose
of this treatment is to prevent losses that have already been deducted from
capital inflating the CCR capital charge, which is understandable.

10.3.4 Once a bank has determined its exposures subject to the CCR framework it
must then apply either the standardised approach or an IRB approach (as the
case may be). Banks applying the standardised approach must use the risk
weights applicable to that counterparty under the standardised approach. IRB
banks apply the relevant IRB approach (standardised or advanced), including
the PD, LGD and EAD, and, where relevant, M, to determine their IRB capital
charge. However, it is the EAD associated with the counterparty credit risk
exposure that must be used in the IRB calculation of RWAs and expected loss
amounts1042.

10.3.5 Banks using the Internal Measurement Method calculate RWAs as the higher of
the following:

the sum of RWA calculated using current parameter calibrations; and

the sum of RWA calculated using stressed parameter calibrations 1043.

10.4 The Standardised Approach to CCR

10.4.1 The standardised approach to CCR (SA-CCR) applies to OTC derivatives,

exchange-traded derivatives and long-settlement positions1044. Securities
financing transactions are excluded. All banks that do not have permission to
apply the Internal Models Method must use the standardised approach 1045.

Netting and netting legal opinions

10.4.2 Basel III recognises netting by novation of derivative transactions. This is a

form of netting under which any existing transaction that has not yet been
performed are novated with any new transaction or transactions with the
same counterparty and in the same currency so as to create a single
performance obligation. For example, if a bank agrees to sell US$ 100,000 for
euro in three months’ time and at a later date agrees to buy US$ 50,000 of
euro to be delivered on the same date then the two contracts can be novated
to produce a single agreement to sell US$ 50,000 for euro on that date.
Netting by novation only applies to derivative contracts with the same
currency and delivery/performance date, and is different from close-out
netting as it applies each time a new transaction in the same underlying for
the same performance date is entered into, as opposed to following a default.
Netting by novation can also be relevant in a default situation where it is not
possible to cancel the outstanding executory contracts following a default,
but only to close them out by entering into an equal and opposite transaction,
with a different counterparty, at a different price after a default. This type
of netting may be used by exchanges to close-out open contracts after a
default and crystallise the profit/loss on the outstanding contracts at or after
a default. However, close-out netting, where each open contract is cancelled
and replaced by a notional profit or loss amount, the sum of which are then

 Back to contents 162

 COUNTERPARTY CREDIT RISK

netted against each other to produce a single sum payable to or by a defaulter,

if appropriate, after realisation and set-off of collateral, is perhaps more
common.

10.4.3 Basel III states that banks may net transactions when determining the
replacement cost (RC) if “any obligation between a bank and its counterparty
to deliver a given currency on a given value date is automatically
amalgamated with all other obligations for the same currency and value date,
legally substituting one single amount for the previously gross obligations”1046.
Other forms of bilateral netting are stated to eligible if they are legally
valid1047. The following detailed requirements apply:

the netting agreement must produce a single legal obligation covering all
included transactions so that the bank will either pay or receive a single
sum representing the positive net mark-to-market values of all included
transactions;

the netting agreement is legally enforceable following default,

bankruptcy or similar circumstances;

the netting agreement does not contain a walkaway clause 1048. This is a
contractual provision entitling a non-defaulting counterparty to make
only limited payments or no payments at all. The prohibition on such
clauses is presumably driven by a concern that if a bank defaults it would
not be able to recover in full any amounts that would otherwise be
payable to it, and this could result in a loss to creditors and, potentially,
depositors;

the bank obtains written reasoned legal advice that in the event of
challenge the relevant courts and administrative authorities would find
the bank’s exposure to be the net amount. This confirmation must be
obtained under the following laws: the law of the place of incorporation
of the counterparty, the law where the branch is located if an overseas
branch, the governing law of the transactions and the law that governs
the netting agreement; and

the legal review referred to above is kept under review in light of possible
changes to the relevant laws1049.

10.4.4 We would make the following observations:

Firstly, only netting by novation seems to be recognised. However, the

more commonly used close-out netting does not. Both are enforceable
under English law, although as English insolvency law is mandatory and
cannot be contracted out of, the netting agreement must follow detailed
legal requirements1050. There is no netting statute in England as it is not
necessary.

Secondly, Basel does not specify what form the legal review must take,
although in view of the requirement that the review is “reasoned” then
an external legal opinion would seem the most straightforward way of
satisfying this requirement. The opinion must also cover all relevant
insolvency procedures applicable to the counterparty, so a standard legal

 Back to contents 163

 COUNTERPARTY CREDIT RISK

opinion with a generic reservation is respect of insolvency opinions is

non-compliant.

Thirdly, the list of laws under which the opinions must cover is logical,
although it will be easier for banks to satisfy the Basel rules by ensuring
that the law governing the individual transactions and the netting
agreement are the same. The term used for incorporation in the Basel
III text is “chartered”, but it must mean the real seat in jurisdictions that
use this test, unless the company is incorporated in one jurisdiction and
has its real seat in a second, in which case both ought to be relevant.
Following the collapse of BCCI in 1991 banking regulators globally
adopted new rules to prevent banks arbitraging between the place of
incorporation and where they actually carry on their business, so the
number of cases where this is permitted should be very rare1051.

Fourthly, some procedure for updating the legal opinions needs to be

established. Unless local supervisors specify otherwise, this could take
the form of refreshing the legal opinions every year (or obtaining a new
one where there are material changes to the relevant law).

10.4.5 Basel III states that national supervisors must be satisfied, after any relevant
consultation with other supervisors, that the netting agreement is enforceable
under all relevant laws. If a supervisor is not so satisfied the netting
agreement is not recognised1052. It is understandable that if a national
supervisor considers netting agreements to be unenforceable then they
cannot be relied on to reduce regulatory capital charges. However, unless a
supervisor is told by a regulator in a different jurisdiction that netting is
invalid then we consider it highly unlikely that a national supervisor would
seek to second guess a legal opinion from a law firm in the relevant
jurisdiction stating that it is. Netting is recognised in most major
jurisdictions, although there may be specific requirements that need to be
met under local law for an agreement to be enforceable.

Calculating the capital charge under the standardised approach

10.4.6 The following factors are taken into account in determining the capital charge
under the standardised approach:

α = 1.4 (it is a scaling factor to generate higher capital charges to reflect

possible weaknesses in the calculation as a measure of risk);

RC = the replacement cost (as calculated in accordance with the

following calculations); and

PFE = the amount for potential future exposure1053.

10.4.7 We now proceed to explain RC and PFE.

Margined and un-margined transactions

10.4.8 Both RC and PFE must be calculated differently for margined and un-margined
netting sets. Margined sets are netting sets covered by a margin agreement
requiring the posting of variation margin. Un-margined transactions are all

 Back to contents 164

 COUNTERPARTY CREDIT RISK

other netting sets, including where the agreement requires only the bank (and
not its counterparty) to post margin1054.

10.4.9 There is a cap on the exposure value for margined netting sets. This is the
exposure amount had the netting set not been margined 1055, which makes
sense. The reason given by the Committee is “the need to ignore exposure
from a large threshold amount that would not realistically by hit by some
small (or non-existent) transactions”1056.

10.4.10 For un-margined transactions the replacement cost (RC) captures the loss that
a bank would suffer should a counterparty default and the bank closed-out all
of its transactions immediately1057.

10.4.11 For margined transactions RC also in intended to capture the loss that a bank
would incur if a counterparty defaulted and all transactions were replaced
immediately. However, there is a difference as there may be a period of time
between the last exchange of collateral and the default of the bank’s
counterparty without variation margin having been provided 1058. The RC for
margined transactions should, however, be lower for margined than un-
margined transactions because of the existence of variation margin under the
agreement.

10.4.12 Any non-cash collateral must be subject to a “haircut” in both cases. The
“haircut” represents the potential change in value of the collateral during the
relevant time frame (one year for un-margined trades, and the margin period
of risk for margined trades)1059.

10.4.13 The Replacement cost (RC) is calculated at the netting set level.

10.4.14 As has been seen, RC is calculated differently for margined and un-margined
transactions. Margined transactions may include both bilateral transactions
(such as under an ISDA Master Agreement) and those that are centrally
cleared1060. However, in the case of centrally cleared transactions the
provisions on margining are likely to be set out in any relevant exchange’s or
CCP’s rules.

RC for un-margined transactions

10.4.15 RC for un-margined transactions is the greater of: (i) the current market value
of the derivative contracts less the market value of the collateral (after
applying haircuts) held by the bank (if any); and (ii) zero 1061. The following
steps must be taken. Firstly, the bank calculates the market value of the
outstanding derivative contracts. Secondly the bank values the collateral it
has received from its counterparty (if any). Thirdly, the bank applies
applicable haircuts to the value of the collateral (which are the haircuts under
the standardised approach to credit risk, adjusted as described below).
Fourthly, the bank deducts the haircut-adjusted value of the collateral from
the exposure amount. If this amount is positive then this is the RC of the
transactions. If the amount is negative as the bank has to make a payment to
the counterparty then the RC is set at zero.

10.4.16 The following formula applies (which is simply a formal way of writing what
has been described in the preceding paragraph)1062:

 Back to contents 165

 COUNTERPARTY CREDIT RISK

RC = max {V – C; 0}1063

where:

V = the value of the derivative transactions in the netting set; and

C = the value of the collateral after application of haircuts 1064.

RC for margined transactions

10.4.17 The RC for margined transactions is defined as the greatest exposure that
would not trigger a margin call (i.e. the posting of variation margin) based on
the terms of the margining agreement 1065. For example, the margin agreement
may include both a threshold amount, triggering the obligation to provide
variation margin, and a minimum amount of margin that must be posted in
that event. Margin may be transferred by way of security (as tends to be more
common in the United States) or by outright transfer of title (as is perhaps
more common under English law).

10.4.18 The calculation of the RC is described in the following paragraphs. As

margined transactions incorporate initial and variation margin, it is necessary
to reflect the actual process for the exchange of collateral over the life of the
transaction. Basel III introduces two new terms: the independent collateral
amount (ICA) and the net independent collateral amount (NICA). These will
now be explained.

10.4.19 The ICA represents: (i) collateral (other than variation margin) posted by a
bank’s counterparty that the bank may realise upon the default of the
counterparty, the amount of which does not change in response to the value
of the transactions it secures and/or (ii) the “independent amount” defined
in market standard documentation1066. This will now be unpacked.

Determining the ICA

10.4.20 There are two limbs to the determination of the independent collateral
amount (ICA): collateral posted by the counterparty that meets specified
requirements and collateral that falls under the definition of an “independent
amount”.

10.4.21 The first part comprises any collateral posted by the bank’s counterparty that
meets two requirements: (i) it is not “variation margin” and (ii) the amount
of the collateral posted is insensitive to the value of the transactions. In most
cases these two will be the same as initial margin will be insensitive to the
value of current transactions, as that is the purpose of variation margin,
although it is possible to draft contractual provisions that operate differently,
so this distinction may be necessary.

10.4.22 The concept of an “independent amount” seems intended to capture amounts

described as such under market standard documentation. There is no
definition in the Basel III text, but a footnote refers to the 1992 and 2002 ISDA
Master Agreements and supporting security documentation1067.

 Back to contents 166

 COUNTERPARTY CREDIT RISK

10.4.23 Because both a bank and its counterparty may be required to post ICA, Basel
III uses the concept of a net independent collateral amount or NICA. NICA is
simply the amount of collateral that a bank may use to off-set its exposure on
the default of its counterparty1068. The point is that if such collateral exists
and can be realised then there is no risk of an actual loss materialising.
Expressly excluded are amounts posted to a segregated and bankruptcy-
remote account that would be returned after the insolvency of the
counterparty1069. Under English law this would require that full title not pass
to the insolvent counterparty. The most usual way of doing this is through a
trust account, although other arrangements may suffice1070.

NICA

10.4.24 NICA is also defined as “any collateral (segregated or unsegregated) posted by

the counterparty less the unsegregated collateral posted by the bank”1071. This
need not be the same thing and, presumably, any collateral posted by the
counterparty must be legally available to the bank in the event of the
counterparty’s default, and any segregated collateral posted by the bank must
not be legally available to the counterparty’s creditors. Segregation may help
achieve this, but it is not identical.

Calculating the RC

10.4.25 Having calculated the ICA and the NICA banks then determine the replacement
cost (RC) for margined transactions in accordance with a formula which is set
out below. The idea is that RC will be equal to the larger of three figures.
The first is the value of the derivative exposures less the haircutted collateral
value (the calculation this time must include variation margin). The second
figure measures the sum of the threshold amount before the counterparty is
required to post variation margin plus the minimum amount of collateral that
will be posted less NICA (including both initial and variation margin). The idea
is that where the net independent collateral amount (i.e. the amount of
collateral the bank can use to reduce its exposure to its counterparty’s
default) is less than the exposure amount before the counterparty is required
to post variation margin plus the amount of total margin that the counterparty
must then post then the bank will suffer a credit loss. Although perhaps
counter-intuitive on a first reading, this makes sense once the reader
remembers that the whole point of the CCR framework is to determine under
which circumstances a bank may suffer a credit loss if its counterparty
defaults.

10.4.26 The third figure is zero. This underscores the policy choice made by the Basel
Committee that the replacement cost can never be a negative amount where
the bank does not suffer a loss if its counterparty defaults. If it could then a
bank could reduce its capital charge for CCR by the amount of those exposures
it must pay which would not be prudent as a payment obligation does not
reduce in any way the loss a bank will suffer if its counterparties default.

10.4.27 The relationship between the three figures will now be explained. The first
figure measures the loss to the bank based on amount of the derivative
exposure less the initial and variation margin the bank can use to reduce the
exposure where its counterparty defaults. The second figure measures the
loss to the bank as a result of its counterparty failing to post variation margin

 Back to contents 167

 COUNTERPARTY CREDIT RISK

after realising all collateral already provided. The third figure is zero so as
to exclude from the calculation the circumstances where the bank is over-
collateralised, to prevent such excess collateral that would need to be
returned to its counterparty in a default scenario 1072 affecting the calculation.
Seen in this light it is logical that the greatest loss a bank may suffer in a
default scenario is the greater of a margin shortfall based on already posted
margin or a variation margin shortfall. Although a bank may suffer both types
of loss, only the highest figure represents the worst case scenario as these are
in both cases actual losses.

10.4.28 Having explained the RC for margined transactions we will now set out the
formula which is again a formal way of writing what has already been
explained.

10.4.29 The following definitions apply:

V = the value of the derivative transactions in the netting set;

C = the collateral value of any initial margin provided (after application

of relevant haircuts). C now includes the net variation margin amount
(i.e. the variation margin received by the bank less variation margin
posted by it);

TH = the positive threshold before the bank’s counterparty must post

variation margin; and

MTA = the minimum amount of collateral that must be posted in this

event1073.

10.4.30 The replacement cost for margined transactions is calculated as follows:

RC = max {V – C; TH + MTA – NICA; 0}1074

10.4.31 TH + MTA – NICA represents the largest exposure that would not trigger a
variation margin call and contains the levels of collateral that need always to
be maintained. The subtraction of NICA is intended to reflect both the actual
level of exposure that would not trigger a margin call and the effect of
collateral already posted to the bank1075.

Potential future exposure (PFE)

10.4.32 Basel III requires banks to calculate an add-on to the RC to address the risk
that at the time of default the size of the exposure may increase due to
movements in market values. For example, an option with a strike price of
$100 may have a current market price of $110. The risk of losses from
movements in market prices is addressed through the market risk framework.
However, if the market price goes up to $120 then the bank may suffer an
additional $10 loss from that currently captured under the CCR framework.
As market prices can change it is prudent to require a capital calculation to
address this potential credit risk. This is called potential future exposure or
PFE and attracts an add-on.

 Back to contents 168

 COUNTERPARTY CREDIT RISK

10.4.33 Over-collateralisation is also relevant to the calculation of PFE. This is

because if the exposure increases excess collateral can then be applied to
reduce the size of the exposure. Providing the bank can apply such excess
collateral this is also sensible. Additionally, negative values for transactions
within a given netting set will reduce future CCR losses as all the transactions
in the netting set are assumed to be netted out. This ensures that the PFE
calculation does not overstate the actual credit risk from market price
movements.

10.4.34 The PFE calculation therefore has three elements: the add-on for future
potential CCR, excess collateral, and negative mark-to-market values. The
last two are addressed through a multiplier1076. This will now be explained.

Add-ons

10.4.35 Add-ons are supervisory adjustments to reflect PFE. The standardised

approach requires exposures to be broken down by asset class, with each asset
class attracting a different add-on, although the process for calculating add-
ons is very similar. The asset classes for this purpose are:

interest rate derivatives;

foreign exchange derivatives;

credit derivatives;

equity derivatives; and

commodity derivatives1077.

10.4.36 Basel III does not permit recognition of any diversification benefits across
asset classes. Instead, the add-ons for each asset class are simply aggregated.
Formally, this is expressed as follows:

AddOnaggregate = Σassetclass AddOn(assetclass) 1078

10.4.37 Derivatives may be more complex than the list of asset classes set out above
and exhibit features of more than one asset class. In this case the allocation
of transactions to any given asset class must be made based on its primary
risk driver. The text states that most derivative transactions have one primary
risk driver, defined by reference to the underlying instrument (e.g. the
interest rate curve for interest rate derivatives, a reference entity for a credit
default swap). If the primary risk driver can be identified it must be used 1079.

10.4.38 For more complex derivatives with multiple risk drivers (e.g. multi-asset or
hybrid derivatives) banks are required to identify the primary risk driver based
on sensitivities and volatility of the underlying 1080. Supervisors may require
more complex derivatives to be allocated to more than one asset class 1081. In
this case the bank must determine separately for each asset class to which
the derivative is allocated the sign and delta adjustment (see below)1082.

10.4.39 In all other cases, the bank will identify the primary risk factor and attribute
each transaction to one of the five asset classes 1083. The add-ons for each

 Back to contents 169

 COUNTERPARTY CREDIT RISK

asset class are then calculated using asset class-specific formulae1084. Although
the formulae differ from asset class to asset class, they all use the steps set
out in the following paragraphs1085.

Effective notional amount

10.4.40 The first factor is determination of the effective notional (D). The effective
notional is calculated for each derivative (i.e. every single trade in the netting
set). D measures the sensitivity of the trade to movements in underlying risk
factors. D is calculated through applying a formula 1086.

10.4.41 The following factors are used to calculate D: (i) the adjusted notional or d;
(ii) the maturity factor or MF; and (iii) the supervisory delta or δ. These will
now be described.

10.4.42 The adjusted notional or d is a measure of the size of a trade. For FX

derivatives it is the notional value of the FX leg of the derivative contract.
For equity and commodity derivatives it is the current price of the relevant
share or unit of the commodity multiplied by the number of shares or
commodity units held by the bank (i.e. the market value of the bank’s current
notional exposure). For credit and interest-rate derivatives the adjusted
notional is the notional amount adjusted by a measure of the duration of the
notional exposure. This is to reflect the fact that that the value of derivatives
with a longer duration are more sensitive to movements in the price of
underlying risk factors1087.

10.4.43 The maturity factor or MF is a parameter that takes account of the period of
time over which the potential future exposure or PFE is calculated. MF
depends on whether or not the netting set is margined or un-margined1088.

10.4.44 The supervisory delta or δ is used to take into account whether the trade is
long or short. This is done through a sign which is positive for long positions
and negative for short positions. It also takes into account whether the value
of the derivative is linear to the value of the underlying (e.g. a forward) or
non-linear (such as an option)1089. The value of derivatives is non-linear to that
of the underlying if a given change in the market price of the underlying result
in a non-proportional change in the value of the derivative e.g. if a change of
10% in the price of the underlying results in a 30% change in the value of the
derivative.

10.4.45 The formula to calculate D is as follows:

D = d * MF * δ1090

D is therefore the product of the adjusted notional representing the value of

the transaction, a maturity factor and the supervisory parameter δ.

Supervisory factors

10.4.46 The second factor used is a supervisory factor or SF. This is the specified
change in the value of the underlying risk factor on which the PFE is
calculated. It is calibrated to take into account the volatility of underlying
risk factors1091.

 Back to contents 170

 COUNTERPARTY CREDIT RISK

Calculating PFE add-ons

10.4.47 In calculating PFE each of the transactions within each asset class are
allocated to supervisory hedging sets. The purpose of a hedging set is group
together transactions within a given netting set in respect of which long and
short positions can be offset against each other when calculating PFE 1092. The
idea that where there are matching long and short positions then the
maximum potential loss to the bank is represented by the net long or short
position and not by aggregating long and short positions.

10.4.48 Aggregation formulae are used to aggregate the effective notional amounts
and supervisory factors across all transactions within each hedging set and
finally at the asset class level to generate the add-on for each asset class on
an aggregate basis. The method of aggregation varies between the different
asset class and for credit derivatives, equity derivatives and commodity
derivatives involves the use of supervisory parameters to take into account
the diversification of transactions as well as basis risk1093. These are set out in
the table below1094.

Summary table of supervisory parameters

Asset class Sub-class Supervisory Correlation Supervisory
factor option
volatility1095
Interest rate 0.5% N/A 50%
Foreign exchange 4% N/A 15%
Single AAA 0.38% 50% 100%
name
AA 0.38% 50% 100%
credit
derivative A 0.42% 50% 100%
BBB 0.54% 50% 100%
BB 1.06% 50% 100%
B 1.6% 50% 100%
CCC 6.0% 50% 100%
Index IG 0.38% 80% 100%
credit
SG 1.06% 80% 100%
derivative
Single 32% 50% 120%
name
equity
Equity 20% 80% 75%
index
Commodity Electricity 40% 40% 150%
Oil/gas 18% 40% 70%
Metals 18% 40% 70%
Agricultural 18% 40% 70%
Other 18% 40% 70%

 Back to contents 171

 COUNTERPARTY CREDIT RISK

10.4.49 The standardised approach uses four time parameters, which are all expressed
in years1096. It is outside the scope of this guide to set out or describe the
detailed calculations.

10.4.50 The add-ons are determined for each asset class. Their calculation is set out
in the text of Basel III.

PFE and excess collateral

10.4.51 The Basel Committee considers that over-collateralisation (i.e. where a bank
holds excess collateral posted by a counterparty) should reduce capital
requirements for CCR. This is because the bank can realise such collateral to
cover a credit loss if its counterparty defaults, and banks may insist on over-
collateralisation precisely in order to mitigate potential future losses from
changes in market variables. Such excess collateral above that needed based
on the current market value of the derivative positions may reduce both the
RC and the PFE calculations1097.

10.4.52 For prudential reasons the Basel Committee does not allow all of excess
collateral to be eligible to reduce capital charges. Instead, a multiplier is
applied so that recognition of excess collateral decreases as the amount of
such collateral increases, without ever reaching zero. A floor of 5% of the PFE
add-on applies to such collateral recognition1098. Where a bank is over-
collateralised then RC is zero and the PFE is less than the full amount of the
add-on.

10.4.53 The multiplier is activated where the current value of the derivative
transactions in the netting set is negative. This reflects the fact that
transactions that are out of the money do not give rise to a credit exposure,
and are less likely to move to being in the money 1099.

10.4.54 In calculating the multiplier the following apply:

exp (…) is the exponential function;

the floor is 5%;

V is the value of the derivative transactions in the netting set; and

C is the value of excess collateral held net of applicable haircuts 1100.

𝑉−𝐶
𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑖𝑒𝑟 = min{1, 𝑓𝑙𝑜𝑜𝑟 + (1 − 𝑓𝑙𝑜𝑜𝑟) * exp( )}
2*(1 − 𝑓𝑙𝑜𝑜𝑟)*𝐴𝑑𝑑𝑂𝑛aggregate

10.4.55 The PFE add-on for un-margined transactions represents a potential

conservative increase in exposure size over a one-year time horizon1101. For
margined transactions the PFE add-on represents the potential change in
value of the exposure since the last exchange of collateral1102.

10.4.56 The PFE add-ons are calculated at the level of each netting set and then
aggregated1103.

10.4.57 Exposures are calculated as follows:

 Back to contents 172

 COUNTERPARTY CREDIT RISK

exposure amount = α x (RC + PFE)1104

10.4.58 Sold options receive a zero exposure value if premiums are paid up front and
are outside any netting or margin agreement 1105. This is because if such options
are outside a netting set, and the premium has been paid up front, the options
constitute a potential liability only as the bank will suffer no loss if its
counterparty defaults.

10.4.59 Where a bank sells credit derivatives that are outside of any netting set the
exposure is capped at the amount of any unpaid premium 1106. This also is
sensible as in this case the bank’s only exposure to its counterparty is the
amount of any unpaid premia. The difference between options and credit
derivatives would seem to reflect market practice that the price of an option
is paid up front, whilst credit derivatives require periodical payments to the
protection provider.

10.4.60 Banks may decompose more complex derivatives (e.g. an option with caps and
floors) into a series of notional derivatives with individual caps and floors to
determine the CCR capital charge. Linear derivatives (e.g. interest rate
swaps) cannot be decomposed1107.

A bird’s eye view of Counterparty Credit Risk

Under Basel III, financial institutions may opt to calculate their counterparty credit
risk (CCR) risk-weighted assets (RWA) using the revised standardised approach for
measuring counterparty credit risk (SA-CCR) or, subject to regulatory approval, the
internal model method (IMM).
The total exposure at default (EAD) under the SA-CCR consists of two components,
the replacement cost (RC) and the potential future exposure (PFE), with alpha as a
constant value set to 1.4 by the Committee, in line with the IMM.
The RC quantifies the immediate loss that would occur if a counterparty were to
default. It is calculated as the total mark-to-market (MtM) of the derivative trades
at the netting set level less collateral.
The PFE consists of (i) a multiplier that allows for the partial recognition of excess
collateral or negative market values and (ii) an “aggregate add-on”, which is the sum
of five asset-class level add-ons.

10.5 Internal Models Method

10.5.1 The second approach to calculating CCR is the Internal Models Approach or
IMM. Banks can only use this approach with prior supervisory consent 1108, and
use of IMM is dependent on meeting the minimum criteria1109. However, the
use of the IMM is independent of which approach banks use to calculate credit
risk i.e. a bank using the standardised approach to credit risk may use the IMM
approach, and an IRB bank may use the standardised approach to CCR if it
does not obtain approval to use IMM 1110.

10.5.2 It follows that banks that apply the standardised approach to credit risk must
use a model that is compatible with the standardised approach when
calculating IMM capital charges. On the other hand, an IRB bank must use the
relevant IRB approach(es) to calculate IRB capital charges for CCR under an

 Back to contents 173

 COUNTERPARTY CREDIT RISK

IMM model. For this reason the Basel III standard refers indiscriminately to
either the exposure value or the EAD.

10.5.3 Banks that use the IMM must, in principle, apply it to all transactions on which
it is required to calculate CCR except long-settlement transactions1111 (which
may be treated under the standardised approach). This is, however, subject
to an important qualification. Banks may adopt the IMM either only for OTC
derivative transactions, only for securities financing transactions (SFTs), or for
both1112. This seems curious as the treatment, in this case, of exchange-traded
derivatives is uncertain. Centrally-cleared exchange-traded derivatives may
be cleared using a CCP, but this is not necessarily the case in all jurisdictions.
Settlement must, of course, take place, but transactions can settle through
other means, for example, by a clearing house guaranteeing performance of
all transactions without becoming a party to them. The Basel III text seems
to assume all exchange-traded derivatives will also be cleared through a CCP.

10.5.4 Where a bank uses the IMM for OTC derivatives or SFTs, it must apply that
method to all relevant exposures within that category, other than exposures
that are immaterial in terms of size and risk. However, during the initial
implementation of the IMM a bank may continue to apply the standardised
approach for a portion of its business. However, the bank must submit a plan
to its supervisor explaining how it will bring all material exposures under
IMM1113.

10.5.5 Capital requirements in respect of CCR arising out of long-settlement

transactions may be calculated using either IMM or the standardised
approach1114.

10.5.6 Once a bank moves from the standardised approach to IMM only under
exceptional circumstances will the bank be allowed to revert to the
standardised approach, and only where this would not lead to an arbitrage of
regulatory capital rules1115.

10.5.7 Transactions not covered by a firm’s IMM approval must be treated applying
the standardised approach1116.

10.5.8 Under IMM, CCR is calculated at the level of individual netting sets. An
internal model must specify the forecasting distribution for changes in the
market value of each netting set attributable to changes in market variables,
such as interest rates, FX rates, etc. The model then computes the bank’s
CCR for each netting set at each future date given the changes in the market
variables. For margined counterparties, the model may also capture future
movements in collateral provided. In respect of transactions in the banking
book, only the forms of collateral recognised under the simple approach to
collateral under the standardised approach are eligible1117. Collateral eligible
under the comprehensive approach and the IRB approach are excluded.
However, for trading book exposures all instrument that are included in the
trading book are eligible for repo-style transactions1118. The qualitative and
quantitative data requirements set out below must be met in respect of the
collateral1119. Collateral cannot be double counted1120.

10.5.9 The capital charge for CCR under the internal models method depends on the
higher of two amounts. The first is calculated using current parameter

 Back to contents 174

 COUNTERPARTY CREDIT RISK

estimates, and the second is based on stressed parameter estimates.

Specifically, the bank must determine the effective expected positive
exposure amount using both current and stressed figures on a total portfolio
basis (i.e. the stressed calibration is applied to the portfolio as a whole and
not to individual transactions)1121.

10.5.10 It should be stated that banks are not required under IMM to employ a single
model. Any models are acceptable so long as they are approved by the
national supervisor, and meet the relevant minimum requirements 1122.

10.5.11 When calculating “expected exposure” and “peak exposure”, banks must take
into account (where appropriate1123) of the possibility of fat tails1124. A “fat
tail” is, basically, a situation where there is a higher than expected probability
of an extreme movement in price occurring1125. This occurred during the global
financial crisis, notably after the collapse of Lehman Brothers.

10.5.12 Under the IMM the exposure amount is equal to the effective expected
exposure amount multiplied by a multiplication factor alpha (α)1126.

10.5.13 Alpha is generally set at 1.41127 representing a 40% uplift. Supervisors may set
a higher α based on an individual bank’s CCR exposures. Factors that could
justify such a supervisory uplift include a low granularity of counterparties,
high “wrong-way” risk (where the risk of loss is positively correlated to the
risk of a counterparty or counterparties defaulting) and other institution-
specific characteristics1128.

10.5.14 Banks can calculate α based on internal estimates with prior supervisory
approval. In this case α is always floored at 1.2 (a 20% uplift). Banks must
estimate α as the ratio of economic capital from a full simulation of
counterparty credit risk exposure1129 to the economic capital requirement
based on the expected positive exposure (EPE) 1130. In order to be able to use
own estimates for α banks must show that their internal estimates capture
the stochastic dependency of distributions of market values of transactions or
portfolios of transactions (e.g. the correlation of defaults across
counterparties and between market risk and default risk)1131. Specified
requirements apply1132. Only banks “in full compliance with the qualitative
criteria will be eligible for application of the minimum multiplication
factor”1133.

10.5.15 The effective expected positive exposure (or effective EPE) is calculated by
the bank estimating the effective exposure (EEt) as the average exposure at
future date t, where the average is taken across possible future values of
relevant market risk factors, such as interest rates, foreign exchange rates,
etc. The model must estimate expected exposure (EE) at a large number of
future dates. Specifically, it is modelled recursively using the following
formula:

Effective EEtK = max (Effective EEtk – 1, EEtk)

10.5.16 A one year time horizon normally applies, unless all transactions in the netting
set expire before then, in which case the time horizon is the weighted average
of all exposures, calculated using a formula 1134, which is modified if there are
exposures with an original maturity of over one year 1135.

 Back to contents 175

 COUNTERPARTY CREDIT RISK

10.5.17 Where transactions are margined, and the internal model captures the effects
of margin, then this is permitted1136. The Basel III text notes that “[s]uch
models are noticeably more complicated than models of EPE for unmargined
counterparties. As such, they are subject to a higher degree of supervisory
scrutiny before they are approved”1137. Detailed requirements apply1138.

Requirements for validation of internal models

10.5.18 There are specific requirements to ensure the integrity and soundness of
banks’ internal models under IMM. The qualitative criteria include:

a regular programme of backtesting (an ex-post comparison of the

model’s performance against actual outcomes);

initial validation and on-going periodic review of the model by persons

not responsible for developing the model;

the Board of Directors must be actively involved in the risk control

process and must regard CCR control as an essential part of the business
devoting significant resources to this task;

there are daily reports prepared by an independent risk control unit for
management with sufficient seniority to be able to order reductions in a
bank’s risk exposure;

the bank’s exposure model is closely integrated into the bank’s day-to-
day risk management process;

there are internal trading and exposure limits;

the bank has a documented set of internal policies, controls and

procedures concerning the operation of the risk measurement system;

there is an internal review at least once per year of the bank’s risk
management system. Specific requirements apply; and

the validation of the model is periodically reviewed by management with

sufficient authority to mandate corrections 1139.

10.5.19 All IMM models must be sufficiently documented to enable third party
recreation of the analysis underlying the model 1140. There are also further
detailed requirements1141.

10.5.20 The operational requirements include:

the internal model’s output is closely integrated into the bank’s daily
management of CCR1142;

the bank has a credible track record in using CCR models, and has been
using a CCR model that broadly meets the minimum requirements for at
least one year1143;

 Back to contents 176

 COUNTERPARTY CREDIT RISK

the bank has an independent control unit responsible for designing and
implementing the bank’s management of CCR risks 1144;

the bank has a collateral management unit that calculates and makes
margin calls, and manages margin disputes 1145;

there is a collateral management unit that meets certain criteria 1146;

the bank’s cash management policies account for liquidity risks on

incoming margin calls, as well as calls to return excess collateral and
margin calls triggered by a credit rating downgrade1147;

banks must be able to calculate EE daily if necessary, although it is not a

requirement that they actually do so1148;

banks must have in place sound stress testing procedures when assessing
capital adequacy requirements1149;

there are data requirements. Where historic market data are used a bank
must have at least three years’ of data1150; and

the general guidance set out by the Basel Committee elsewhere on the
use of internal models (this is undefined and therefore unclear) must be
followed1151.

10.5.21 There are specific requirements for so-called “wrong-way” risk. As

mentioned, this is the risk that potential future credit exposures to a
counterparty or counterparties is/are highly correlated to the risk of that
counterparty or those counterparties defaulting. Banks must employ stress
testing and scenario analysis to address such risk, and monitor wring-way risk
by product, region, industry or other factors, with reporting to management
and an appropriate committee of the Board of Directors1152. An example of
wrong-way risk is a company writing put options on its own shares that are
purchased by a bank. The value of such options is directly correlated with
the bank’s own creditworthiness as the stock price may be expected to
decline in response to a fall in the company’s credit standing. Another
example is a counterparty selling credit default swaps to a bank on a related
party whose insolvency would affect the seller 1153.

Cross-product netting

10.5.22 This is possible for OTC derivatives1154, long-settlement positions and securities
financing transactions. Banks may include either SFTs or both SFTs and OTC
derivatives1155 under a cross-product netting agreement. There is no mention
in the Basel III text of long-settlement positions, which may be included within
the CCR IMM framework, as we have seen.

10.5.23 National supervisors are permitted to impose approval requirements for the
prior use of cross-product netting agreements1156. An example is that
published by ISDA. There are specific legal criteria for such agreements. A
bank must have a written bilateral netting agreement with its counterparty
that creates a single legal obligation covering all master agreements and
transactions. The effect of the agreement must be that the bank has either

 Back to contents 177

 COUNTERPARTY CREDIT RISK

an obligation to pay, or a claim to receive, the net sum of the close-out values
of all individual master agreements and the mark-to-market value of all
included transactions, following a default or insolvency 1157. This is a slightly
circular requirement as normally under a cross-product netting agreement all
transactions under each master netting agreement entered into between the
parties are closed-out and then the sums payable under each netting
agreement are then netted inter se. Walkaway clauses are prohibited1158.

10.5.24 Banks relying on a cross-product netting agreement must have “written and
reasoned legal opinions that conclude with a high degree of certainty that, in
the event of a legal challenge, the relevant courts or administrative
authorities would find the bank’s exposure under the cross-product netting
arrangement to be the cross-product net amount under the laws of all relevant
jurisdictions. In reaching this conclusion, legal opinions must address the
validity and enforceability of the entire cross-product netting agreement on
the material provisions of any included bilateral master agreement”1159.
Where the cross-product master netting agreement is sponsored by a trade
association (like ISDA) we expect that there will be a reasoned legal opinion
provided to all members of the association that can be relied on by members
(in which case a bank seeking to rely on the opinion will need to become a
member). In case of a bespoke bilateral arrangement (which is rare in
practice) then specific legal opinions will need to be obtained.

10.5.25 The definition of “all relevant laws” is the same as has been considered above
under the standardised approach1160, which is unsurprising. The legal opinion
must also “be recognised as such by the legal community in the bank’s home
country or a memorandum of law that addresses all relevant issues in a
reasoned manner”1161. There is no process for the recognition of legal opinions
“by the legal community” in England. However, a reasoned legal opinion
addressing all insolvency and resolution procedures should suffice.

10.5.26 Banks are required to ensure that any given transaction treated under the
cross-product netting rules is covered by the relevant legal opinion(s) 1162,
which is a question of legal due diligence. Opinions must be updated 1163 (the
frequency of which is not specified). Banks must also retain all required
documentation in their files1164. Presumably, this refers to the relevant
netting agreements and not legal opinions where they are obtained by a trade
association and issued to all their members.

10.5.27 National supervisors must be satisfied that banks manage counterparty credit
risk on covered transactions on a net basis 1165. Credit limits and economic
capital processes must do the same1166.

10.6 CCR in the Trading Book

10.6.1 CCR can arise in exactly the same way on transactions booked in the trading
book. Indeed, it is perhaps more common in the trading book. For this reason
the CCR framework applies to both banking and trading book transactions,
with the same set of approaches outlined above. However, there are a few
changes.

10.6.2 Unsurprisingly, the CCR capital charge must be calculated independently of

the market risk capital charge (which deals with risks arising out of

 Back to contents 178

 COUNTERPARTY CREDIT RISK

movements in market prices, as well as credit valuation adjustments) 1167. Risk

weights in the trading book must be “consistent” with those used in the
banking book. Banks using the standardised approach must use standardised
risk weights, and IRB banks are required to use the IRB risk weights 1168.

10.6.3 As mentioned above, for repo-style transactions any instruments included in

the trading book are eligible collateral 1169. The “haircut” applicable to such
instruments is that applicable to non-main index securities under the
comprehensive approach to collateral1170. Where a VaR approach is used under
the standardised approach to credit risk for securities financing transactions
then this model may likewise be used for any trading book transactions 1171.

10.7 Exposures to CCPs

10.7.1 As was noted above a bespoke capital treatment applies to exposures to

central counterparties. This approach applies to all transactions subject to
the CCR framework with CCPs, whether in the trading or banking books. The
only exception is for cash transactions (equities, bonds, spot foreign exchange
and spot commodity transactions)1172. Such transactions are subject to the
treatment described in chapter 8.

10.7.2 Where a clearing member-to-client leg of an exchange-traded derivatives

transaction is conducted under a bilateral agreement, both the client bank
and the clearing member must treat the transaction as an OTC derivative 1173.
In such a case the clearing member enters into one transaction with the CCP
and a second back-to-back transaction with its counterparty. The CCR risk on
the centrally-cleared transaction is the credit risk to the clearing member
should the CCP default, and the CCR risk on the bank is the risk of the clearing
member defaulting. However, this blanket rule does not take into account of
the possibility that the exchange (where there is one) or the CCP may close-
out the transactions between a clearing member and the bank. If this is
discretionary, then the treatment seems logical, but if mandatory then it
seems unduly prudent.

Definitions

10.7.3 The following definitions apply to transactions subject to the specific

mandatory treatment of CCP-cleared transactions:

A central counterparty is defined as a clearing house that interposes

itself between counterparties to contracts traded in one or more
financial markets, becoming the buyer to every seller, and the seller to
every buyer, thereby ensuring the performance of open contracts 1174.
According to Basel III a CCP becomes a party to trades with market
participants either through novation, an open offer system, or other
legally binding arrangement1175. For the purpose of the Basel framework,
a CCP is treated as a financial institution. This definition is curious as
becoming a party to each buy or sell transaction brokered between
participants on an exchange is the function of a central counterparty. A
clearing house is the entity that is legally responsible for ensuring the
settlement of transactions on an exchange. In the UK all recognised
investment exchanges are required by law to make arrangements for the
settlement of transactions1176. However, neither an investment exchange

 Back to contents 179

 COUNTERPARTY CREDIT RISK

nor a clearing house need be a central counterparty 1177, nor has this
historically been the case. A clearing house can use other methods of
ensuring settlement without becoming a party to transactions, and
clearing houses have often done so. Purely OTC clearing houses
(unaffiliated to any exchange) also exist. The reference to a CCP being
a clearing house therefore is confusing.

Central counterparties are further subdivided into “qualifying” and non-

qualifying central counterparties.

A qualifying central counterparty is an entity licensed to operate as a

CCP (including a license granted by way of confirming an exemption) and
is permitted to operate as such. The CCP must be based and prudentially
regulated in its jurisdiction in accordance with the Principles for
Financial Market Infrastructures published by the Committee on
Payments and Market Infrastructures and IOSCO 1178. If, however, the CCP
is established in a jurisdiction where there is no regulation of CCPs then
the banking supervisor can assess if the foregoing requirements are
met1179. In addition, specified information must be provided either by the
CCP or another person to enable calculation of the capital charge for
exposures (see below), as well as enabling the supervisors of clearing
member banks to verify this calculation1180. Basically, the difference
between a qualifying CCP and other CCPs seems to be prudential
regulation in accordance with international standards.

A clearing member is a member of, or a direct participant in, a CCP that

is entitled to enter into transactions with the CCP, regardless of whether
it acts as a financial intermediary for its own clients or on its own account
(e.g. proprietary trading). This definition seems to require clearing
members to trade on their own behalf1181. Although this may be common,
the defining characteristic of a clearing member is normally to be able
to enter into transactions directly with the CCP, whether through a
novation of previous bilateral contracts, or by other means.

Where a CCP has links to a second CCP, the second CCP is treated as a
clearing member of the first CCP. Whether the second CCP’s margin
contribution to the first CCP is treated as initial margin or a payment into
the default fund depends on the precise legal arrangements. In case of
doubt, national supervisors should be consulted to determine the correct
treatment1182.

A client is a party to a transaction with a CCP through either a clearing

member acting as a financial intermediary, or a clearing member
guaranteeing the performance of the client to the CCP 1183. This would not
seem to reflect how clearing takes place in many cases. Generally,
clearing members enter into a transaction as principal with the CCP even
if it is a client trade (in which case there will be a back-to-back
transaction between the clearing member and the CCP. The reason is to
ensure the mutuality of all transactions entered into by clearing members
with the CCP to permit legally enforceable netting in a default scenario.
Other structures may be possible under national insolvency law, though
it seems odd for the Basel Committee to require a direct contractual

 Back to contents 180

 COUNTERPARTY CREDIT RISK

nexus between the client and the CCP when in practice this may not be
possible.

A multi-level structure is one where a bank can centrally clear as an

indirect client. This refers to the provision of clearing services to a bank
that is itself a client and not a clearing member or other clearing
client1184.

Initial margin is defined as a clearing member’s or client’s funded

collateral posted to the CCP to mitigate the potential future exposure of
the CCP to the clearing member arising from possible future changes in
the value of transactions1185. In practice, initial margin may be a fairly
standard minimum margin contribution based on the clearing member’s
status and projected trading volume. The definition of initial margin
expressly excluded contributions in respect of mutualised loss sharing
arrangements (i.e. where a clearing member uses initial margin to
mutualise losses following a default by a clearing member) 1186. This
practice, which used to be common, results in the margin being treated
as a default fund exposure1187. Initial margin also includes margin
deposited in excess of the minimum amount required provided the CCP
or (in the case of client margin) the clearing member may prevent
withdrawal1188.

Variation margin means a clearing member’s or client’s funded collateral

posted on a daily or intra-day basis to a CCP based on price movements
in cleared transactions1189. Whilst daily or intra-day re-margining is
certainly common, it is unclear to us whether it should be mandatory.
For example, there may be materiality thresholds.

Trade exposures include the current and potential future exposure of a

clearing member or a client to a CCP arising from OTC or exchange-traded
derivatives, securities financing transactions, or initial margin
contributions1190. Potential future exposure measures the risk that any
given exposure may increase in size prior to the counterparty’s default.

The default fund consists of clearing members’ funded or unfunded

contributions to, or underwriting of, a CCP’s mutualised loss sharing
arrangements. Basel III clarifies that the actual description used by a
CCP is not determinative, but the substance of the arrangements1191. In
practice, almost all CCPs require such contributions to ensure that should
a clearing member or members default the loss is mutualised amongst
those making use of the CCP’s services. Historically, some CCPs had a
separate designated default fund, while others had a contractual right to
resort to all margin provided by clearing members. The Basel standard
applies regardless of local practices or legal frameworks.

Capital requirements

10.7.4 Banks must maintain adequate capital for their exposures to CCPs, and must
consider if they need to hold capital in excess of the minimum set out in this
section. This may be necessary if, for example, a bank:

has exposures to a CCP that give rise to more risky exposures;

 Back to contents 181

 COUNTERPARTY CREDIT RISK

is unclear if the CCP is a qualifying CCP or not; or

an external body (Basel III refers to the IMF Financial Sector Assessment
Programme) has found material shortcomings in a CCP which have not
been publicly addressed1192.

10.7.5 For banks that are clearing members they must assess through appropriate
scenario analysis and stress testing whether the level of capital required to
be held adequately addresses the inherent risks of those transactions. Such
assessments must include potential future or contingent exposures by the CCP
from default fund contributions and/or any commitments to take over or
replace offsetting transactions from other clearing members should they
default1193. This seems understandable, but one may wonder why such risks
are not adequately taken into account under the capital adequacy framework.
There are certain monitoring and internal reporting requirements 1194.

10.7.6 The capital requirements distinguish between qualifying and non-qualifying

CCPs. The definition of a “qualifying” CCP has already been set out.

Non-qualifying CCP exposures

10.7.7 The capital charge differs between trading exposures and contributions to the
default fund of the CCP.

10.7.8 Trade exposures are risk-weighted using the standardised approach to credit
risk based on the appropriate risk weight to each counterparty 1195. The IRB
approach cannot be used.

10.7.9 Default fund contributions are risk-weighted on a blanket 1250% basis

(equivalent to a deduction from capital) 1196. The default fund contribution,
for this purpose, includes both funded and unfunded (i.e. agreed but not paid)
contributions. In the case of unfunded commitments the bank’s national
supervisor is to determine, through the Pillar 2 process, the amount to which
the 1250% risk weight applies. This is curious as an unfunded commitment
will be determined in accordance with the CCP’s internal rules. If a CCP can
itself determine the amount of margin on a discretionary basis then the
amount of the deduction, logically, should be whatever amount is set by the
CCP from time to time.

Qualifying CCPs

10.7.10 The rules are considerably more complex.

10.7.11 The rules distinguish between trade exposures, client exposures, posted
collateral and default fund contributions. We will take each of these in turn.

10.7.12 Only clearing members will normally have an exposure to the CCP’s default as
non-clearing members CCR exposure will be on their exposure to the clearing
member.

 Back to contents 182

 COUNTERPARTY CREDIT RISK

Trade exposures

10.7.13 If a bank is a clearing member and enters into trades for its own account then
the CCR capital charge is 2%1197.

10.7.14 If the clearing member additionally offers clearing services to clients then a
2% risk weight applies based on the amount the clearing member is obligated
to reimburse to its client(s)1198. This seems ambiguous as it is not clear if
“clearing services to clients” means client clearing, transactions entered into
as principal by the clearing member but on behalf of clients (so there is a
back-to-back transaction with the customer) or both. Posted-collateral is
subject to the treatment detailed below.

10.7.15 The exposure amount to which this 2% risk weight applies is calculated under
the normal CCR framework1199. Provided the netting set does not contain any
exotic derivatives or illiquid collateral, and there are no disputed trades, the
margin period of risk may be under 20 days 1200, subject to a floor of 10 days 1201.

Client exposures

10.7.16 These rules apply to banks’ exposures to their clients in respect of centrally
cleared transactions. Specifically, the rules apply to the following
transactions:

a bank’s exposures to a clearing member if it is a client and the clearing

member acts as a financial intermediary (in this case the Basel III text
makes clear that a back-to-back transaction is covered)1202;

a bank’s exposures where the bank is a client of a clearing member and

the clearing member guarantees performance of the transaction1203; and

exposures of lower-level clients to higher-level clients in a multi-level

client structure if certain criteria are met 1204.

10.7.17 This covers: traditional back-to-back contracts between the clearing member
and its client bank, (ii) client clearing; and (iii) so-called “multi-level
structures” (see above), insofar as different.

10.7.18 If the client bank is exposed to losses if its clearing member and another client
of the clearing member both default the risk weight is 4% 1205. If not, the
transaction is treated as not centrally cleared and is instead subject to the
CCR treatment under the framework applicable for un-cleared transactions1206.

Posted collateral

10.7.19 Any posted collateral is subject to the credit risk treatment or the market risk
treatment that would apply under the credit risk framework for banking book
transactions (standardised or IRB) and the market risk framework for trading
book transactions1207. In other words, the bank is treated under both the credit
risk and market risk framework as still being exposed to the posted collateral,
which is logical as if the collateral becomes worthless the bank will face a
direct credit risk or market risk loss (as the collateral would need to be
replaced).

 Back to contents 183

 COUNTERPARTY CREDIT RISK

10.7.20 However, as there may be CCR losses in addition, this attracts a capital
charge, both in the trading and banking books 1208. There will be a risk of CCR
losses in all cases where the collateral is not held in a bankruptcy remote SPV,
but is, say, transferred to the CCP or to a custodian on its behalf 1209.

10.7.21 The following risk weights apply:

banks that are clearing members apply a risk weight of 2% for client
exposures;

banks that are clients of a clearing member and are not protected from
losses in the case that the clearing member and another client of the
clearing member default apply a risk weight of 4% to their exposures; and

banks that are clients of a clearing member that are not so exposed apply
a risk weight of 2%1210.

10.7.22 Collateral held by a bankruptcy remote custodian for a CCP attracts no capital
charge as the bank posting the collateral will not suffer loss if the CCP fails 1211.

Default fund exposures

10.7.23 The capital charge for default fund exposures depends on whether the CCP
segregates default fund contributions by product or exposure class. If the CCP
does do this (meaning, say, that default fund contributions for SFTs cannot be
used to satisfy losses on OTC derivatives) the formula and methodology
described below applies on a product-class by product-class basis1212. If not,
then no apportionment to separate asset classes can be made1213. Exposures
to the default fund attract the approach set out in the following
paragraphs1214.

10.7.24 The capital charge for default fund contributions depends on a risk sensitive
formula that considers: (i) the size and quality of a qualifying CCP’s financial
resources; (ii) the CCR exposures of the CCP; and (iii) the application of the
CCP’s financing resources under the applicable (contractual or statutory)
waterfall in the event of the default of one or more clearing members 1215.

10.7.25 It is not necessary for the calculations described below to be performed by

the bank if another party, such as the CCP, a regulator or other person with
access to the relevant data is able to do so1216. However, to ensure the capital
calculation is reliable minimum criteria must be met1217.

10.7.26 The bank clearing member’s capital contribution is calculated in two steps:
the hypothetical capital requirement of the CCP based on (i) the CCR
exposures of the CCP to all of its clearing members and clients; and (ii) the
capital requirement for the clearing member bank1218.

10.7.27 We will now describe these two calculations.

 Back to contents 184

 COUNTERPARTY CREDIT RISK

Hypothetical capital requirement of the CCP

10.7.28 It should be understood that this calculation has nothing to do with the actual
capital adequacy requirement (if any) imposed on a CCP by its own regulator.
It is a purely hypothetical calculation.

10.7.29 The first step in this process is to calculate the wholly notional capital
requirement for the CCP (KCCP). This calculation is based on the CCP’s CCR
exposures to its clearing members and clients1219. The sole purpose of this
calculation is to determine the capitalisation of the CCP’s default fund
contributions, and does not represent the CCP’s actual capital requirement
set by its supervisor1220.

10.7.30 In calculating KCCP the following factors apply:1221

the risk weight = 20%;

the capital ratio = 8%;

CM means the clearing member of the CCP; and

EADi = the exposure amount of the CCP to clearing member i1222.

10.7.31 The formula is calculated as follows:

𝐾CCP = ΣCMi 𝐸ADi * 𝑅𝑊 * 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑟𝑎𝑡𝑖𝑜 1223

10.7.32 The calculation of the notional capital requirement for the CCP is therefore
equal to the sum of all exposures of each clearing member to the CCP
multiplied by 20% multiplied by 8%. National supervisors may increase the
20% figure1224.

10.7.33 There are detailed rules governing the specific calculation.

Capital requirement for clearing members

10.7.34 This is the second part of the capital calculation. A complex equation applies
which is set out below. The basis of this calculation is the amount of the CCP’s
own resources that it contributes towards the funding of defaults by clearing
members. The idea is that the amount of the loss attributable to defaults by
clearing members will be reduced to the extent that the CCP agrees to fund
this loss by its own injection of capital in a default scenario. This will be
described formally below.

𝐾CMI= the capital requirement on the default fund contribution of clearing

member i;

DFCMpref = the total prefunded default fund contributions by clearing

members;

DFCCP = the CCP’s prefunded own resources contributed to the default

waterfall if pari passu or junior to funded clearing member contributions;
and

 Back to contents 185

 COUNTERPARTY CREDIT RISK

DFipref = the prefunded default fund contributions provided by clearing

member i1225.

10.7.35 Based on the foregoing factors the notional capital charge for each clearing
member I is calculated as follows:

𝐷𝐹ipref
𝐾CMi = max (𝐾CCP* ( ) ; 8% * 2% * 𝐷𝐹ipref) 1226
𝐷𝐹CCP+𝐷𝐹CMpref

10.7.36 The formula places a minimum risk weight under this calculation of 2% 1227.

Capital charge

10.7.37 The consolidated Basel III text does not actually specify how the capital charge
is to be calculated, which seems to be a mistake. It is therefore unclear if
the capital charge is equal to the greater of the two figures calculated above
or is their sum. The Basel III text does not provide any guidance on this matter,
although approaching the matter from first principles, it would seem that only
the maximum of the two amounts should be taken into account as
representing the greatest loss that the bank could suffer should both the CCP
and its clearing members all default.

Floor

10.7.38 There is a floor for both trade exposures and default fund exposures equal to
the capital charge had those same exposures been to a non-qualifying CCP. In
this case the capital charge for a non-qualifying CCP applies (see above)1228.

 Back to contents 186

 MARKET RISK

11. MARKET RISK

11.1 Introduction

11.1.1 The financial crisis was centred on losses made possible from a regulatory
perspective by the 1996 Market Risk Amendment, that was retained, basically
unamended, by the Basel II framework, and permitted banks to use their own
value-at-risk (VaR) models to determine capital requirements. Most failures
of financial institutions in the crisis were attributable to losses incurred in
banks’ “trading book” for market risks arising out of their trading book
positions (e.g. proprietary trading). Unsurprisingly, in the development of
Basel III searching attention has been given to what aspects of the 1996
standard went wrong. In fact, most of the criticisms of the standard were
well known before the crisis in the relevant academic literature, but were
dismissed as unlikely to be relevant in practice. The financial crisis disproved
this approach. Unsurprisingly, the Basel Committee has adopted a new
approach, that while theoretically superior as a measure of risk, suffers from
possibly greater downsides than the former standard, in terms of data
limitations and model risk.

11.1.2 The new models-based approach is supplemented by a new “standardised

approach”, that unlike its predecessor, is based extensively on banks’ internal
pricing models. Because of its complexity, and, some could say its
impenetrability to those not familiar with the mathematics of quantitative
risk management, this led, in a late amendment, to the 2019 decision that
allowed the continued use of an amended form of the 1996 standardised
approach, scaled up to reflect perceived deficiencies in the former in terms
of risk measurement. Supervisory approval is required for both the new
models-based approach and the simplified standardised approach.

11.1.3 Given the intention, and clear capital incentives, to adopt the revised internal
models approach this will be described first. We follow with the revised
standardised approach and then a description of the simplified standardised
approach, which is not intended to be available to most internationally active
banks, but reflects the Committee’s acceptance that the Basel standards are
applied more widely (especially in the European Union) to other banks.
Discussion of the models-based approach first is also justified by the intention
of the Committee that the standardised approach should offer a suitable fall-
back for banks that fail, in whole or in part, to satisfy the new models-based
approach, and is expressly intended to be compatible with it, unlike the 1996
standardised approach. Both approaches are heavily (perhaps too heavily)
based on theories of quantitative risk management making a direct read of
the Basel III text difficult to those without such a background, as well as
acquaintance with risk models. Whether this is desirable given the need for
senior management to understand risk and risk models only time will tell.

11.1.4 This lengthy chapter is organised as follows. We first outline the 1996 internal
models-based approach (the Market Risk Amendment). This is followed by a
description of the Basel Committee’s criticisms of the Market Risk Amendment

 Back to contents 187

 MARKET RISK

in the light of the financial crisis. We consider, in brief, the relevant academic
literature, including a 2011 publication (sadly, not updated since) by the Basel
Committee, justifying a move from VaR to expected shortfall (ES) as a
measure of risk. The new ES standard is then described followed by the new
standardised approach and finally the simplified standardised approach. The
difference in format in this chapter (which includes extensive discussion of
the Basel consultative documents leading up to the new standard published in
2019) from earlier chapters is justified by the role market risk losses played
in the financial crisis, as well as the scale of the changes made from the
former regime. The degree of improvement actually made will only be
revealed when the next financial crisis comes around, which is likely to take
a different path to the last. The revisions to the market risk framework were
originally intended to come into force much earlier, but were then pushed
back, initially due to deficiencies in the proposed standard and now to January
2023 (due to the Covid-19 pandemic), 15 years after the outbreak of the
financial crisis. The UK and the EU have announced that they will defer
implementation to 2025.

11.1.5 Readers not interested in the origins of and limitations to VaR as a risk
measure can skip the following sections and proceed directly to discussion of
the Basel III standard.

11.2 The 1996 Market Risk Amendment

11.2.1 The discussion that follows will focus on the internal models-based approach
used by most sophisticated banks to calculate capital charges for market risk.
Basel I had been concerned with credit risk, being then the predominant risk
applicable to banks in 1988. However, subsequent developments in the
financial markets made it necessary to take account of the risks presented by
banks’ trading activities, especially in then new financial products such as
derivatives1229. Basel I had adopted a very rough and ready set of rules but by
the 1990s it seemed clear that a more sophisticated approach was called for.
The choice made was, essentially, to rely on the concept of “value at risk”
(VaR). This had been developed as a concept in the early 1990s and made
popular as cutting edge technology in risk management with the publication
by J.P Morgan of its Risk Metrics document in 1994 (subsequently updated), a
simplified version of the bank’s own internal capital model.

11.2.2 The basic idea behind VaR is very simple and appealing (and still remains so
after the financial crisis). VaR determines the maximum loss a bank can expect
to suffer based on a given “confidence level”. The confidence level sets the
number of occasions in which the maximum postulated loss may be expected
to be exceeded over a specified period (which can be one day or any longer
period the data set available to the bank can reasonably allow). If the
confidence level is 95% and the period is one day then in 5 days out of 100 the
loss may be greater than that postulated by the model. If, as the Basel
Committee decided, the confidence level is 99% then the maximum loss will
only be exceeded in one time period out of a hundred. The Committee opted
for a 10 day “holding period”, implying that trading positions could be
liquidated or hedged over 10 trading days (two weeks). A 99% confidence level
with a 10 day holding period implies that the maximum loss would be
exceeded about once in every four years, assuming a normal distribution of
losses (of which more later). As this was not considered to be sufficiently

 Back to contents 188

 MARKET RISK

prudent, national supervisors were required to set a “scaling factor”,

essentially a multiplication factor, of three or more. Assuming a normal – or
Gaussian – distribution of losses this seemed a very conservative capital
measurement of potential losses, and if losses were in fact normally
distributed in a crisis, meaning clustering around the mean, with no “fat tails”
i.e. low risk but high impact events, it would have been.

11.2.3 It should be understood that, as originally developed by J.P Morgan, VaR was
intended to be a day-to-day risk metric applied to the bank’s trading positions
with a one day horizon. As designed, it was, and, in normal times remains, a
good risk measure (which does not mean there are not better ones 1230).
Financial markets generally vary relatively little from day to day, so the
likelihood of an extreme loss on any given day is low. (This is not to ignore
periodic crises, of which the prime one before the Market Risk Amendment
was adopted was the 1987 Stock Market crash, the causes of which remain
debated1231). Hence, the assumption of normal distribution works well for a
day-to-day measurement of risk.

11.2.4 The problem arises when VaR is used as a measure for regulatory capital
requirements. Inevitably, regulators are more concerned about unlikely but
severe events that may pose a serious risk to the soundness of individual
institutions, or to the financial system as a whole (systemic risk). Thus the
95% confidence level was raised to 99% and the holding period – simulating
the longest period for a bank to exit its holdings of tradable securities and
derivatives – extended to 10 trading days. Hence, also, the scaling factor,
which was set on a case-by-case basis by individual banking supervisors,
contributing to widely varying VaR figures for the same asset held by different
banks. All these decisions by the Basel Committee can be justified by a focus
on real risks, and by a desire to ensure the banking sector was properly
capitalised. Unfortunately the result was opacity, a non-level playing field,
and when the financial crisis hit, a gross underestimation of actual bank
solvency requirements, essentially caused by the interaction of three factors
not taken into account expressly in the VaR model: a non-normal distribution
of loss events, liquidity risk and the impact in a crisis of mark-to-market (or
available for sale) accounting.

11.2.5 Capital adequacy has traditionally been about solvency based on real losses,
as opposed to a catastrophic collapse in market liquidly, which traditionally
was seen as capable of being capable of being addressed through the
monetary policy of the central bank. Mark-to-market (or fair value)
accounting also makes sense for assets held for short term trading purposes
to profit from market price movements. In such cases, historic cost
accounting (the main alternative) would seem irrelevant. However, a collapse
in market liquidity combined with the effect of re-pricing based on market
prices, or where unavailable, models, in a crisis produced a toxic mix, which
when combined with other factors, triggered the 2008 financial crisis and
ensuing global recession.

11.2.6 When applied to short-term risk management, VaR is effective. It also has the
advantage of quantifying risk (within the selected confidence level) to a
simple numerical figure that, conceptually, can be extended from market risk
to other risks, such as operational risk and credit risk. This was the hope of
its advocates, and when Basel II introduced the internal ratings-based

 Back to contents 189

 MARKET RISK

approach, this was widely seen as a precursor to adoption of credit risk models
similar to VaR models for all regulatory capital requirements.

11.2.7 The Market Risk Amendment was deliberately not specific as to the specifics
of the models allowed by banks to be used. Instead, it set both quantitative
and qualitative criteria to be met, and relied on banks persuading their
national supervisor that the proposed model was both conceptually sound and
based on correct data. This should not necessarily be seen as a failing as the
literature on VaR models shows that there are many ways to build such a
model and that there is no a priori reason to prioritise one type of model over
another, as conceptual advantages need to be weighed against data
limitations, and possible model error. This reflects the common sense view
that it is better to be approximately right than precisely wrong.

11.2.8 One common approach to VAR is to rely on historical data, weighted, if

necessary, to give greater prominence to more recent – and presumably more
relevant – data. More sophisticated parametric models relying on copulas are
mathematically more likely to generate correct calculations of VaR.
Unfortunately, they are also more prone to model error if the copulas are
incorrectly calculated, and in the case of non-elliptical distributions, likely to
lead to incorrect results unless such distributions are known in advance and
the model is correctly calibrated to address them. Perfect information, often
assumed by economic theory, would solve this, but unfortunately, as FA Hayek
demonstrated1232 such a state of affairs never has never existed, and models
that make this assumption are practically useless.

11.2.9 Given data limitations, as well as lack of knowledge as to how market prices
would perform in a crisis (especially, correlations), many VaR models, that
were approved by regulators, relied on simplistic and, in the event, incorrect
assumptions, such as normal distribution and Brownian (i.e. random) shifts
that failed to take account of the fact that in the case of many financial
instruments (especially, derivatives) there are “fat tails” i.e. low probability
but high loss events that are far more likely than a normal distribution would
allow.

11.2.10 Further, many financial instruments feature clustering effects, and in a crisis
hitherto established correlations break down, liquidity vanishes, and market
prices triggered by “fire sales” of assets in a thin market will, through mark-
to-market accounting trigger exponential losses at other institutions holding
the same or similar assets, triggering further fire-sales at lower prices. Where
liquidity disappears, firms are required to mark-to-model, with all the
uncertainties, and possibly, incentives to under estimate prices where no
market actually exists. In 2007-2008 this was not solely an issue for well-
known illiquid assets, such as super-senior and mezzanine tranches of re-
securitisations of sub-prime mortgages, but of assets like commercial paper
the liquidity of which simply vanished. The VaR numbers, translated through
mark-to-market accounting in some cases (not all) grossly undervalued many
of the assets held in banks’ “available for sale” portfolio, as a point in time
market price did not factor in possible (and, in many cases, real) recovery in
values after the crisis. A case in point is the administration of Lehman
Brothers’ UK subsidiary where the holders of regulatory capital instruments
are still litigating over the distribution of the approximately £5 billion surplus
after all customers and senior creditors have been repaid1233. The Lehman

 Back to contents 190

 MARKET RISK

subsidiary was certainly cash flow insolvent once its parent company entered
liquidation, but accounting rules that require the liquidation of a company
that turns out to have billions of surplus assets after more than a decade of
protracted insolvency proceedings and extensive litigation may be open to
question.

11.2.11 A detailed description of the 1996 Market Risk amendment will not be
provided, as the purpose of this publication is to describe the new framework
that is scheduled to replace it in 2023.

11.3 The Concept of Market Risk

11.3.1 Market risks are, basically, the risks of a bank suffering losses (or making
profits) as a result of changes in the value of financial instruments, exchange
rates or commodity prices. The market risk framework only applies to
financial instruments that are held for trading purposes where the bank is
exposed to risk from changes in such prices, as well as FX and commodity risk
in the banking book (as there is no banking book treatment). The most obvious
example is proprietary trading in equities, bonds, and derivatives. If a bank
purchases an equity stake as a long-term investment, as is common in
countries like Germany and Japan, then the credit risk framework applies.
The same applies where a bank makes a loan, or buys a bond for long-term
investment purposes, or enters into a securities financing transaction (like a
repo) for similar purposes.

11.3.2 In reality the definitions are stricter than that, and have been significantly
tightened by Basel III, and will be considered below when considering the
definition of the trading book. However, the basic reason for the different
treatment is that in long-term transactions the main risk a bank is exposed to
is credit risk, as the bank will only make a loss if the obligor defaults, the
bank makes a provision, or accounts for an impairment to the value of the
position, under its applicable accounting framework. A bank will not normally
account for a loss if the daily share price changes in a company that it holds
as a strategic investment, as such price movements are, in almost all cases,
irrelevant.

11.3.3 Where, however, a bank actively trades equities, bonds or derivatives for its
own account, then it will suffer a loss every time the price goes down and
make a profit if it goes up. This does not mean that a trading position cannot
also generate credit risk too. There is a risk of loss on bonds or equities of a
sudden insolvency event happening while the asset is held on the bank’s
balance sheet. This risk must also be taken into account in the market risk
capital charge (the term used is default capital risk or DRC). Credit risk on
derivative and certain other positions with positive net value is taken into
account through the capital charge for counterparty credit risk which has
already been described in chapter 10.

11.3.4 Traditionally (and this remains the case under Basel III) the market value on
positions held for trading purposes (i.e. the bank’s trading book) is calculated
daily based on the market value. This is also required by accounting
standards. Profits and losses on trading may generate an effect on available
capital through changes reflected in the profit and loss account, although the
circumstances where profits can be treated as capital are described in the

 Back to contents 191

 MARKET RISK

chapter on the definition of capital. Losses must, in principle, always be

deducted from core Tier 1 capital. The purpose of the market risk framework
is then to take account not of actual losses or expected losses (which are
addressed through the P&L account and the bank’s policy on provisioning) but
the potential for future losses as a result of falls in market values 1234. Thus
capital is only required for financial instruments with positive market value.

11.3.5 The same applies, in principle, to derivatives, although account must be taken
of out-of-the money derivatives that may generate future losses from changes
in market prices (as opposed to counterparty default). Thus an out-of-the
money sold option may result in large losses if the market price of the
underlying falls, making the option valuable to the counterparty and
generating a loss when exercised. This differs from counterparty credit risk
as the loss is wholly attributable to changes in the market value of the
underlying. Equally, an in the money purchased option may lead to losses if
the price falls before the expiry date making the option worthless. We have
assumed for simplicity that the options in both cases are only exercisable on
expiry, i.e. they are European options. Options exercisable at any time prior
to expiry (American options) are more complicated, but the principle is the
same, as if the value of a sold option increases there will be losses that will
only be fully crystallised on exercise. Options that expire with the option
being out of the money will generate no further loss as the option will simply
not be exercised, which is why the maximum loss on a purchased option is
limited to the premium paid, whereas on a sold option it is theoretically
infinite.

11.3.6 Forward transactions, on the other hand, commit the parties to settle
(whether in cash or in kind) at the pre-agreed market price. Swaps may
generate daily profits or losses as the two assets being swapped e.g. a fixed
for a floating interest rate may change, or as in an equity swap with the value
of the equity. Credit default swaps (CDS) are different, and they are
functionally equivalent to a sold option where the profit is capped at the
premium paid by the counterparty, and the loss is capped at the amount
payable on the default of the reference obligation, which, in the case of CDS
documented under current standard ISDA documentation, will be calculated
following an auction process after a determination by the ISDA Determinations
Committee that a credit event has occurred. Clearly other possibilities exist
and parties are free to contract on bespoke terms that do not reference the
ISDA documentation, or make changes to it.

11.3.7 It will be seen from the above, which simplifies greatly, that the process of
calculating the appropriate capital charge for unexpected losses is extremely
complex, as the framework has to cope not just with instruments whose losses
may be relatively simple to calculate, such as equities and bonds, where you
normally just look at the market price, or forwards on such instruments, which
behave the same way, but also instruments the value of which exhibit
extensive “non linearities” i.e. derivatives whose value is not directly
correlated to the price of the underlying. Options provide a good example,
as a small change in the value of the underlying may trigger a much bigger
change in the market value of the option.

11.3.8 The value of positions in foreign currencies and commodities held for trading
purposes will move in accordance with daily market prices. Derivatives on

 Back to contents 192

 MARKET RISK

such instruments will exhibit the same characteristics as other derivatives,

such as on bonds or equities.

The development of VaR

11.3.9 Over the years traders and risk managers developed various techniques for
valuing, and trying to limit losses that trading activities could give rise to. It
is not proposed to provide a history of the evolution of risk management, and
the techniques that have been used, as this is set out in books on risk
management as well as monographs dealing with pricing and trading specific
instruments. Instead we will proceed directly to VaR. The idea behind VaR is
very simple as it claims to reduce the maximum overall loss on a bank’s
trading book to a simple number in the bank’s reporting currency across its
whole portfolio. The concept was developed by J.P. Morgan, and published in
1994. A revised and refined version was released in 1996. The concept sought
to measure the highest likely loss across a bank’s trading portfolio over the
next day. This reveals the two basic components of all VaR models:

a time horizon over which the losses are to be estimated. RiskMetrics,

and many internal economic capital models, look at a one trading day
period; and

a percentage below 100% setting the probability of the maximum loss

being exceeded.

11.3.10 The first is determined by the length of time over which the bank is concerned
about the maximum likely loss materialising. The appropriate length of this
period, which is essentially arbitrary, should be determined by reference to
the purpose we are using the VaR model for. From a day-to-day risk-
management perspective, or when setting daily position limits for traders, a
one day period may be adequate. For more long-term decisions on portfolio
strategy it is likely to be inadequate. In particular, if used as a measure of
the maximum loss likely to be suffered if the bank wants to liquidate or
completely hedge its entire trading portfolio at other than “fire sale” prices,
a one day period could be a very poor indicator, especially if the positions are
not perfectly liquid and the bank’s positions are large.

11.3.11 The period of time over which one looks is referred to in VaR models as the
“holding period”.

11.3.12 The second factor used by all VaR models is referred to as the “confidence
level” as it specifies the degree of certainty that the maximum likely loss will
not be exceeded. A low percentage (say, 50%) would be useless as a risk
management tool as losses would exceed the VaR on 50% of all trading days.
Originally, most banks using VaR models for internal risk management purposes
chose a confidence level of 95% which means the likely maximum loss will only
be exceeded in five days out of one hundred trading days if a one day holding
period is taken. If, on the other hand a 99.9% confidence level is taken, with
the same holding period, then the likely maximum loss will only be exceeded
in 1 out of 1000 trading days, which works out to approximately once in four
years. In theory, the higher the confidence level the better the model is at
predicting the maximum likely loss, as there will be fewer exceptions.
However, there is a significant caveat. This is only true if the data set is big

 Back to contents 193

 MARKET RISK

enough and relevant, and the assumptions behind the model, such as
correlations between movements in asset values, are in all material respects
correct. If not, very high confidence levels are likely to produce results that
are progressively more and more inaccurate, as the data may not exist (and
for very high confidence levels certainly will not) and therefore the figures
will have to be extrapolated from a much smaller data set based on models.
In practice, there is a trade-off between the set confidence level and how
likely in reality it will be accurate.

11.3.13 The choice between increasing the confidence level, the holding period, or
both, to improve the measurement of risk is essentially a pragmatic one based
on data limitations as well as the purpose which the model is intended to
serve.

11.3.14 The Basel Committee chose a 99% confidence level, meaning that the
maximum expected loss should only be exceeded 1% of the time over a ten
day holding period. The choice of a ten day holding period was based on the
assumption that a bank will be able to close-out or fully hedge the portfolio
within ten trading days. As the instruments held in banks’ trading books in
1996 consisted of mainly equities, fixed-income securities and relatively
simple derivatives this does not seem unreasonable.

11.3.15 Then, to instil further conservatism in the VaR figures, national supervisors
were required to apply a scaling factor of between three and four. Without
the scaling factor, the VaR could lead to a loss exceeding the capital
requirement about once in every four years 1235, which was not thought
sufficiently prudent. Assuming a scaling factor of four and a normal
distribution such an event would happen less than once in the period since
the Big Bang1236.

11.3.16 The Committee can be excused from not anticipating market developments
up to the crisis eleven years later. Credit default swaps had been pioneered
by J.P. Morgan in 1994 and the first synthetic CDO followed in 1997, also
created by J.P Morgan, under the title of Broad Index Secured Trust Offering
(BISTRO). Needless to say BISTROs referenced underlying commercial loans,
bonds and municipal bonds, the risk characteristics of which were fairly well
known, backed by decades of data, and investors were informed of the
underlying pool of obligors. Also, the transaction was designed for J.P Morgan
to obtain regulatory capital relief as opposed to actively trade credit risk. It
was therefore an innovative credit risk hedging strategy.

11.3.17 The subsequent metamorphosis of BISTROs into sub-prime CDOs and even
more complex and less understandable instruments such as CDOs of CDOs
(CDOs squared, cubed, etc.) was perhaps an inevitable consequence of
financial ingenuity and a search for apparently safe, but higher yielding
securities at a time of historic low interest rates. Arbitraging capital charges
between the banking book and the trading book by booking highly complex,
poorly understood and untested financial instruments in the trading book
when little actual trading took place was an unintended consequence. In
practice, in the absence of market prices, positions were (as permitted)
marked to model.

 Back to contents 194

 MARKET RISK

Limitations of VaR models

11.3.18 It needs to be stressed that VaR was never designed as a means for banks to
assess their regulatory requirements, as opposed to being a guide for banks
when deciding how internally to manage market risk on their trading portfolio.
The limitations of VaR as a regulatory measure of risk in the academic
literature are summarised below, as well as described in the Basel
Committee’s publications leading up to the adoption of the final Basel III
market risk standard. However, perhaps its greatest limitation as a regulatory
measure of risk is that the confidence level only told you the most likely
maximum loss (based on the given data set used by the bank) at the specified
confidence level. What VaR cannot tell you is the likely loss in those cases
where the VAR is exceeded (1% of cases under the 1996 Market Risk
Amendment. The loss may be small or it may be sufficient to bankrupt the
institution.

11.3.19 Simply put, VaR provides absolutely no information about states of the world
outside the confidence level1237. And simply increasing the confidence level is
not a solution either because of the inevitable data limitations. For example,
had the Basel Committee set a 99.9% confidence level with a ten day holding
period, assuming a Gaussian – or normal – distribution (with no scaling factor),
implies a risk of losses exceeding that based on loss data that will occur
perhaps once in 1000 years. No bank has ever had such a data set. Monti dei
Paschi di Sienna is the oldest bank still existing in the world, and was founded
as a pawn agency in 1472. However, even though it has records from its early
activities, it would patently be absurd to use statistics taken from fifteenth
century Sienna to price losses on shares or bonds in Italy today. Such data (if
it exists) is simply irrelevant to the risks in trading financial instruments today.

11.3.20 As mentioned above, the Basel Committee required a ten day holding period.
To meet this directly a bank would need 3000 trading days of data to directly
estimate the regulatory VaR1238. Assuming 250 trading days per year this is a
12 year data set for all instruments included within the portfolio being
modelled measured on a daily basis. Apart from the fact that very few
institutions have such data sets for all instruments currently held in their
trading book (some did not exist 12 years ago), it is questionable whether such
old data would really tell us anything about the likely loss over the next ten
days. Giving equal weight to all the data would result in the VaR being
unresponsive to recent shifts in market volatilities or prices, as the new data
would be overwhelmed by very old data resulting in an almost flat VaR, which
is not a good risk measurement, whether for regulatory or other purposes. Of
course, one can discount the weight of old data, but this may not provide any
clear guide to what is happening in the market now, given the presence of
probably irrelevant old data.

11.3.21 Where the data do not exist then it is necessary to imply such a data set from
a more limited set of relevant data. While there are valid mathematical
techniques for doing so, the higher the confidence level, the more the VaR
will be dependent on whether extrapolations from such limited data are
representative for the presumed distribution in all states of the world up to
the chosen confidence level. Unless we know how representative the data
used are, the mathematically generated figure is likely to tell us little about
what would really happen in a crisis1239.

 Back to contents 195

 MARKET RISK

11.3.22 The solution to this problem adopted by the Basel Committee is the so-called
square-root-of-time. This relies on a one day VaR multiplied by the square
root of ten to estimate the ten day figure. This is simple and easy for banks
as they calculate their VaR on a daily basis anyway. However, this rule only
holds if it is assumed: firstly, all losses are normally distributed; secondly,
volatility is independent over time; and thirdly, volatility is identical across
all time periods1240. Danielsson argues that “all three assumptions are
violated”1241. What happened in the financial crisis suggests he is correct.

11.3.23 The basic problem with using VaR as a regulatory measure to ensure financial
soundness and stability is not with the mathematical models, or their
correctness, which can be proven to be either valid or false, and
mathematically false models would never be presented to regulators for
approval. The real issue lies in the assumptions made that play such a large
role in the outcome of the VaR calculation that, unless based on near-perfect
data, and correct assumptions, may be meaningless. Hayek referred, in a
speech in 1978 to the American Enterprise Institute (in another context) to
the “beautiful systems of equations with which we can show in imagination
what would happen if all these data were given to us. But we often forget
that these data are purely fictitious, are not available to any single mind, and,
therefore, do not lead us to an explanation of the process we observe”1242.
Similar points have been made by economists working in a very different
school of economic thought by Kay and King 1243 and Mandelbrot1244.

11.3.24 There are further assumptions made by most VaR models used to calculate
regulatory capital requirements before the financial crisis: the distribution of
losses is known and (usually) assumed to be normal, apparent correlations are
real and do not change over time, price movements are random and do not
exhibit sharp “jumps”, liquidity is constant and the market behaves in a panic
in exactly the same way as in normal times. Of course, none of this is inherent
in the nature of a VaR model. One can model any chosen distribution, assume
price movements are correlated and not random, and factor in panics. Such
a model will be just as sound mathematically as a VaR model that does none,
or only some, of those things. So it comes down to the choices made by the
modeller and the regulator when granting approval.

11.3.25 Unfortunately, assumptions such as normal distribution, etc. proved

unfounded in the financial crisis. The distribution of losses can only be
inferred from the best fit of the available and relevant data. This involves
three judgments: (1) what counts as enough data, (2) what data to exclude as
not relevant, or to be discounted compared to other, more recent or relevant,
data and (3) which model best fits that data. Of course, if all that were known
then there would be only one solution and the model would always be correct.
But where this does not obtain, there may be multiple plausible solutions
based on expert judgment. This is not because of the use of subjective
criteria or biases, but simply that there are multiple ways of explaining the
data, none of which can be proved ex ante to be superior to another.

11.3.26 The problem is exacerbated by the process by which correlations between

different instruments are measured1245. There are many possibilities, including
historical models (which look back as far as the modeller chooses), and
alternatives include so-called “parametric” models, of which the best are
probably those using copulas, which are mathematical correlations assumed

 Back to contents 196

 MARKET RISK

to exist between events and losses. Theoretically, parametric models are

more accurate than historical simulations 1246, but depend extensively on the
assumptions made by the modeller. If those assumptions are wrong then the
model is more likely to give an inaccurate result.

11.3.27 Secondly, market data show that the distribution of losses is rarely normal or
Gaussian, especially in financial crises or panics 1247. Instead, losses tend to
exhibit fat tails, the more so in crises, meaning that high loss low probability
events are factually more common than we might expect them to be. This
can be addressed in a VaR model by presuming a non-normal distribution, but
which of the almost infinite possible distributions between the mathematical
extremes of normal and Cauchy distributions should be chosen, and should
this be the same for all instruments and over all time periods? A Cauchy
distribution is a statistical distribution with no mean or variance, but with a
well-defined mode and median. As it is stable it can be calculated, and has
the property of having much fatter tails, making it suitable for modelling
extreme events.

11.3.28 A VaR model with a 99.9% confidence level, a ten day holding period, a scaling
factor of four and assuming a Cauchy distribution would see the loss in excess
of the bank’s VaR capital requirement occurring not less than once in the
history of the universe, but in less than one year 1248, which is not particularly
conservative if the purpose of regulation is to avoid bank failures or systemic
risk. This shows the importance of being able accurately to choose the correct
distribution, as well as measuring its stability or instability over time. In real
life the true distribution is likely to lie somewhere in between the Gaussian
and Cauchy distributions, and be nearer the Gaussian one in normal market
conditions, but may approach closer to the Cauchy one in a market panic like
the last quarter of 2008. Further, properly calibrated, VaR is generally very
good at estimating losses at the mean, but if we are concerned about the tail
loss it is practically useless as it provides no information at all on likely losses
beyond the predefined confidence level.

11.3.29 As was seen the financial crisis, prices, in stressed situations, often do not
move randomly, but may often exhibit “jumps”, and liquidity, that is plentiful,
as it was in 2005-2006 may disappear as happened in 2008 after the collapse
of Lehman Brothers. Where liquidity evaporates then there may be no market
price, or the market price may be driven by “fire sales” of assets by
institutions (they need not be banks) that have to sell at any price (for
example, funds with mandates only to hold investment grade instruments).
The combination of illiquidity and mark-to-market accounting proved
particularly toxic when institutions required, either by capital ratios or
investment restrictions tied to a credit rating, were forced to dump illiquid
assets, such as sub-prime mortgage re-securitisations, in a market with very
few buyers, resulting in “fire sale” prices that were translated into further
losses based on that market price at institutions marking prices to market, as
accounting standards treated all market prices as the “fair” market price.
This is a perfectly defensible and reasonable assumption in normal
circumstances as historic cost makes no sense when valuing assets held for
trading purposes. However, it can create a doom loop where each price fall
triggers further sales. In a rational market, this process will end when well-
capitalised market participants regard the market as having fallen too low and
start buying bringing the panic to an end. However, if potential investors

 Back to contents 197

 MARKET RISK

don’t have deep pockets, or are incentivised to short term actions, as their
bonus or employment is based on avoiding losses, as opposed to taking a risk
on potentially higher, but uncertain, profits, then this need not happen, which
was the intellectual justification for the original Troubled Assets Relief
Program (TARP) that would stabilise the market by buying up assets for which
there was no real market1249.

11.3.30 Incentives may also play a role in determining risk if traders are not subject
to adequate risk management and monitoring, given the current remuneration
structure based on short-term performance (quarterly, annually), with a
significant part in discretionary bonuses. For example, if a trader takes the
same risks as most other traders then even if everyone makes a loss he is less
likely to be fired than if he pursues a strategy that overall generates higher
profits over the long-term through a few large profits with most days showing
a loss. On the other hand a trader that adopts a strategy known to him to be
likely to generate higher profits consistently, but with a very small risk of
catastrophic losses is from his perspective (if not the bank’s) preferable as he
will then receive higher compensation or bonuses for his perceived superior
performance, while if the loss materialises he will at worst get fired, and if
the likelihood of the loss materialising is small may have moved on. It is
possible that, the extremely risky trading strategy rogue trader Nick Leeson
adopted in taking huge unauthorised positions based on the assumption that
the Nikkei index would continue to trade at a very low level of volatility
(through complex options known as straddle options) would have turned a
significant profit if the Kobe earthquake had not happened, which triggered
sharp movements in the Nikkei, and therefore vast losses on the options. Yet
Leeson had a personal incentive to take the gamble as he had already made –
and hidden through his control of the back office – vast losses so he had an
incentive to take any risk to try to recover his position.

11.3.31 It may be asked what this has to do with VaR? As mentioned above, VaR tells
you nothing about losses beyond the specified confidence level, which the
Basel Committee set at 99%. Therefore, absent effective risk management of
the 1% of cases where losses may exceed the likely maximum, traders may
decide to adopt trading strategies that are more highly profitable 99% of the
time but may result in a catastrophic loss 1% of the time. With a one day
holding period this would probably not be a rational trading strategy (as the
trader might be fired within six months when the loss occurred), but if the
risk is seen as extremely unlikely it might. For example, in August 2007 the
CFO of Goldman Sachs, David Viniar, said “We are seeing things that were 25-
standard deviation moves, several days in a row”1250. How likely is such a loss
occurring? Assuming a normal distribution, an 8 sigma standard deviation
event will happen on any given day less than once in the period of time
corresponding to the age of the universe. A 25 sigma standard deviation will
occur once in 3.057 multiplied by 10 135 years. And Goldman Sachs was
reporting such events occurring several days in a row, which is exponentially
even less likely. It is, of course, possible that Goldman Sachs were just that
unlucky, but it seems improbable. Moreover, changing the distribution to a
non-normal one does not materially change the probability of a 25 standard
deviation move in market prices 1251. Nor was Goldman Sachs alone, as many
other firms were reporting at the same time, and subsequently, massive losses
on subprime mortgage re-securitisation positions.

 Back to contents 198

 MARKET RISK

11.3.32 The more plausible explanation for such price movements is that the models
used were wrong in material respects. However, no trader, or risk manager,
setting and monitoring position limits would be concerned by events believed
to occur three times in 10135 years, and would consider the risk to be zero.

11.3.33 None of the above means that VaR is useless as a risk measurement tool. If
we are concerned about losses around the mean over a short term, it can play
a very useful role in risk management. There may be better ones, but all have
limitations as we shall see. Unless a better risk measurement is identified
then having much knowledge is better than having none, and while intuition
may be useful to a trader acting within defined and enforced position limits,
betting the solvency of the bank on intuition would be reckless.

Academic assessment of VaR as a risk measure

11.3.34 It should be noted that most of the limitations on VaR as a risk management
tool were known and acknowledged in the academic and risk management
literature, including those favourable to the use of VaR, before it was retained
in the Basel II standard in 2004. An early, and outspoken critic, Taleb, wrote
in Derivatives Strategy in 1996 “[t]o me VaR is charlatanism because it tries
to estimate something that is not scientifically possible to estimate, namely
the risks of rare events. It gives people misleading precision that could lead
to the build up of positions by hedgers. It lulls people to sleep. All that
because there are financial stakes involved. To know the VaR you need to
know the probabilities of events. To get the probabilities right you need to
forecast volatilities and correlations. I spent close to a decade and a half
trying to guess volatility, the volatility of volatility, and correlations … [y]ou’re
worse off relying on misleading information than on not having any
information at all. Before VaR, we looked at the positions and understood
them using what I call a non parametric method. After VaR, all we see is
numbers, numbers that depend on strong assumptions”.

11.3.35 In a 1997 article in the same publication Taleb argued “VAR is the alibi bankers
will give shareholders (and the bailing-out taxpayer) to show documented due
diligence and will express that their blow-up came from truly unforeseeable
circumstances and events with low probability – not from taking large risks
they did not understand. … I maintain that the due-diligence VAR toll will
encourage untrained people to take misdirected risk with the shareholder’s,
and ultimately the taxpayer’s, money” (emphasis added).

11.3.36 Jorion, in the same publication, countered that “VAR is an essential

component of sound risk management systems. VAR gives an estimate of the
potential losses due to market risks. In the end, the greatest benefit of VAR
lies in the imposition of a structured methodology for critically thinking about
risk. Institutions that go through the process of computing their VAR are
forced to confront their exposure to financial risks and to set up a proper risk
management function. Thus the process of getting to VAR may be as important
as the number itself”.

11.3.37 Jorion’s book on Value at Risk has so far gone through three editions 1252. In his
conclusion he writes “VAR is no panacea. As we have seen, VAR makes no
attempt to measure the losses beyond the specified limit. Even with a 99
percent confidence interval, unusual events happen, and they sometimes do

 Back to contents 199

 MARKET RISK

so with a vengeance. … While VAR techniques are firmly grounded on a

scientific basis, their interpretation remains more of an art than a science.
Thus VAR should be considered only as a first-order approximation. The fact
that the value is generated from a statistical method should not hide the fact
that it is only an estimate. Users should not be lulled into a state of
complacency but rather recognise the limitations of VAR, which have been
amply documented in this book. As Steve Thieke, chairman of J.P. Morgan’s
risk management committee says. ‘There has to come a point when this stops
being a risk management methodology and becomes a management issue –
what is the experience of the people in this business, and the firm’s tolerance
for risk’”1253.

11.3.38 Jorion is supportive of “[a]ppropriate use of VAR”1254 and wrote that “[s]tudies
of bank portfolios based on historical data have shown that while the 99
percent VAR is often exceeded, a multiplier of 3 provides adequate protection
against extreme losses” citing a Basel Committee publication in 1999 focussing
on the 1998 Asian and Russian financial crises 1255. That said, his focus is not
on appropriate regulatory standards but quantitative internal risk
management at banks.

11.3.39 Jorion also acknowledges the argument that as VaR totally ignores liquidity,
“faced with binding VAR-based capital requirement, a ‘bank is then faced with
two choices: put in extra capital or reduce its positions, whatever and
wherever they may be. This is what happened last autumn’1256. In turn, these
forced sales depress prices, causing increased volatility, which further feeds
into VAR. This is the vicious-circle hypothesis advanced by Persaud”1257. Jorion
accepted that “[t]his line or argument should be a serious source of concern
given the generalized trend towards risk-sensitive capital adequacy
requirements”, but considered the evidence in their support “anecdotal” and
“[t]o be valid, this explanation requires most VAR-constrained institutions to
start from similar positions”1258. He concluded that “there is no empirical
evidence to support this theory”1259. Factually, this was true of the financial
turbulence of the later 1990s, although the Federal Reserve Bank considered
the failure of the Long Term Capital Management hedge fund to pose sufficient
systemic risk to justify an orchestrated bail-out by the private sector. None
of the major financial institutions that failed in the financial crisis did so
because of VaR capital constraints – indeed the most high profile failures such
as Northern Rock, Bear Stearns and Lehman Brothers exceeded their
regulatory capital requirement until the day they failed. It was a lack of
liquidity and not capital that brought them down.

11.3.40 As VaR under the Basel standards ignored liquidity we are not aware of a more
recent example of the doom-loop operating in the way referred to by Jorion.
However, the doom-loop certainly did occur if we move the focus from capital
to liquidity1260. But as VaR was never designed to address liquidity risk this is
not really a valid criticism of VaR or its regulatory application.

11.3.41 The collapse of Long Term Capital Management may perhaps better be seen
as a case study in poor risk management. Jorion ascribes it mainly to “its
inability to manage its risk” due “in no small part to the fact that LTCM’s
trades were rather undiversified” and that “it did not foresee that it would
be unable to raise new funds as its performance dived”1261. Shin regards the
fall of LTCM as being due to “the endogeneity of correlations [between asset

 Back to contents 200

 MARKET RISK

prices], and the pitfalls of relying on historical correlations as a guide to

portfolio choice, especially when high levels of leverage are
contemplated”1262. Both explanations may be valid, although neither proves
VaR to be ultimately flawed as making unwarranted assumptions about factors
that are not integrated into a risk model may be a failure of risk management
(and LTCM, like Barings before, exhibited that). However, it does not mean
that VaR is wrong for not doing a task it was not designed for.

11.3.42 The limitations on historical correlations highlighted by Shin are well known
in the literature, and are not the only – or theoretically best – way of modelling
market risk. They may, however, be the simplest 1263. As Dowd writes “[w]e
should never rely on non-parametric methods [such as historical simulation]
alone”1264.

11.3.43 Jorion was certainly aware of the importance of liquidity risk and devoted a
chapter to it identifying both asset liquidity risk (where changes in liquidity
affect the price of assets traded in thin markets) as well as funding liquidity
risk (where due to changing conditions leveraged institutions are unable to
raise funding against the provision of collateral) 1265. He proposed some
possible solutions, including liquidity-adjusted VaR measures, such as
incorporating bid-ask spread effects and extending the holding period for
illiquid positions, concluding that while there is no clear answer “the main
basis of this analysis is not so much to come up with one summary risk number
but rather to provide a systematic framework for thinking about the
interactions among market risk, asset liquidity risk, and liquidity funding
risk”1266.

11.3.44 The importance of banks addressing liquidity risk is now clear 1267, as a lack of
liquidity was the proximate (if not underlying) cause of failure in many recent
cases including Enron1268, Long Term Capital Management1269, Northern Rock1270,
Bear Stearns1271, Lehman Brothers1272, RBS1273 and perhaps others. But this is far
removed from the limitations of VaR. Nor is this to deny there were other
more fundamental factors in play in the collapse of those institutions.

11.3.45 Dowd’s first two books on VaR seem broadly 1274 accepting of VaR1275 as a risk
measure1276, although in the second edition of Measuring Market Risk he sees
the main problems arising in model risk i.e. “inadequacies in our risk
models”1277. His prescient conclusion was:

“Model risk is one of the most important and least appreciated areas of market
risk measurement. We go about our work in risk management as if we know
a lot that we actually don’t; we often treat our models as if they are correct,
we might treat parameters as if they are known, and so on. And yet in the
strict sense of the word we actually know very little at all. Instead, we only
ever work with assumptions and have no choice but to do so. However, it is
then all too easy to fall into the trap of starting to think of our assumptions
as if they were true knowledge. We are particularly vulnerable to this trap
because it is a basic human characteristic to seek confirmation of our beliefs:
we all want the world to confirm our views of it, and we tend to brush aside
inconvenient evidence that we might have got it wrong. … Ultimately, model
risk is like the proverbial ghost at the banquet – an unwelcome guest, but one
that we would be very unwise to ignore”1278.

 Back to contents 201

 MARKET RISK

11.3.46 A more critical assessment of VaR was provided in a paper prepared by

eminent economists submitted to the Basel Committee as An Academic
Response to Basel II1279 in 2001. In respect of VaR, after stating criticisms
already noted above, the authors write “VaR models are not robust. They do
not provide accurate and consistent risk forecasts across different assets, time
horizons and risk levels within the same asset class. Moreover, estimates by
Danielsson (2000) show that VaR forecasts are almost as volatile as returns
themselves”1280. The Executive Summary states in its first bullet point “Value-
at-Risk can destabilise an economy and induce crashes when they would not
otherwise occur”1281. VaR models are “problematic in two senses. First, by
failing to acknowledge the endogeneity of risk and liquidity at the systemic
level they produce inaccurate volatility estimates. Second, by encouraging
all market participants to employ similar risk modelling techniques regulation
renders them more homogeneous in risk-aversion and trading strategies, thus
rendering the financial system less stable”1282. The Introduction concludes
“Reconsider before it is too late”1283. Subsequently, as noted below, Danielsson
appears to have changed his mind on the failures of VaR.

11.3.47 A further influential criticism of VaR was advanced by Artzer et al. in a paper
entitled Coherent Measures of Risk1284. This article presents a theory of
“coherent” measures of risk. Artzer at al. posit four requirements for a
measure of risk to be coherent. The only one we will consider here is that it
is “subadditive” i.e. that “a merger [of portfolios of assets] does not create
extra risk”1285. The authors prove that VaR does not satisfy this test as a
combination of two separate portfolios (say the trading book of a bank in
London and New York) can, under VaR, result in a lower number than the sum
of the two separate portfolios calculated independently. Secondly, they show
that VaR can result in situations where a highly concentrated portfolio
generates a lower number than a well-diversified one, violating the principle
that diversification should reduce risk1286, although this is not necessarily true,
and may actually increase risks1287.

11.3.48 McNeil et al1288 conclude that “VaR is not subadditive in general” and
“measuring risks with VaR can lead to nonsensical results”, although “VaR is
subadditive in the idealized situation where all portfolios can be represented
as linear combinations of the same set of underlying elliptically distributed
risk factors … We have seen … that an elliptical model may be a reasonable
approximate model for various kinds of risk-factor data, such as stock or
exchange-rate returns”1289. An “elliptical distribution” is a family of
distributions that include “normal” or Gaussian distributions, as well as many
other distributions with much fatter tails, as well as potentially skewedness,
such as Cauchy distributions, and t-distributions, which conform to certain
specified properties.

11.3.49 The essential question, on which academics differ, is not whether the absence
of subaditivity of VaR is correct, except where distributions of the financial
instruments are elliptical, but whether it matters in the real world 1290, and
further, if it does, whether distributions of losses on financial instruments are
generally elliptical1291. This is clearly a factual question1292. Doubtless the
debate will continue.

11.3.50 On the other hand, Danielsson et al. in Subadditivity Re-Examined: the Case
for Value-at-Risk1293 argues that when focussing on the tails (low risk high

 Back to contents 202

 MARKET RISK

impact events) of heavily tailed assets “VaR is subadditive in the tails, at

probabilities that are most relevant for practical applications. … The results
suggest that there is a strong case for using VaR, and it is not necessary to
consider other risk measures solely for reasons of coherence”1294. For
Danielsson, VaR works where it is most important for risk management
purposes. This is also a factual question. If correct, then it may raise
questions about the decision of the Basel Committee to adopt Expected
Shortfall (ES) given its known limitations (including incalculability in many
cases, which is why, presumably, the Committee still assigns a material role
to VaR under the Basel III framework).

Why did the Basel Committee choose VaR?

11.3.51 Given the criticisms of VaR outlined above, it may be asked why the Basel
Committee adopted the Market Risk Amendment in 1996, implemented it in
1998 and retained it in 2006 as part of Basel II. Perhaps the best place to start
is Charles Goodhart’s analysis in his semi-official history The Basel Committee
on Banking Supervision: a History of the Early Years1295 based on partial access
to the Committee’s archives1296. He traces the recognition of VaR models in
the Market Risk Amendment to a recognition by the Committee that:

“it had somehow failed to take notice, or sufficient account, of the

considerable advances being achieved by the major international banks on
modelling their own internal risk. The banks were now applying both up-to-
date academic (portfolio) theory and empirical measures of risk, for example
based on historical data on volatilities and correlations. The [original] BCBS
proposals had no backing in either theory or empirical evidence. The BCBS
had been looking inward within the regulatory community, rather than
outward to the innovations in risk modelling achieved by the major banks.
Admittedly the adoption of these new risk measures, generally based on
‘value at risk’ (VaR) measures, had only occurred recently, and the pace of
innovation had been swift, but how and why bank supervisors, or at least those
represented at the BCBS, appeared to have become so out of touch with
banking best practice, is not easy to understand” 1297.

11.3.52 Goodhart added:

“To their credit, however, officials at the BCBS immediately recognised both
the validity of the banks’ complaints about the consultative document, the
superiority of the banks’ own techniques and the need for regulators to learn
and master the new modelling techniques. It is always difficult for any set of
authorities to eat ‘humble-pie’ and to accept external criticism, and it is
praiseworthy that the BCBS did so, though admittedly more openly so in their
internal papers than in their public response” 1298.

11.3.53 The Basel Committee in 1994 set up a Models Task Force which “found the
work of the major international banks in assessing their own market risk to be
impressive, and far in advance of their own ‘building blocks’ approach”1299. He
concludes:

“The Market Risk Amendment was, in the end, seen as a considerable success
for the BCBS. … Even so, this exercise also marked the point at which the BCBS

 Back to contents 203

 MARKET RISK

began to become more influenced by the technical modelling expertise of the

large international banks”1300.

11.3.54 Goodhart seems to be arguing that faced with their limited knowledge the
members of the Committee did their best by learning from market practice.

11.3.55 However, in a later chapter on The BCBS and the Social Sciences, he writes
more critically, relying on hindsight derived from the experience of the
financial crisis. Discussing the Market Risk Amendment he adds:

“Unfortunately, the regulators, having been persuaded of the better

analytical foundations of the banks’ VaR models, did not fully appreciate their
limitations from the viewpoint of regulatory objectives. Co-variances and
correlations that determine portfolio risk and VaRs are not stable over time.
In particular, during a crisis co-variances and correlations that were, under
all normal conditions, low would suddenly rise dramatically towards unity. If
one based risk estimates on behaviour during such normal times, for example
relating to the last 250 days, and a crisis strikes (as at the end of September
2008 or August 2008 [surely 2007]), then one can, and does, easily find events
occurring all too often that appear, on the basis of normal times, to be totally
impossible, for example a 25 standard deviation effect.

What that means is that standard VaR models, though of great use as a
management tool during normal times, give regulators little feeling for what
may be flung at them if a crisis should occur, or in jargon terms past
experience provides little indication of future tail risk. What one can see is
what actually happened during the occasional sporadic crises that have
punctured past history.

The regulators did have some appreciation that standard VaR-type models
were only a fair-weather guide, whereas they were supposed to protect the
system against bad weather. They responded in two generic ways. First, they
would introduce multiplication factors, so whatever the VaR figure turned out
to be, the commercial bank would have to hold capital multiplied by X, but
generally the choice of X could only be done by guesswork. The commercial
banks would protest if X was so large as to be commercially damaging for
them.

Secondly, if VaR-type models were only fair-weather guides, why not use the
occasional examples of crises and whatever other crisis scenario regulators
might think up to examine how banks might fare in such circumstances, in
other words to use stress tests as a means of supplementing CARs [capital
adequacy ratios] based on other approaches. One problem is that crises are
not easily foreseeable; if they were, they would not happen. Virtually no one
foresaw the likelihood or the path of the crisis that began in August 2007. Had
regulators asked banks, prior to that time, to assume a scenario in which most
wholesale financial markets closed, this would have been too improbable to
be worthy of consideration”1301.

11.3.56 Goodhart concludes the chapter with thoughts on how the Basel system of
capital adequacy could be reformed.

 Back to contents 204

 MARKET RISK

11.3.57 As Goodhart’s book expressly does not deal with events after 1997, it provides
no evidence for the decision of the Basel Committee not to revise the Market
Risk Amendment as part of its lengthy work on Basel II. Perhaps the most
likely explanation is that other aspects of the framework were in more urgent
need of reform, VaR models seemed to be working well in practice, and no
empirical evidence had yet come to light (as opposed to academic criticisms)
that the assumptions were unsound as none of the financial crises in the
intervening period had involved a banking crisis, or could reasonably be
attributed to the regulation of market risk, and, on their part, the banks were
content with their VaR models, and had not developed new or more
sophisticated ones, unlike for credit risk and operational risk which were the
prime focus of Basel II.

11.3.58 The theoretical underpinnings (and limitations) of VaR seemed to remain

broadly intact, and the criticism that VaR was not a “coherent” measure of
risk was only published in 1999 when work on Basel II was well underway, and
while accepted as theoretically correct by academics, its implications had not
(and still yet have) to be worked out in terms of real results. A committee of
regulators and central bankers, with limited resources and time, and a focus
on what were perceived to be the major issues in upgrading a global standard
of prudential regulation, cannot reasonably be faulted in not launching a
major reform of rules that seemed to be working well when the deficiencies
of Basel I were so manifest. As Goodhart argues above, no one would have
anticipated in 2006 that two years later wholesale funding markets would
close as a result of a chain reaction to a problem in securitised mortgage
loans.

11.3.59 Criticisms based on hindsight may be useful in devising new regulatory

standards but are usually unfair when evaluating decisions made at the time
unless the particular matters either were or ought to have been known to the
decision-makers concerned, and were of a sufficient magnitude to have
required the time of the members of the Committee. Most of the major
theoretical limitations with VaR were known in 1998 and 2006, but the way
they would play out in a crisis that virtually no one foresaw, and seemed so
unlikely as to be not relevant, is not evidence that the regulators did not do
their best, that they should have come up with a better measure when their
initial attempt at devising standards for market risk had been a failure, or
that they were “captured” by the banking industry, as opposed to being
influenced by bank best practice1302. Critics fail to explain what the Basel
Committee should have done instead. That the new chosen metric of market
risk in Basel III, which is discussed in detail below, Expected Shortfall, was
also known in academic literature in 1996, and had actually been proposed in
a research report by JP Morgan before VaR in 1993, is not really relevant. VaR
was the standard that all the most sophisticated banks were using at the time,
and was clearly superior to the original proposals devised by the Basel
Committee. There was no clearly better model on the table, and the jury is
out on whether ES will, in fact, prove superior to VaR in a future financial
crisis, as opposed to in theoretical models.

11.4 The Basel Committee Responds to the Financial Crisis

11.4.1 Given the failures that have been summarised above it is unsurprising that the
Basel Committee engaged in a very lengthy iterative process with banks

 Back to contents 205

 MARKET RISK

involving the publication of four main consultation papers before the new
rules in Basel III for market risk were finalised in 2019. However, as a
preliminary step the Basel Committee implemented a number of measures in
2010 as a stop-gap until a final solution could be found. This is usually
referred to in the literature as Basel 2.5. As these measures are either
superseded or have been amended by Basel III we will not consider them
further here.

11.4.2 The Basel Committee’s response started in January 2011 with a (selective)
review of the academic literature, followed by a first consultative document
in May 2012 and a second consultative document in October 2013. A final
standard was released in 2016, which was then subject to two further
consultations and then finalised in 2019. These documents were
complemented by various quantitative impact studies (QISs), the results of
which were also published, and informed the development of the new
standards. It seems reasonable to conclude from the above that the
Committee has sought to seriously engage with the shortcomings exposed by
the academic literature, while seeking to engage actively with banks’
experience in the financial crisis as well as the views of other stakeholders,
with a more pronounced transparency than is often seen.

The Basel Committee Reviews the Academic Literature

11.4.3 The first major publication by the Committee was Messages from the
Academic Literature on Risk Measurement for the Trading Book published in
January 2011. The five lessons drawn by the Committee from the literature
reviewed will be summarised below and are mainly taken from the Executive
Summary:

On VaR models, there is no unique solution to the problem of the

appropriate horizon for risk measurement (i.e. the holding period), as it
depends on characteristics of the asset portfolio. The square root of time
rule used for deriving a ten day holding period has been found to be
inaccurate in many studies, although there is no widely accepted
alternative. There are also difficulties with volatility over time and
backtesting where only few violations are recorded.

VaR does not address liquidity. The literature distinguishes between

endogenous and exogenous liquidity, as well as normal and stressed
liquidity. Exogenous liquidity (i.e. average transaction costs) can be
captured by a liquidity-adjusted VaR. Endogenous liquidity (i.e. the price
impact of a liquidation of specific positions) depends on trade size as well
as whether other market participants are liquidating their positions at
the same time. However, the practical implications of liquidity in the
trading book have yet to be developed.

The lack of subadditivity (as discussed above) of VaR is relevant in

practice. This conclusion seems based on a 2007 paper by Degen et al. 1303
on operational risk modelling. The other studies citied, cited in the body
of the report, many of which have been referred to above, set out
conditions under which VaR is subadditive, except in cases considered
“artificial”1304, or where there is no other coherent and practicable
alternative1305. Expected shortfall is, however, stated to be a coherent

 Back to contents 206

 MARKET RISK

risk measure, which has been generally recognised since the idea of
coherence was developed in 1999. It also seeks to take account of losses
beyond the specified confidence level (which is also correct).

Stress testing would better be integrated into the risk modelling

framework. The “stressed VaR” adopted in Basel 2.5 (see above) has not
been analysed in the literature but in the Committee’s view may not lead
to an accurate assessment of risk in a stressed environment.

Unified measures of risk are generally preferable as opposed to

compartmentalised measures (e.g. adding the VaR on market risk to the
VaR on credit risk), as the addition of separately calculated VaRs only is
conservative in certain circumstances: “it is always questionable to
calculate different risks for the same portfolio in a compartmentalised
fashion and to hope that adding up the compartmentalised measures will
be a conservative estimate of the true risk”1306.

VaR measures tend to be pro-cyclical and can exacerbate the economic

cycle, although the literature does not offer convincing solutions as to
how such risks could be addressed in the regulatory framework. VaR
models could also induce herd-like behaviour in booms and busts.

The first consultative document

11.4.4 The Basel Committee followed this with its consultative document on the
Fundamental Review of the Trading Book in May 2012. This is a complex
publication and we will consider here its considerations of the shortcomings
of the 1996 Market Risk Amendment, the weaknesses of the Basel 2.5 changes,
the basis on which a new market risk framework should be constructed and
proposals for a revised models-based approach. The approach to these issues
is inevitably selective. The development of the new standardised approach
will be considered in its context. It should be noted that in a late amendment
after the market risk framework had supposedly been finalised an approach
based-on the old standardised approach was re-introduced, essentially due to
a recognition that some banks with limited trading books could not calculate
the new standardised charges at a reasonable cost.

The failings revealed by the financial crisis

11.4.5 The concrete lessons drawn from the financial crisis are set out in Annex 1 of
the publication. This identified three key failings:

weaknesses in the design of the framework;

weaknesses in risk management methodology; and

weaknesses in valuation methodologies.

11.4.6 These will be described in turn.

 Back to contents 207

 MARKET RISK

Weaknesses in the design of the framework

11.4.7 Concerning the design of the market risk framework, the Committee
identified three issues: firstly, the role of the boundary between the banking
book (subject to the credit risk framework) and the market risk framework;
secondly, significant differences between the models-based approach and the
standardised approach; thirdly, the absence of a credible option for the
withdrawal of model approval.

11.4.8 The role of the boundary between the trading book and the banking book is
“an operational one. It seeks to classify instruments into a capital regime
that is equipped to deliver the appropriate level of capital given the nature
of the risks that regulators choose to capitalise under Pillar 1”1307. The
document identified three “features of the boundary”1308 that were
problematic:

The different capital charges for instruments held in the trading book
and the banking book created incentives for banks to arbitrage the
boundary, given the test for allocation was a subjective one based on
intent to trade, rather than whether actual trading took place. The
differences in capital treatment were justified by the assumption that
trading book positions were liquid and could easily be traded out of.

As the test for inclusion in the trading book was one of subjective intent,
banks could arbitrage the boundary by choosing to place instruments that
were inherently hard to trade or held for long-term investment, into the
trading book by claiming that they intended to trade them. Equally,
during the crisis, some banks reallocated positions that had become
illiquid and subject to extreme price volatility on a mark-to-market basis
to the banking book preventing further losses unless the instruments
became impaired.

The interaction with accounting standards caused problems. While all

assets that are fair valued through the profit and loss account and
available-for-sale must be marked-to-market daily, not all such positions
were required to be held in the trading book, such as options that were
fair valued. Under Basel II fair valued gains and losses were not
automatically taken forward to core Tier 1 capital, so fair value losses
would avoid an immediate hit to capital.

11.4.9 The second main criticism identified by the Committee is the significant
difference between capital calculations under the standardised and VaR
models-based approach. “The design of the current framework does not
embed a clear link between models-based and standardised approaches,
either in terms of calibration or in terms of the conceptual approach to risk
management. Historically, the two approaches have been seen as catering to
different sets of banks”1309. This was seen as acceptable “partly in the belief
that there should be significant capital benefits for the models-based
approach. This has been justified on the basis that it is appropriate to provide
regulatory capital incentive, over and above the private incentives that banks
should have on their own, for good risk management”1310.

 Back to contents 208

 MARKET RISK

11.4.10 The third main criticism follows: “a key weakness of the design of the current
framework has been the lack of credible options for the withdrawal of models
approval”, particularly in stressed periods1311.

Weaknesses in risk measurement

11.4.11 In this section we will discuss only weaknesses identified in respect of the
models-based approach.

11.4.12 The Committee questioned whether calibration on a 99% confidence level

based on a ten day holding period “meets prudential objectives”1312. The
following limitations are noted in the document:

VaR does not adequately capture credit risk in the trading book. The
market for traded credit grew significantly since 1996 (e.g. CDOs and
credit default swaps). Basel attempted to address this partially with the
incremental risk charge in Basel 2.5 which is abolished under Basel III so
will not be discussed further.

VaR models proved inadequate in capturing ex ante market liquidity risk:

“banks were often unable to exit or hedge certain trading positions in a
short time period due to market illiquidity. Moreover, the sharp rise in
liquidity premia at the height of the crisis led to banks incurring
substantial mark-to-market losses on a range of positions”1313. (This is
obviously true, but VaR models were never designed or required to
consider liquidity risk). Although liquidity and market risks are related
they are not the same thing, and pre-crisis liquidity was taken as a given
and not something to be modelled.

Banks had incentives to take on tail risk: “[b]y not looking beyond the
99th percentile, VaR – and hence regulatory capital requirements – fails to
capture so-called ‘tail risks’”1314. (As we have seen this is a design feature
of VaR, but nothing prevented banks’ risk management internally from
considering such risks).

VaR models did not adequately capture basis risk. “Ahead of the crisis,
internal models often did not capture the basis risk between market
parameters as they were often ‘mapped’ to the same underlying risk
factor”1315, such as a long position in a corporate bond mapped by
equivalent CDS protection, which could deliver a zero VaR. (VaR models
were not required to have this effect). “More broadly, the entire
framework was based on the estimates of correlations derived from
historical data based on ‘normal’ market conditions”1316. (Again, this is
not a necessary feature of VaR models if the data set includes non-normal
conditions. The problem was that there had been no recent banking
crises in the US or Europe, and no regulatory requirement to take such
risks into account).

VaR models relied on a bank-specific notion of risk: “individual banks’

risk assessments might not be adequate from the perspective of the
banking system as a whole. A clear manifestation of that was the lack of
incorporation of market liquidity risk”1317.

 Back to contents 209

 MARKET RISK

VaR-based capital charges are pro-cyclical, allowing banks to accumulate

more risk in the boom, and being under pressure to shed risk by exiting
positions in the crisis contributing to further asset price falls and
liquidity. (This is again true, but applies equally to credit risk as all risk-
based measurements that are not fixed once and for all – as under Basel
I – are pro-cyclical. There is little evidence of major banks shedding risky
positions in the crisis solely because of regulatory capital constraints.
The main difference between pro-cyclicality in the trading book, as
opposed to the banking book, lies in the difference in accounting
treatment: historic cost versus mark-to-market/fair value, even if there
are some exceptions as noted above).

There were a large number of backtesting exceptions (i.e. breaches of

the VaR figure on the next trading day) during the financial crisis. This
led to further work by the Trading Book Group, whose main conclusions
were that “VaR models did not perform adequately during the crisis”1318,
there were two clear peaks in August 2007 and October 2008
corresponding to “the significant rise in inter-bank funding spreads” in
the summer of 2007 and “the fall-out of the Lehman Brothers collapse in
autumn 2008”1319, “[t]he main cause of exceptions were interest rate
movements”1320, and, finally, “many banks did not regularly update time
series data and some key risk factors driving the observed losses were
not incorporated into their VaR models”1321.

Weaknesses in valuation practices

11.4.13 The report notes that the “crisis served to highlight the importance of
valuation practices, especially of complex or illiquid financial instruments in
times of stress, for the regulatory assessment of capital adequacy”1322.

11.4.14 Perhaps unsurprisingly, the consultative document does not mention the
possibility of regulatory failure either in the design of the Market Risk
Amendment, or failures by leading regulators. Purely by way of example, the
FSA decided on 29 June 2007 to approve Northern Rock’s application for a
waiver to use the advanced IRB approach to credit risk that significantly
reduced its capital requirements. Northern Rock then declared a 30.3%
increase in its interim dividend on 25 July 2007 1323. It failed when it ran out
of money on 13 September 2007, making the timing for a significant reduction
in its capital seem unfortunate.

11.5 The Basel Committee’s May 2012 Proposals

11.5.1 The next sections describe the choices ultimately made by the Basel
Committee in designing the new models-based standard.

11.5.2 We pass over the criticisms made by the Basel Committee of the adequacy of
the Basel 2.5 changes. These were always intended as a stop-gap, necessary
to address some lessons of the crisis, but never as a comprehensive new
framework. The Committee sets out five key elements to the proposed new
framework:

 Back to contents 210

 MARKET RISK

a reassessment of the boundary between the trading book and the

banking book. The Committee rejected abolishing the boundary, whilst
insisting on improvements;

a new metric to replace VaR by expected shortfall (ES);

factoring in market liquidity;

the treatment of hedging and diversification risk; and

a revised relationship between the standardised and models-based

approaches.

11.5.3 All of these principles are reflected in the final 2019 Basel III standard. It may
be questioned why, given the introduction of a new metric based on expected
shortfall, we have devoted so much space to VaR in the pages above. There
are four reasons for this.

Firstly, ES is essentially an adjusted, albeit less flexible in some respects,

measure than VaR. In other words, in most cases, if you can calculate
the VaR then systems can be adapted to generate the ES for the portfolio
concerned. Understanding VaR is therefore necessary to understand ES.

Secondly, a number of the criticisms directed against VaR apply in a

similar way to ES. This is particularly the case when it comes to data
limitations when calculating holding periods (now renamed liquidity
horizons). By definition, in seeking to look into the tail beyond a 99%
confidence level you will need more relevant data. The problems with
data limitations described above in respect to VaR apply a fortiori to ES,
as there will be smaller relevant data sets than for VaR. Essentially, Basel
III adopts the square root of time followed by the original VaR standard
to determine longer liquidity horizons than 10 trading days in calculating
ES. Problems with the VaR measure due to limited data are therefore
likely to be exacerbated with the move to ES.

Thirdly, for some portfolios ES cannot be calculated, in which case VaR

may be the only risk management tool available to banks that do not
want internally to rely on the new standardised approach. If an ES
measure is unavailable, or the bank wishes to manage risk for purposes
other than regulatory capital (and there are many reasons why regulatory
requirements are not the sole determinant of internal risk management)
then banks may still continue to rely on VaR (say for daily risk
management based on the previous day’s trading in setting the trading
strategy for the next day, even if the results reported to regulators
differ). ES may be a better measurement for regulatory purposes but
there may be other reasons for using a different risk measurement for
day-to-day decisions, particularly if that is already embedded in internal
systems and reliable data sets to calculate it exist. Of course, there is a
cost in operating internal risk management models different from
regulatory reporting, but few sophisticated institutions totally outsource
risk management to regulators, or third party vendors of data, or would
consider it prudent to do so, as regulatory requirements may be over
simplified.

 Back to contents 211

 MARKET RISK

Finally, although ES is now the regulatory measure for capital held

against market risk, Basel III still relies in important ways on VaR. Firstly,
the default capital charge (DRC) is based on VaR, to ensure consistency
with the credit risk framework. Further, the model validation and
integrity standards that must be met for a bank to use an ES model are
themselves based on VaR. Implicitly, despite the criticisms referred to
above, the Basel Committee continues to view VaR as either a useful or
even an essential measurement of risk for some purposes.

11.5.4 We will not describe in detail the proposed 2012 revised models-based
approach as it was subject to substantial revision in 2013. However, we note
the Committee’s “objective for the models-based approach to calculating
regulatory capital for the trading book is to estimate the amount of capital
required to cover a potential loss in a period of stress from all sources of
risk”1324.

11.5.5 There are three steps to the new framework that are retained by the final
standard.

The first step of the new framework is to determine eligibility for the
models-based approach. This involves a focus on an overall assessment
of a bank’s internal model. If the bank fails this step, then it is required
to use the standardised approach for all its trading book exposures.

The next step is to break down the model approval process. Under the
Market Risk Amendment, this was an all or nothing approach: if a model
was approved then it applied across all trading activities in all
instruments. Now, the Basel Committee’s assessment of model approval
applies at the level of a “trading desk” (not an individual trader) level.
What this means is considered below when we come to consider the
detailed new rules. This means that a bank could have approval for some
but not all of its trading desks, depending on meeting the required
criteria. Also, if a trading desk or desks cease to meet the criteria then
model approval will be withdrawn, without forcing the bank to revert to
the standardised approach for all market risk positions.

The third step is to assess, on a trading desk-by-trading desk, which risk

factors those desks are eligible to model, based essentially on the quality
of data and how rapidly they can be updated. Those risk factors capable
of modelling are capitalised using an expected shortfall approach.
Factors deemed non-modellable, or not included in a particular trading
desk’s risk model, are capitalised in accordance with supervisory
requirements specified by the Basel Committee.

11.5.6 The capital charge for eligible trading desks is the aggregated ES model
requirement for modellable risk factors, plus the sum of the capital charge
for non-modellable factors, with an additional capital charge for default and
other risks. The aggregate capital charge for market risk consists of the
capital charge (computed as above) for eligible trading desks, plus the
standardised capital charge for all non-eligible trading desks.

 Back to contents 212

 MARKET RISK

11.6 The Second Consultative Document

11.6.1 The second consultative document is much broader in scope than the first,
makes numerous changes in detail and also proposes a revised text. Only
those aspects that remain relevant to the new market risk framework will be
described. Although most of the main elements of the models-based approach
are described in detail in the second consultative document, further
refinements were made in the third consultative document (December 2014),
the Market Risk Standard (January 2016), the fourth consultative document
(March 2018) and the final market risk standard (February 2019) 1325. This
clearly reflects prolonged engagement with the banking community.

11.6.2 The second consultative document has five sections of which only the first
two will be considered here: overall revisions to the market risk framework
and the revised models-based approach. Proposals relevant to the
standardised approach are considered under that heading in this chapter.

11.6.3 The first part of the consultative document (overall revisions) sets out
proposals in six areas: (1) the boundary between the trading book and the
banking book; (2) the treatment of credit risk; (3) factoring in market
liquidity; (4) the new chosen risk metric (ES) and the treatment of stress; (5)
the treatment of hedging and diversification; and (6) the relationship between
the models-based approach and the standardised approach.

The boundary between the trading and banking book

11.6.4 This “has been a source of weakness in the current regime. A key element of
the existing boundary has been between banks’ effectively self-determined
intent to trade, an inherently subjective criterion that has proved difficult to
police and insufficiently restrictive from a prudential perspective”1326. The
Committee designated as relevant criteria:

objectivity in the definition of the boundary;

mitigation of opportunities for capital arbitrage;

making the boundary less permeable;

alignment with banks’ risk management processes; and

ease of application.

11.6.5 The 2012 consultation advanced two alternatives: a trading evidence-based

approach and a valuation-based approach. Both were rejected in 2013 in
favour of a revised boundary based on how positions are risk managed, as well
as to prevent banks from obtaining capital benefits from re-allocating
positions between the trading and banking book. The Committee also
endorsed a set of “presumptions” as to how particular positions should be
allocated: “[g]enerally, instruments presumed to be in the trading book are
so designated because they are held with the intention of short-term price
movements or arbitrage profits; and/or to hedge risks resulting from such
instrument types. … These objective requirements and presumptions should
facilitate the development of a common understanding amongst supervisors

 Back to contents 213

 MARKET RISK

on the types of instrument that would typically be included in the different

books”1327. However, “[t]he Committee has not pursued specific quantitative
thresholds for determining whether certain instruments would be allowed in
the trading book. For example, the draft Accord does not include a regulatory
definition of ‘stale positions’. This decision has been taken on the grounds
that it is difficult to determine universal quantitative criteria, applicable
across different instruments, portfolios, banks and jurisdictions”1328.

11.6.6 Regulatory arbitrage is addressed by “strict constraints on switching

instruments between books. After initial designation of an instrument to
either the trading or banking book, a bank may not re-designate the
instrument to the other book, except in extraordinary circumstances. …
Where an instrument is switched to another book, the bank will not be allowed
to benefit from a lower regulatory capital requirement from this switch.
Supervisors are also explicitly given the authority to require a bank to re-
designate a given instrument”1329. The “second measure to mitigate the risk
of regulatory arbitrage is an effort to better align the trading book and
banking book charges”1330.

Addressing default and credit risk migration

11.6.7 The document states that “[c]redit-related instruments were a key source of
losses during the crisis and the regulatory treatment of these positions proved
particularly flawed”1331. Following the responses to the 2012 consultative
document the Committee decided that “the total capital charge for credit
risk will have two separate components; an integrated credit spread risk
capital charge, which will also cover migration risk, and an Incremental
Default Risk (IDR) capital charge”1332.

11.6.8 The “objective of the capital requirement for credit spread is to capture the
risk of changes in the market value of credit instruments with respect to the
volatility of credit spreads”, whereas “[t]he objective of the capital
requirement for default risk is to capture the incremental loss from defaults
in excess of the mark-to-market (MtM) loss from changes in credit spreads and
migration. The capital charge for MtM loss captures the risk of changes in
credit spreads. It does not capture the risk of loss from jump to default”1333.

11.6.9 According to the Committee, “[c]redit spreads capture the expected loss from
default (i.e. PD multiplied by LGD) and are a measure of the mean of the
distribution of default losses (see chart). A change in credit spreads
represents a shift in the mean of the mean of the default distribution.
Therefore, the risk of an MtM loss from MtM risk is a capital charge for the
volatility of the mean of the default distribution changes. The ES measure of
MtM risk is a capital charge for the mean of the default distribution. This
measure does not capture the risk of a jump to default (a jump in the mean
of the default distribution to 100%). Thus, banks must hold an incremental
amount of capital against default risk. The capital requirement for default
risk is an IDR charge to capture the risk that the severity of defaults over the
capital horizon will fall in the extreme tail of the default distribution. To
avoid double counting, the capital charge for default risk should be quantified
in terms of incremental default loss relative to mark-to-market losses already
recognised in market values”1334. The proposed incremental capital charge is

 Back to contents 214

 MARKET RISK

based on a VaR calculation over a one year horizon calibrated to a 99.9%

confidence level, which is the same as that applied to the banking book 1335.

 Back to contents 215

 MARKET RISK

Frequency distribution of losses from default

Shift of mean of distribution

= shock in credit spreads

95% percentile of
initial distribution

Loss amount
μ μ’ 95%

 Back to contents 216

 MARKET RISK

11.6.10 Given variations in banks’ existing default and migration models, the
Committee decided to require a more prescriptive treatment based on a two-
factor default simulation model. Default correlations must be based on listed
equity prices estimated over a one-year time horizon (using stressed
figures)1336.

11.6.11 Basel 2.5 introduced a capital charge to capture the risk of changes in credit
valuation adjustment (CVA). This requires banks to capitalise the impact of
changes in their counterparties’ credit spreads on all OTC derivatives after
taking account of permitted hedges. A standardised and an advanced CVA
approach are available. Basel III modifies this stand-alone capital charge
within the market risk framework, although the models-based approach is
based on the ES model used to capture bond credit spreads1337. It is described
in chapter 12.

Changes to the liquidity horizon

11.6.12 The liquidity horizon (referred to as the holding period under Basel II) was 10
trading days. This reflected a judgment that banks would be able either to
trade out, or effectively hedge, any position held within the trading book
within two weeks. However, the evaporation of market liquidity during the
acute stage of the financial crisis meant that banks were often unable to so.
“This violated a key assumption that was implicit in the 10-day VaR treatment
of market risk. Moreover, large swings in liquidity premia, defined as the
additional compensation required by investors to hold illiquid instruments led
to substantial mark-to-market losses on fair-valued instruments as liquidity
conditions deteriorated”1338. Unsurprisingly, the Committee made changes to
the Basel II treatment. The proposal is to introduce varying liquidity horizons
to take account of the fact that banks may be unable to hedge or exit positions
without triggering material market price changes. The following changes
apply:

the liquidity horizon is defined as “the time required to execute

transactions that extinguish an exposure to a risk factor without moving
the price of the hedging instruments, in stressed market conditions”1339;

risk factors are assigned to five generic categories ranging from 10 to 120
trading days1340;

liquidity horizons are set based on supervisory estimates and not left to
banks’ own internal modelling. This is “in recognition of the fact that
market liquidity is a systemic concept: while individual banks might judge
that they can all promptly exit or hedge their risk exposures without
affecting market prices, the market is likely to turn rapidly illiquid in
times of banking system stress if the banking system as a whole holds
similar exposures”1341;

risk factors are grouped into separate categories. “[S]upervisors will be

able to require additional capital against exposures to particularly illiquid
risk factors within a broad category”1342. Firms must map their exposures
to the relevant category. For each of the five broad risk categories there
is an “other” category “to capture all risk factors under any of the other

 Back to contents 217

 MARKET RISK

buckets”, which is “likely to be a relatively heterogeneous category”1343;

and

in calculating ES a key practical consideration is how to apply risk factor

shocks over longer and varying time horizons i.e. to extrapolate a longer-
term holding period from data that are available only for a shorter period
and still produce results that are meaningful.

11.6.13 Further revisions of a highly technical nature were made to the calculation of
ES in the December 2014 third consultative document. The main change was
to adopt a revised ES calculation for a base horizon for all risk factors, which
would be scaled up to the longer specified risk horizon, using the square root
of time, the limitations of which have been discussed above.

Expected shortfall

11.6.14 Expected shortfall was outlined above and is the basis of Basel III’s models-
based market risk framework, and has enjoyed academic support as being
superior to VaR, although it is based essentially on the same risk models as
VaR. Kevin Dowd, in the second edition of Measuring Market Risk1344, writes
that “the ES easily dominates the VaR as a risk measure”. His reasons are
essentially as follows:

ES tells us what to expect in bad states;

an ES-based risk-expected return decision rule is valid under more

general conditions than VaR-based risk-expected return decision rule;

ES satisfies “sub-additivity” while VaR does not (this has been considered
above); and

the ES portfolio risk surface is convex, meaning that portfolio-optimising

problems have a unique well-behaved optimum1345.

11.6.15 However, Dowd concludes that “ES is also rarely the ‘best’ coherent risk
principle”1346, essentially as ES risk weights “imply that the user is risk-neutral
between tail outcomes. Since we usually assume that agents are risk-averse,
this would suggest that the ES is not, in general, a good risk measure to use,
notwithstanding its coherence. If a user is risk-averse, it should have a
weighting function that giver higher losses a higher weight”1347. He advocates
spectral risk measures based on risk aversion 1348. As such measures are
necessarily subjective, based on the “user’s risk aversion function”1349, it is
hard to see how such metrics could ever form part of a regulatory standard,
as allowing banks to determine their own risk-appetite when setting capital
requirements creates inescapable problems of self-selection and moral
hazard.

11.6.16 Further possible problems with ES identified by Auer 1350 include:

ES might effectively capture an incomplete or distorted tail risk, possibly

providing a false sense of risk coverage;

 Back to contents 218

 MARKET RISK

poor quality or missing data can arbitrarily distort ES calculations; the

same is not so for VaR as individual outliers are likely to have a very small
effect on VaR; and

backtesting ES against actual outcomes is harder than VaR, and less

transparent1351.

11.6.17 Auer concludes “while the ES is indeed smartly sub-additive, it is more

pretentious in its assertions, resentful with respect to outliers, and more
elusive to validate. In a bit more earnestness, the ES is a fine measure with
no agenda of its own. We should probably just try not to infuse it with an
expressiveness unsupported by the underlying data”1352. Only time will tell
whether these are real, as opposed to theoretical, concerns.

11.6.18 The confidence level for the new ES metric is set at 97.5%, meaning that
banks’ models must be 97.5% accurate for each given liquidity horizon. The
Committee states that this will “provide a broadly similar level of risk capture
as the existing 99th percentile VaR threshold, while providing a number of
other benefits, including generally more stable model output and often less
sensitivity to extreme outlier observations”1353. We cannot express any view
on this given the change to the risk metric and other changes made, including
to the liquidity horizon for different classes of instrument. The only thing
that can be said for certain is that the capital charge will be quite different
than under Basel II1354.

11.6.19 Basel III requires banks to use a stressed calibration. The Basel 2.5 reforms
introduced a “stressed VaR” calculation to take into account the fact that VaR
figures produce an inadequate measurement of risk in a time of financial
stress owing to its pro-cyclicality. However, simply adding a stressed
calculation to a market-based calculation is duplicative and, although likely
to be conservative, is not really a sensible way to measure risk. Hence the
stress element is directly integrated into the ES calculation, and there is no
additional charge.

11.6.20 As it is not practicable to develop an ES model that captures the full set of
current risk factors for other than a relatively short period of time, the
Committee originally decided that the data set must go back to 2005 (2007 in
the final standard) based on a reduced set of risk factors relevant to individual
banks’ portfolios and for which there is a sufficiently long history of data.
However, banks are not given total freedom to determine the set of reduced
risk factors for the stressed figure, and the chosen factors, in addition to
meeting specified requirements must explain at least 75% of the variation of
the full ES model.

11.6.21 The ES calculated using this reduced set of risk factors is then calibrated to
the most severe 12 month period of stress since 2007 (or earlier, if a bank can
calculate this). The stressed period is calculated based on the bank’s
aggregate portfolio and not on specific risk factors. For most banks, this is
likely to be 2007 or 2008 at the time of writing 1355. This value is then scaled
up by the ratio of the current expected shortfall using the full set of risk
factors divided by the current expected shortfall using the reduced set of risk
factors.

 Back to contents 219

 MARKET RISK

Hedging and diversification

11.6.22 Under Basel II banks were permitted to internally model correlations within
market risk measures (VaR, stressed VaR, IRC, etc.) and then add up the
relevant capital charges. The Committee observed that as trading book
portfolios contain a mix of long and short positions, the correlation
assumptions (and implicitly, therefore, the capital treatment for hedging and
diversification benefits) can materially affect the capital charge. Further,
higher correlations may, depending on the composition of the portfolio, result
in lower capital charges, which the Committee does not consider to be a
conservative treatment. Hence it was decided that “the capital framework
should only recognise hedges if they are likely to prove effective – and can be
maintained – during periods of market stress”1356.

11.6.23 With this lengthy introduction we now proceed to set out, at a more granular
level, the Basel III standard for market risk.

11.7 Basel III - The Definition of the Trading Book

11.7.1 Unlike earlier iterations of the Basel Accord, there is no single definition of
the trading book1357. Rather, instruments that meet the criteria set out in the
following paragraphs are allocated to the trading book or to the banking book.
Any residual positions are allocated to the banking book 1358. Perhaps
surprisingly, the rules are not set out in the modular section of the
consolidated Basel framework on market risk (MAR) but in that on risk-based
capital (RBC).

11.7.2 Instruments are defined as “financial instruments, foreign exchange (FX) and
commodities. A financial instrument is any contract that gives rise to both a
financial asset of one entity and a financial liability or equity instrument of
another entity. Financial instruments include primary financial instruments
(or cash instruments) and derivative financial instruments. A financial asset
is any asset that is cash, the right to receive cash, the right to receive another
financial asset or a commodity, or an equity instrument. A financial liability
is the contractual obligation to deliver cash or another financial asset or a
commodity. Commodities also include non-tangible (ie non-physical) goods
such as electric power”1359.

11.7.3 Logically crypto-currencies or crypto-assets should be treated as a commodity

if they are not intangible assets, money or a financial instrument. Some US
courts and regulatory agencies have classified such digital assets as either a
commodity or a security under federal law. However, the approach of the
Basel Committee is to regard then as outside the current scope of the Basel
framework, although in June 2021 the Committee published a consultative
document on the prudential treatment of crypto-asset exposures. A second
consultative document was published in June 2022. As no final standard has
yet been released, this will not be considered further in this publication,
although the intention is to incorporate the final standard into the
consolidated Basel III text.

11.7.4 Instruments, FX and commodities can only be included in the trading book if
there is no legal impediment to selling or fully hedging them1360. Trading book

 Back to contents 220

 MARKET RISK

instruments must also be fairly valued on a daily basis with valuation changes
recorded in the profit and loss (P&L) account.

11.7.5 Any instrument that a bank holds for one or more of the following purposes
must be allocated to the trading book unless Basel III requires it to be
allocated to the banking book:

short-term resale;

profiting from short-term price movements;

locking in arbitrage profits; or

hedging risks arising from any instrument falling within the three previous
points1361.

11.7.6 All of the following instruments are deemed to be held for one of the above
purposes (and therefore allocated to the trading book) unless required to be
allocated to the banking book:

instruments in the correlation trading portfolio;

instruments that would give rise to a net short credit or equity position
in the banking book (i.e. if the present value of the banking book
increases when an equity price decreases, or when a credit spread on an
issuer or group of issuers of debt increases); and

instruments resulting from underwriting commitments, in relation to

securities, where the bank expects to purchase those securities on the
settlement date1362.

11.7.7 There is a general (but rebuttable) presumption that the following

instruments are trading book instruments (unless explicitly required to be
allocated to the banking book; see below):

instruments held as accounting trading assets or liabilities;

instruments resulting from market-making activities;

equity investments in funds;

listed equities1363;

trading-related repo-style transactions; and

options, including embedded derivatives, from instruments issued out of

the bank’s own banking book which relate to credit or equity risk 1364.

11.7.8 The following instruments are, in all cases, allocated to the banking book:

unlisted equities;

instruments designed for securitisation warehousing;

 Back to contents 221

 MARKET RISK

real estate holdings and derivatives on real estate holdings;

retail and SME credit;

equity investments in a fund unless at least one of the following two

following criteria are met: (1) the bank can “look through” the fund to
its component parts and there is sufficient and frequent information,
verified by an independent third party, proving the fund’s composition;
or (2) the bank obtains daily price quotes for the fund and has access to
the fund’s mandate;

hedge funds;

derivatives and funds that have the same instrument types as underlyings
listed above; and

instruments held for the purpose of hedging a particular risk of a position

in the types of instruments listed above 1365.

11.7.9 Any other instruments not referred to in the preceding paragraphs are
allocated to the banking book1366.

11.7.10 Basically, the scheme is as follows:

some instruments must be allocated to the trading book (unless required

to be allocated to the banking book in which case the banking book
treatment prevails);

some instruments must in all cases be allocated to the banking book;

some instruments are presumed to fall within the trading book, but this
presumption may be rebutted; and

any residual class of instruments falls within the banking book.

11.7.11 Compared with the original definition based on subjective trading intent the
purpose of these rules is to restrict banks’ ability to select which instruments
are held in the trading book and which in the banking book. However, the
Committee accepts that some instruments (e.g. repos) may be entered into,
or used to hedge exposures, in either book. Hence the “general presumption”
referred to above. Further, repo-style transactions entered into for liquidity
management and valued at accrual are not presumed to be trading book
transactions1367 for obvious reasons.

11.7.12 For the presumptive list, banks may choose to classify a position in the banking
book only if it receives approval for a banking book treatment on the basis
that it is not held with trading intent (e.g. a 5 year repo intended to be held
to maturity)1368. If approval is denied the bank must include the instrument in
the trading book1369.

 Back to contents 222

 MARKET RISK

Supervisory powers

11.7.13 National supervisors can require banks to provide evidence that an instrument
on the presumptive list is actually held for trading purposes. If the supervisor
considers the instrument would customarily belong in the banking book it may
require the bank to reassign it unless it must be held in the trading book 1370.
The same applies to instruments held in the banking book that the national
supervisor considers customarily to belong to the trading book 1371.

Restrictions on transfers between the trading and banking books

11.7.14 Basel III introduces strict restrictions on moving instruments between the
trading and banking books. This is to prevent regulatory arbitrage. During
the financial crisis certain institutions moved positions previously held in the
trading book to the banking book to avoid the need to report mark-to-market
losses where trading was taking place at distressed levels in thin markets on
the basis the bank no longer intended to trade the instruments. It may be
questioned whether this is properly attributed to actual regulatory arbitrage
as opposed to the inappropriate use of fair value accounting in a crisis.
However, the Basel Committee’s decision to address possible regulatory
arbitrage is understandable.

11.7.15 Basel III states that “[s]witching instruments for regulatory arbitrage is strictly
prohibited. In practice, switching should be rare and will be allowed by
supervisors only in extraordinary circumstances. Examples are a major
publicly announced event, such as a bank restructuring that results in the
permanent closure of trading desks, requiring termination of the business
activity applicable to the instrument or portfolio or a change in accounting
standards that allows an item to be fair-valued through P&L. Market events,
changes in the liquidity of a financial instrument, or a change of trading intent
alone are not valid reasons for reassigning an instrument to a different
book”1372.

11.7.16 Moreover, “[w]ithout exception, a capital benefit as a result of switching will

not be allowed in any case or circumstance. This means that the bank must
determine its total capital requirement (across the banking book and trading
book) before and immediately after the switch. If this capital requirement is
reduced as a result of this switch, the difference as measured at the time of
the switch will be imposed on the bank as a disclosed Pillar 1 capital
surcharge. This surcharge will be allowed to run off as the positions mature
or expire, in a manner agreed with the national supervisor. To maintain
operational simplicity, it is not envisaged that this additional capital
requirement would be recalculated on an ongoing basis, although the positions
would continue to be subject to the ongoing capital requirements of the book
into which they have been switched”1373.

11.7.17 It follows that if a national supervisor permits a switch, then there could be
a capital benefit if the switch results in lower on-going capital requirements
than would have been the case had the switch not occurred. This would seem
most likely to be the case where the switch is to the banking book and the
instrument is held at value on transfer in circumstances where the application
of fair value accounting would have resulted in further losses, but there is no

 Back to contents 223

 MARKET RISK

impairment of the asset when held in the banking book. Such cases are
intended to be very rare in practice.

11.7.18 Given the general restriction on switching we do not discuss further the
specific rules on transfers between the two regulatory books.

11.7.19 In terms of internal risk transfers (i.e. without switching), there is no

regulatory capital recognition of internal risk transfers from the trading book
to the banking book1374. Specific rules apply to internal risk transfers to the
trading book from the banking book1375. Further discussion is outside the scope
of this publication. Internal risk transfers between trading desks within the
trading book generally receive regulatory recognition 1376. Specific rules apply
to internal risk transfers in respect of the credit risk valuation (CVA) capital
requirement1377.

11.8 Basel III - The New Market Risk Framework for Banks Using Internal Models

11.8.1 We now consider first the minimum requirements – both quantitative and
qualitative – that apply under the internal models-based approach, before
providing a description of the capital charges and how they are determined.

11.8.2 Market risk is defined as “the risk of losses in on- and off-balance sheet risk
positions arising from movements in market prices”1378.

11.8.3 A bank requires the consent of its supervisor to apply an internal models
approach1379. Any significant changes to the model must also be approved in
advance1380. Approval will only be granted if (inter alia):

the bank’s risk management system is conceptually sound and

implemented with integrity;

the bank has sufficiently skilled staff in the use of models, including
trading and risk management; and

the bank’s model has a proven track record of reasonable accuracy in

measuring risk1381.

11.8.4 A four stage approach applies to the models-approval process. Firstly, the
bank must show that its internal risk management model meets the
requirements set out in MAR. Secondly, the bank must nominate trading desks
for supervisory approval for use of the internal models approach. Thirdly, the
supervisor determines whether the nominated trading desks individually meet
the requirements. Finally, the bank must meet the profit and loss application
tests, as well as the backtesting requirements on an on-going basis1382. This is
illustrated in the diagram below:

 Back to contents 224

 MARKET RISK

The Internal Models Approach for Market Risk

Step 1
Standardised approach for
Overall assessment of the banks’ firm-wide entire trading book
internal risk capital model Fail
Pass

Step 2(i)
Banks nominate which trading desks are in-scope
for model approval and which are out-of-scope Out of scope

Pass Standardised approach for

specific trading desks
Step 2(ii)
Assessment of trading desk-level model Fail
performance against quantitative criteria: Clear
thresholds for breaches of P&L attribution and
backtesting procedures Securitisation exposures in the
trading book are fully out of
Pass scope of internal models and
capitalised in the revised
Step 3
standardised approach
Individual risk factor analysis
 Risk factors must be based on real,
verifiable prices
 Frequency of observable prices

Modellable Non-modellable

Global Expected Shortfall Default Risk Charge Stressed capital add-on

(ES) (DRC) (SES)
Equal weighted average of Captures default risk of Aggregate regulatory capital
diversified ES and non- credit and equity trading measure for non-modellable
diversified partial ES capital book exposures with no risk factors in model-
charges for specified risk diversification effects eligible desks
allowed with other market
risks

The definition of trading desks

11.8.5 It has been explained that Basel III requires model approval on a trading desk-
by-trading desk basis. We now proceed to describe the relevant requirements
for a “trading desk”. This is “a group of traders or trading accounts that
implements a well-defined business strategy operating within a clear risk
management structure”1383. It is therefore not simply an individual trader
acting on his or her own.

 Back to contents 225

 MARKET RISK

11.8.6 Banks can always define their trading desks, but the choice of desks is subject
to supervisory approval if the bank wishes to use an internal ES model to
calculate the capital requirements for that trading desk 1384. Banks may also
designate sub-desks for internal risk management purposes, but these cannot
be used to calculate regulatory capital1385. Basel III sets out three “key
attributes” of a trading desk. These are as follows:

A trading desk is an unambiguously defined group of traders or trading

accounts. The following specifications apply:

a trading account is an indisputable and unambiguous unit of

observation in accounting for trading strategy;

the trading desk must have one head trader, but can have two
provided either one has authority over the other, or their
respective roles and responsibilities are clearly separated;

each trading account must be assigned to a single trading desk;

there is a presumption that traders (and head traders) are assigned

to only one trading desk, although an individual trader may work
across several trading desks with supervisory approval, if it can be
justified on sound management, business or resource allocation
reasons; and

the trading desk must have a clear reporting line to senior

management, and a compensation policy linked to the objectives
of the desk1386.

A trading desk must have a well-defined and documented business

strategy, including an annual budget and regular management reports.
In particular:

there must be a clear description of the economics of the business

strategy for the trading desk, its primary activities and trading
strategies;

the management team at the trading desk must have a clear annual
plan for budgeting and staffing of the trading desk; and

a trading desk’s documented business strategy must include

regular management information reports1387.

A trading desk must have a clear risk management strategy. This

encompasses the following:

the bank must identify key groups and personnel responsible for
overseeing risk-taking activities at the trading desk;

the trading desk must have clearly defined trading limits based on
the desk’s business strategy. These must be reviewed at least
annually by senior management;

 Back to contents 226

 MARKET RISK

a trading desk must produce, at least weekly, appropriate risk

management reports, including profit and loss reports, and risk-
measurement reports that include both desk VaR and ES
measures1388.

11.8.7 Banks must prepare and make available to supervisors on request the
following for all trading desks:

inventory aging reports;

daily limit reports, including exposures, limit breaches and follow-up

action;

reports on intra-day limits and utilisation for banks with active intra-day
trading; and

reports on the assessment of market liquidity1389.

11.8.8 As foreign exchange and commodity positions held in the banking book are
required to be treated under the market risk framework (there is no bespoke
banking book treatment), such positions are treated as held on purely notional
trading desks within the trading book1390.

Requirement for a bank to use the models-based approach

11.8.9 At least 10% of a bank’s capital charge for market risk must be calculated in
accordance with the internal models approach for such an approach to be
allowed. This is calculated on a quarterly basis 1391. Securitisation positions
are not eligible for the internal models-based approach1392. Accordingly, the
standardised approach must be used.

Requirement to calculate the standardised approach capital charge for banks using
internal models

11.8.10 All banks that have supervisory approval to use the internal models-based
approach for one or more trading desks must:

calculate the standardised approach capital charge across all trading

desks, including those desks with model approval; and

calculate on a stand-alone basis the standardised approach capital charge

for each trading desk with model approval (with no netting allowed
between trading desks)1393.

11.8.11 According to the Basel Committee, these calculations will:

serve as an indication of the fall-back capital requirement for trading

desks that fail the eligibility criteria (e.g. because of backtesting results);

generate information on the capital outcomes of internal models both

across banks and relative to the standardised approach;

enable monitoring over time of the relative calibration of standardised

and model-based approaches; and

 Back to contents 227

 MARKET RISK

provide macro-prudential insight1394.

Qualitative standards

11.8.12 The following requirements apply:

the bank must have an independent risk control unit responsible for the
design and implementation of the market risk system, which must
produce and analyse daily reports;

the risk unit must carry out the profit and loss attribution and backtesting
tests (see below);

a separate unit at the bank must conduct initial and on-going validation
of all internal models on at least an annual basis;

the board of directors and senior management must be actively involved

in the risk control process, and devote appropriate resources to risk
control. The daily reports must be reviewed by a level of management
with sufficient seniority and authority to enforce reductions in positions;

the “core design elements” of the bank’s regulatory approved model and
any internal models used by the bank in day-to-day internal risk
management must be the same;

stress testing results must be reviewed at least monthly by senior

management, used in the bank’s internal assessment of capital adequacy
and be reflected in the policies and limits set by senior management and
the board of directors; and

various documentation requirements must be fulfilled 1395.

11.8.13 The bank’s internal models must cover the full set of positions within the
scope of the model1396, and on at least an annual basis be reviewed by internal
or external audit1397.

11.8.14 Internal models must be conceptually sound and adequately reflect all risks,
and must be revalidated periodically, particularly if there are significant
changes to the market, or the composition of the bank’s portfolio, making the
models no longer adequate. Specific standards apply 1398.

11.8.15 Banks are required to have rigorous and comprehensive stress testing
programmes at the trading desk, as well as at the bank-wide level1399. The
stress tests must cover a range of factors that can create extraordinary losses,
or make the control of risk in portfolios very difficult1400. Stress tests must
have quantitative and qualitative aspects, including plausible stress scenarios,
and the capacity of the bank to absorb potentially serious losses 1401. Stress
testing results must be communicated routinely to senior management and
periodically to the board1402.

 Back to contents 228

 MARKET RISK

Model requirements

11.8.16 The basic concept used by Basel III is that of “risk factors”. Risk factors are
“the market rates and prices that affect the value of the bank’s trading
positions”1403. The risk factors must be sufficient to represent the risks
inherent in the bank’s portfolio of on- and off-balance sheet items1404. The
internal model must include all risk factors used for pricing, and any omissions
from risk factors used for pricing, but not in the firm’s internal model, must
be justified to the satisfaction of the relevant supervisor1405. The requirement
to use all risk factors extends to all those factors specified under the
standardised approach, except for securitisation exposures which are
ineligible for the internal models approach, and are treated under the
standardised approach1406.

Risk factors

11.8.17 The risk factors specified under the standardised approach are summarised in
the section dealing with the standardised approach and are not repeated
here. Instead we concentrate on types of risk that must be taken into account
under an internal model.

11.8.18 Interest rate risk is the risk that a bank suffers losses as a result of a change
in interest rates on its portfolio of bonds and interest rate derivatives.
Interest rate risk is divided into two general sub-categories: general interest
rate risk and specific interest rate risk. General interest rate risk arises as a
result of a change in the general level of interest rates in a currency (e.g. the
Bank of England base rate). Specific interest rate risk arises as a result of
factors idiosyncratic to an issuer (e.g. an improvement or downgrading of the
bank’s assessment of the creditworthiness of a counterparty).

11.8.19 The capital charge for general market risk applies to government debt,
corporate bonds and derivatives that include interest rate risk.

11.8.20 For interest rate risk a bank must use a set of risk factors that corresponds to
the interest rates associated with each currency in which the bank has on- or
off-balance sheet exposures. The bank must model the yield curve, divided
into segments along the curve. Banks are required to model at least six risk
factors for material exposures in major currencies and markets 1407. If a bank
cannot model specific risk it is treated as unmodellable.

11.8.21 Equity risk is the risk that a bank may suffer losses as a result of movements
in the market prices of equities, or their derivatives, held in the trading book.
Banks are required to use risk factors that correspond to each of the equity
markets in which the bank holds significant positions. These factors must
reflect both market-wide movements in equity prices and may reflect
movements in prices of individual equities to which the bank is exposed. The
former is normally referred to as general equity market risk and the latter as
specific risk. Banks may use risk factors that reference specific industry
sectors (e.g. retail) as well as cyclical and non-cyclical factors. The
sophistication and nature of the modelling technique should correspond to the
bank’s exposure to the overall market, as well as the concentration on
individual equities1408.

 Back to contents 229

 MARKET RISK

11.8.22 Equity investments in funds where look-through is possible are treated as

exposures to each of the underlying exposures in the fund, and the bank must
assign the positions to the trading desk to which the fund is assigned. Where
look-though is not possible, but the bank has access to daily prices and the
mandate of the fund, then the standardised approach applies 1409. If not a
capital requirement equivalent to a deduction from capital applies.

11.8.23 Exchange rate risk is the risk of a bank suffering losses from movements in
exchange rates. Banks’ models must incorporate risk factors that correspond
to the individual foreign currencies in which a bank’s positions are
denominated. Foreign exchange risk arises on all exposures that are
denominated in a currency other than the bank’s reporting currency1410. FX
risk in banking book exposures is required to be treated under the market risk
framework.

11.8.24 Commodities risk is the risk of loss as a result of exposure to commodity price
changes (as well as their derivatives). Banks must utilise risk factors
corresponding to each of the commodity markets in which the bank holds
significant positions. The required sophistication of banks’ models depends
on the scale of their commodities trading, with those with an active trading
portfolio required to model for the convenience yield (e.g. the ability to profit
from temporary market shortages where the bank owns the commodities in
question) between forwards and cash positions 1411.

Modellable and non-modellable risk factors

11.8.25 Having identified the relevant risk factors for each asset class in accordance
with the above rules, banks must next determine whether the particular risk
factor is modellable or not. The capital treatment differs depending on
whether this is the case or not. This is a significant change from Basel II where
all risk factors were considered modellable. Under Basel III a risk factor is
only modellable if it satisfies the risk factor eligibility test (RFET). This test
requires a sufficient number of real prices representative of the risk factor.
At least one of the following must be satisfied for a price to be considered a
real price:

it is a price at which a bank has conducted a transaction;

it is a verifiable price from a transaction between arms’ length third

parties;

it is a price derived from a committed quote made by the bank or another

third party; or

it is a price from a third party vendor satisfying specified criteria1412.

11.8.26 The intention is to restrict reliance on internal models to those risk factors
for which the bank has sufficient actual prices, rather than relying on the
output of a model.

11.8.27 Either of the two following criteria must be met on a quarterly basis for a risk
factor to be considered as modellable:

 Back to contents 230

 MARKET RISK

the bank must identify at least 24 real price observations per year, with
no more than one price observation per day, and there must be no 90 day
period with fewer than four real price observations. These criteria must
be monitored and met on a monthly basis; or

the bank must identify for the relevant risk factor at least 100 real price
observations over the past 12 months, with no more than one price
observation per day1413.

11.8.28 Where real price data are obtained from a third party vendor, the vendor must
provide specified information and be subject to audit on its pricing
information1414.

11.8.29 If a risk factor is a point on a curve then in order to count real price
observations banks may either use their own internal risk bucketing approach,
or use a regulatory approach1415. In the former case each risk factor must be
allocated to a bucket1416.

11.8.30 Once a risk factor has passed the RFET test (above) banks are required to
ensure that the data used to calibrate their ES model satisfy specified
principles. Supervisors may also decide on a case-by-case basis that data are
unsuitable to calibrate the model and, if so, the risk factor is excluded from
the ES model1417. The following principles apply:

the data used may include combinations of risk factors;

the data used must allow the model to pick up both idiosyncratic and
general market risk. General market risk is the tendency of an
instrument’s value to change with the change of the broader market,
whereas idiosyncratic risk is the risk associated with a specific issuance
(i.e. specific and general equity and interest rate risk, supra). If both
are not captured by the bank’s model then the uncaptured risk factor is
treated as unmodellable (but the bank may model the other risk factor);

the data must allow the model to reflect volatility and correlation of the
risk positions;

the data must be reflective of prices observed and/or quoted in the

market, and if not derived from real price observations must be
reasonably representative of such prices;

data must be updated with sufficient frequency, preferably on a daily

basis, but at a minimum on a monthly basis;

the data used to determine the stressed ES calculation must be reflective

of the prices observed and/or quoted during the chosen stress period.
Any divergence from historical prices (for example, if the relevant
instrument differs in its characteristics from that available for the
relevant period) must be empirically justified; and

the use of proxies for data must be limited, and the chosen proxies must
have sufficiently similar characteristics to the transactions they
represent1418.

 Back to contents 231

 MARKET RISK

The capital adequacy calculation

11.8.31 According to Basel III, “[b]anks will have flexibility in devising the precise
nature of their expected shortfall (ES) models”, provided that the minimum
standards are adhered to1419. ES must be calculated on a daily basis for bank-
wide internal models used to calculate market risk capital requirements, as
well on a daily basis for each trading desk using the internal model1420.

11.8.32 ES is calculated using a 97.5% one tailed confidence level 1421. This means that
for each 200 liquidity horizons (which, as has been seen, vary under Basel III)
the model should fail to predict the actual loss no more than five times.

11.8.33 As seen above, a variable liquidity horizon replaces the 10 day holding period
under Basel II. However, this is achieved by calculating a base liquidity
horizon of 10 days and then scaling up this base result in accordance with a
formula1422. Hull explains as follows the calculation:

“In FRTB [Basel III], banks are required to consider changes over a period of
10 days that occurred during a stressed period in the past. Econometricians
naturally prefer that non-overlapping be used when VaR or ES is being
estimated using historical simulation, because they want observations on the
losses to be independent. However, this is not feasible when 10-day changes
are considered, because it would require a very long historical period. The
first simulation trial assumes that the percentage changes in all risk factors
over the next 10 days will be the same as their change between Day 0 and Day
10 of the stressed period; the second simulation trial assumes that the
percentage changes in all risk factors over the next 10 days will be the same
as their changes between Day 1 and Day 11 of the stressed period; and so on.

Banks are first required to calculate ES when 10-day changes are made to all
risk factors. (We will denote this by ES1). They are then required to calculate
ES when 10-day changes are made to all risk factors in category 2 and above
with risk factors in category 1 being kept constant. (We will denote this as
ES2). They are then required to calculate ES when 10-day changes are made
to all risk factors in categories 3, 4 and 5 with risk factors in categories 1 and
2 being kept constant. (We will denote this as ES 3). They are then required
to calculate ES when 10-day changes are made to all risk factors in categories
4 and 5 with risk factors in categories 1, 2 and 3 being kept constant. (We
will denote this as ES4). Finally, they are required to calculate ES 5, which is
the effect of making 10-day changes only to category 5 risk factors”1423.

11.8.34 The categories referred to above are the liquidity horizons of 10 days, 20 days,
40 days, 60 days and 120 days1424. According to Hull:

“[Basel III] represents a movement away from basing calculations on one-day

changes. Presumably, the Basel Committee has decided that, in spite of the
lack of independence of observations, a measure calculated from 10-day
changes provides more relevant information than a measure calculated from
one-day changes. This could be the case if changes on successive days are
not independent, but changes in successive 10-day periods can reasonably be
assumed to be independent”1425.

11.8.35 The author concludes:

 Back to contents 232

 MARKET RISK

“After 20 years of using VaR with a 10-day time horizon and 99% confidence
to determine market risk capital, regulators are switching to ES with a 97.5%
confidence level and varying time horizons. The time horizons, which can be
as high as 120 days, are designed to incorporate liquidity considerations into
the capital calculations. The change that is considered to a risk factor when
capital is calculated reflects movements in the risk factor over a period of
time equal to the liquidity horizon in stressed market conditions” 1426.

11.8.36 The liquidity horizon for each type of risk factor is set out in the table below.
Banks are required to map each of their chosen risk factors to the categories
set out in the table, and satisfy specified documentation and validation
requirements1427.

Liquidity horizon by risk factor

Risk factor Liquidity horizon in trading days
Interest rate (EUR, USD, GBP, 10
AUD, JPY, CAD and the domestic
currency of the bank)
Interest rate (other currencies) 20
Interest rate: volatility 60
Interest rate: other types of risk 60
factors
Credit spreads: sovereign – 20
investment grade
Credit spreads: sovereign – high- 40
yield
Credit spread: corporate – 40
investment grade
Credit spread: corporate – high- 60
yield
Credit spread: volatility 120
Credit spread: other types of risk 120
factor
Equity price: large cap 10
Equity: small cap 20
Equity: large cap-volatility 20
Equity: small cap-volatility 60
Equity: other types of risk factors 60
Foreign exchange: specified 10
currency pairs
Foreign exchange: other pairs 20
Foreign exchange: volatility 40
Energy and carbon emissions 20
prices

 Back to contents 233

 MARKET RISK

Liquidity horizon by risk factor

Risk factor Liquidity horizon in trading days
Precious and non-ferrous metals 20
Other commodities 60
Energy and carbon emissions 60
prices: volatility
Precious and non-ferrous metals: 60
volatility
Other commodities: volatility 120
Other commodities: other risk 120
factors

11.8.37 The maximum liquidity horizon is 120 days, which is a little less than six
months’ of trading days.

11.8.38 For the actual ES figure banks are required to update their data sets at least
every three months, and also reassess data sets when market prices are
subject to frequent changes1428 (e.g. during a crisis).

11.8.39 Where a bank hold options in its trading book then the ES model must
accurately capture the risks associated with interest rate risk, equity risk,
foreign exchange risk, commodity risk and credit risk. Further the non-linear
price characteristics of options, as well as volatilities associated with option
prices (vega risk), must be captured1429.

11.8.40 No specific type of model is prescribed, provided the minimum requirements

are met. Permissible models therefore include historical simulation, Monte
Carlo simulation, certain parametric approaches and other analytical
techniques1430. However, some models that could be used to calculate VaR
under Basel II would appear to be ruled out by the RFET requirement, or by
the nature of ES as a market risk measure. (While VaR can always in theory
be calculated, the same is not true of ES, as it takes into account losses
beyond the specified confidence level).

11.8.41 Additionally, banks must calculate a stressed ES figure. This replicates the ES
outcome that would be generated by the bank’s current portfolio if the
relevant risk factors were experiencing a stress. The calibration of this
stressed ES is based on a reduced set of risk factors selected by the bank that
are relevant for their current portfolio and for which there is a sufficiently
long data set of observations. The reduced risk factors must be approved by
the bank’s supervisor, meet the requirements for a risk factor to be
modellable, and the reduced set of risk factors chosen must explain at least
75% of the variation of the full ES model (i.e. all risk factors used to calculate
the non-stressed ES figure)1431.

11.8.42 The stressed ES figure is based on the most severe 12 month period of stress
available for the observation horizon of the bank 1432. This is the period over
which the bank experiences the greatest loss, spanning back to and including
20071433 (which is generally reckoned to be the start of the financial crisis).
The stressed figure must be updated at least quarterly, or whenever there are

 Back to contents 234

 MARKET RISK

material changes in the risk factors in the portfolio. The reduced set of risk
factors used to calculate the stressed ES must also be updated at the same
time1434.

Calculation of the capital requirement

11.8.43 Those trading desks that are approved to use an internal models approach
must calculate the ES using all risk factors that are deemed modellable.
Under this calculation the bank may recognise any cross-risk class
correlations. The bank must then calculate a series of partial ES capital
requirements (holding all other risk factors constant) for the range of risk
classes (interest rate risk, equity risk, foreign exchange risk, commodity risk
and credit risk spread). These partial and constrained ES values are summed
to provide an aggregate cross-risk class ES calculation1435.

11.8.44 The stressed ES is incorporated as follows. The calculation of the stressed ES

has already been explained above. The stressed ES calculated on the reduced
set of risk factors is scaled up by the ratio of (1) the current ES calculated
using all of the risk factors to (2) the current ES measure using the reduced
set of risk factors. This ratio is floored at one 1436 so the stressed ES figure sets
a floor to the ES calculation.

11.8.45 The calculation may be set out as follows:

𝐸𝑆𝐹,𝐶
𝐸𝑆 = 𝐸𝑆𝑅,𝐶 𝑥
𝐸𝑆𝑅,𝐶

Where:

ESR,C is the stressed ES figure calculated based on the restricted set of risk
factors.

ESF,C is the most recent ES figure based on the most recent 12 month
observation period with the full set of risk factors1437.

11.8.46 The aggregate capital requirement for modellable risk factors is calculated
based on the weighted average of the constrained and unconstrained ES
capital requirements, taking into consideration the stressed ES calculation
referred to above1438.

11.8.47 The actual capital calculation for those trading desks with model approval and
meet the various requirements summarised in the next section is equal to the
maximum of the most recent observation and a weighted average of the
previous 60 days scaled up by a multiplier1439. The multiplier is generally set
at 1.5, although national supervisors may set a higher level in defined cases 1440
(e.g. deficiencies in the model that do not justify model approval being
revoked).

Capital requirement for non-modellable risk factors

11.8.48 We have mentioned the existence of risk factors that cannot be modelled as
they do not satisfy the RFET test. Capital requirements for non-modellable
risk factors must be determined using a stress scenario that is calibrated to

 Back to contents 235

 MARKET RISK

be at least as prudent as the 97.5% ES calibration used for modellable risk

factors. In selecting the stress scenario the bank must determine a common
12 month period of stress across all non-modellable risk factors in the same
risk class. With supervisory approval the bank may calculate stress scenario
capital requirements at the bucket level for risk factors that belong to curves.
For each such non-modellable risk factor the bank is required to set a liquidity
horizon that is the greater than that given for modellable risk factors in the
table above and 20 days. Non-modellable risk factors arising from
idiosyncratic (i.e. issuer) credit spread risk banks may be based on a common
12 month stress period. Banks that cannot satisfy their supervisor that their
stress scenario is adequate have to use the maximum potential loss as the
stress scenario i.e. a total loss1441.

Capital requirement for market risk other than default risk

11.8.49 This is simple. It is the aggregate capital requirement for all approved and
eligible trading desks plus the capital charge for all trading desks calculated
under the standardised approach for those trading desks without model
approval, or which are ineligible to use their model 1442. If any trading desk is
in the “amber zone” (see below) then a capital surcharge applies1443.

11.8.50 The capital charge is then multiplied (as under the Basel II framework) by
12.51444 to integrate the final figure into the bank’s overall capital charge in a
manner consistent with the risk-asset ratio for credit risk (12.5 multiplied by
8% equals 1).

Backtesting

11.8.51 In order for a bank to be permitted to use an internal market risk model it
must also satisfy quantitative and qualitative requirements for backtesting
and the profit and loss account attribution test (PLA). These will be briefly
described below. Both are intended to ensure the integrity of the model as
well as its accuracy over time.

11.8.52 Basel III defines “backtesting” as “the process of comparing daily actual and
hypothetical profits and losses with model-generated VaR measures to assess
the conservatism of risk measurement systems”1445.

11.8.53 Backtesting must be performed at both the bank-wide and trading desk level.
The PLA test only applies at the trading desk level 1446. Backtesting therefore
relies on VaR methodology.

11.8.54 Backtesting commences on the date that a bank receives approval to use an
internal model for one or more trading desks 1447. The requirement is to
compare the VaR measure over a one day holding period against each of: (1)
the actual profit & loss (APL) and (2) the hypothetical profit and loss (HPL)
over the past 12 months1448.

11.8.55 The APL is defined as “the actual P&L derived from the daily P&L process. It
includes intraday trading as well as time effects and new and modified deals,
but excludes fees and commissions as well as valuation adjustments for which
separate regulatory capital approaches have been otherwise specified or
which are deducted from Common Equity Tier 1. Any other valuation

 Back to contents 236

 MARKET RISK

adjustments that are market risk-related must be included in the APL”1449.

Foreign exchange and commodity risk in the banking book is included in the
APL as there is no banking book treatment.

11.8.56 The HPL is defined as “the daily P&L produced by revaluing the positions held
at the end of the previous trading day using the market data at the end of the
current day. Commissions, fees, intraday trading and new/modified deals,
valuation adjustments for which separate regulatory capital approaches have
been otherwise specified as part of the rules and valuation adjustments which
are deducted from CET 1 are excluded from the HPL. Value adjustments
updated daily should usually be included in the HPL”1450.

11.8.57 As will be seen the main difference between the APL and HPL is the exclusion
from the latter of intraday trading and new or modified deals.

11.8.58 When applied at a bank-wide level the VaR model must be calibrated at a 99%
confidence level. An exception occurs if the actual loss or the hypothetical
loss (as described below) of the bank-wide trading book exceeds the
corresponding VaR measure given by the model 1451. If either the daily P&L or
the VaR is unable to be calculated then it is also treated as an exception 1452.
Exceptions may, however, be disregarded if they relate to a non-modellable
risk factor, the capital charge for that non-modellable risk factor is greater
than the actual or hypothetical loss, and the bank’s supervisor is notified and
does not object to disregarding the exception 1453.

11.8.59 The number of exceptions are grouped into green, amber and red zones, with
the number of exceptions increasing the multiplier referred to above. The
green zone comprises 0-4 exceptions. This comprises results that do not
suggest a problem with the quality or accuracy of a bank’s model. The
multiplier is 1.5. The amber zone comprises 5-9 exceptions. This
encompasses situations that raise questions as to the quality or accuracy of a
bank’s model. The multiplier varies from 1.7 to 1.92 depending on the
number of exceptions. The red zone consists of 10 or more exceptions and
indicates that there is almost certainly a problem with the bank’s risk model.
In this case the multiplier is 21454.

11.8.60 The number of exceptions is calculated over 250 trading days, corresponding
to a one year horizon1455. Where a bank’s model falls in the amber or red zone
the supervisor may revoke model approval if satisfied that the model is
inaccurate1456. Such revocation is not automatic even in the red zone as it is
still possible that the model is adequate, or can be improved to make it so.
As stated by the Basel Committee in 2016 in this case “the supervisor should
… begin investigating the reasons why the bank’s model produced such a large
number of misses, and should require the bank to begin work on improving its
model immediately”1457.

11.8.61 The Basel Committee explained further in January 2016 the rationale around
the three zones as follows:

“there is no threshold number of exceptions that yields both a low probability

of erroneously rejecting an accurate model and a low probability of
erroneously accepting all of the relevant inaccurate models. It is for this
reason that the Committee has rejected a single threshold.

 Back to contents 237

 MARKET RISK

Given these limitations, the Committee has classified outcomes for the
backtesting of the firm-wide model into three categories. In the first
category, the test results are consistent with an accurate model, and the
possibility of erroneously accepting an inaccurate model is low (green zone).
At the other extreme, the test results are extremely unlikely to have resulted
from an accurate model, and the probability of erroneously rejecting an
accurate model on this basis is remote (red zone). In between these two
cases, however, is a zone where the backtesting results could be consistent
with either accurate or inaccurate models, and the supervisor should
encourage a bank to present additional information about its model before
taking action (yellow zone)”1458.

11.8.62 As mentioned above, banks must also carry out backtesting at the trading desk
level. This is carried out daily 1459. Backtesting is effected at both the 97.5%
and 99% confidence level using each trading desk’s VaR over a one year period,
with all data equally weighted1460. An exception occurs when either the actual
or hypothetical loss of the trading desk exceeds the corresponding VaR
determined by the bank’s model. The same applies where the P&L, or other
risk factor, is not available or is impossible to compute.

11.8.63 If the number of exceptions exceeds 12 at the 99% confidence level, or 30 at

a 97.5% confidence level, in the most recent 12 month period the bank is no
longer able to use its model for that trading desk. Instead, the trading desk
must calculate its capital requirements under the standardised approach 1461.
However, “[t]he designation of being ineligible for internal modelling is not,
however, envisaged as being permanent. If P&L attribution and backtesting
performance sufficiently improved for a sufficient period of time, the
designation for the relevant internal models-based approach could be changed
from ineligible to eligible”1462.

The profit & loss attribution (PLA) test

11.8.64 This is the second aspect of ensuring enduring model validity. The PLA test
compares the daily risk-theoretical P&L (RTPL) with the daily hypothetical
P&L (HPL). Its purpose is to:

measure the materiality of simplifications in banks’ internal models used

for determining market risk capital requirements driven by missing risk
factors and differences in the way positions are valued compared with
front office systems; and

prevent banks from using internal models where such simplifications are
material1463.

11.8.65 The PLA is defined as “a method for assessing the robustness of banks’ risk
management models by comparing the risk-theoretical P&L predicted by
trading desk risk management models”1464.

11.8.66 The RTPL is defined as the daily trading desk P&L that is produced by the
valuation of the trading desk’s risk management model. The model must
include all risk factors that are included in the bank’s ES model, with
supervisory parameters and any risk factors deemed not modellable, but
disregarding risk factors not taken into account in the trading desk risk

 Back to contents 238

 MARKET RISK

management model1465. Time effects must be treated in a consistent manner

in both the HPL and RTPL1466.

11.8.67 The HPL must be identical to that used for backtesting purposes. The
comparison between the RTPL and HPL is intended to determine whether the
risk factors in the trading desk’s risk management model capture the material
drivers of the bank’s P&L by assessing if there is a significant degree of
association between the two P&L measurements over a suitable time
period1467.

11.8.68 As described in the (now replaced) 2016 Market Risk Standard “[t]he P&L
attribution is designed to identify whether a bank’s trading desk risk
management model includes a sufficient number of the risk factors that drive
the trading desk’s daily P&L. For the assessment, all of the instruments held
within a particular trading desk should be identified and considered as a
distinct portfolio. The risk factors for that portfolio that are included in the
desk’s risk management model must be used to calculate a ‘risk-theoretical’
P&L. This ‘risk-theoretical’ P&L is the P&L that would be produced by the
bank’s pricing models for the desk if they only included the risk factors used
in the risk management model. … This risk-theoretical P&L would be
compared to the hypothetical desk-level P&L, based on the mark-to-market
value of the trading desk’s instruments derived from the bank’s pricing models
including all risk factors. The risk-theoretical P&L used in P&L attribution
must not take into account any risk factors that the bank does not include in
its desk’s risk management model”1468.

11.8.69 The frequency and design of the test metrics were changed in 2019 after a
2018 consultation on problems with the original PLA test 1469. The 2018
consultation document explained: “the objective of the PLA test is to assess
the materiality of risks that may be missing from the risk management model
due to risk factors that are not included in the model or simplifications in the
model’s approach to valuation. Beyond these sources of discrepancy between
the HPL and RTPL of a given trading desk, additional differences between the
two measures of P&L may arise as the result of acknowledged differences or
misalignments of data that the bank uses as inputs to calculate each
measure”1470. These could include differences in time at which the data are
collected, and a bank using different data providers to source data 1471.

11.8.70 The calculation of the PLA is based on two test metrics 1472. These are:

the correlation between the two time series of P&Ls to assess the level
of dependence between the HPL and the RTPL (the Spearman correlation
metric). The metric separately ranks (from lowest to highest) the
historical 12 month series of daily HPL and RTPL values. A strong
correlation will only be observed if the rank ordering of values is closely
related between the two time series; and

an assessment of how “similar” the distributions of HPL and RTPL are over
time by calculating the maximum absolute difference between the
probability distributions of the HPL and RTPL over the time series (the
Kolmogorov-Smirnov metric)1473.

 Back to contents 239

 MARKET RISK

11.8.71 Based on the results of the application of these metrics, each trading desk is
allocated to either a green, red or amber zone 1474. Trading desks in the green
zone are eligible for the internal models approach to market risk. Trading
desks in the amber zone are subject to a capital surcharge 1475, whilst trading
desks in the red zone are disqualified from use of the internal models
approach, and must instead use the standardised approach for market risk
capital charges1476.

Default risk capital requirement

11.8.72 As has been mentioned above, Basel III requires banks to determine the
default risk capital requirement. The default risk model must meet the
general criteria and qualitative standards referred to above1477. Default risk is
defined as the risk of direct loss due to an obligor’s default as well as the
potential for indirect losses that may arise from a default event 1478. Default
risk models are based on VaR models. The specific requirements are:

the default simulation model must have two types of systemic risk
factors;

default correlations must be based on credit spreads or listed equity

prices based on 10 years of data that includes a stress period (which may,
in practice, mean going back to the financial crisis 15 years’ ago);

clear policies and procedures are required for the correlation calibration
process; and

the VaR calculation is based on a 99.9% one tailed confidence level 1479.
This allows only one excess in one out of 1000 holding periods.

11.8.73 All trading book positions not subject to the standardised approach must
satisfy the default risk capital requirement 1480. Sovereign exposures must be
included, and equity positions modelled based on a total loss 1481.

11.8.74 Due to the relationship between credit spread risk and default risk, banks
need approval from their supervisors for each trading desk to model default
risk. If this is not forthcoming, the trading desk is required to use the
standardised approach1482.

11.8.75 If the bank has an IRB approved model for credit risk (see chapter 5) then the
probability of default (PD) figures from that model must be used to calculate
default risk. If it does not, or if the supervisor deems the PD estimates to be
insufficiently robust1483, then the bank must compute PD figures using a
methodology consistent with the IRB approach to credit risk, as well as
satisfying certain specified conditions 1484. The same applies for loss given
default (LGD)1485.

11.8.76 The default capital requirement is the greater of:

the average model requirement measured over 12 weeks; or

the most recent model measure1486.

 Back to contents 240

 MARKET RISK

11.8.77 Default risk must be measured for each obligor. Probabilities of default have
to be corrected to obtain an objective probability of default. A footnote
states that market-implied probabilities of default are not acceptable.
Secondly, there is a floor of 0.03%1487. It is unclear to us on what basis
probabilities of default can be calculated other than based on market factors
as by definition the entity will not yet have defaulted. What other evidence
can there be than CDS spreads, equity, bond or subordinated debt prices, etc.
all of which are market-implied measures of default? Banks cannot be
required to have inside information from the entity’s management or auditors
as this would be illegal. How such “objective” measurements are to be
determined is therefore unclear.

11.8.78 Netting of long and short positions to the same obligor is allowed, although
differences in seniority of positions must be taken into account 1488. Basis risk
between long and short positions must be modelled explicitly1489. The model
must recognise the impact of correlations between defaults of obligors,
including the effect of stress on correlations. This reflects the increase in
correlations of defaults observed during the financial crisis. Specified
requirements for the determination of stressed correlations are set out
including calibrating correlations over a period of at least 10 years 1490.

11.8.79 A bank’s model must reflect the effect of issuer and market concentrations,
as well as concentrations that may arise within and across product classes
during a period of stress1491. This is another lesson of the financial crisis.

11.8.80 Derivatives with material non-linear behaviour in respect of default (e.g.

options) must be reflected in the model1492.

11.8.81 A bank must calculate the incremental loss relative to the current value that
the bank would incur in the event of a loss for each and every position subject
to the default capital risk model1493. Capital is required to be held against the
incremental loss from default in excess of any mark-to-market losses already
taken into account through the profit and loss account 1494. This is to prevent
double counting of losses.

11.8.82 Where a model cannot be used for any given trading desk, then the capital
charge will be determined using the standardised approach and then
aggregated across all such desks1495. This will be added to the capital charge
for default risk from all trading desks with model approval.

11.8.83 Basel III recognises that given the very high confidence level (99.9%) for
default risk models standard backtesting will not be possible. Accordingly,
indirect methods to validate model performance are required which may
include stress tests, scenario analysis and sensitivity analysis 1496.

11.9 Basel III - The Standardised Approach to Credit Risk

11.9.1 We now proceed to examine the standardised approach to market risk. It

bears no comparison to the former standardised approach under Basel II
(which is broadly retained subject, to a scaling factor, as the simplified
standardised approach). As with the internal models-based approach the
standardised approach has been successively refined through four consultative

 Back to contents 241

 MARKET RISK

documents, quantitative impact assessments and two standards, the current

being published in 2019.

11.9.2 As with the models-based approach we will describe the genesis and basis of
the new standard as set out in the relevant preparatory documents before
moving on to an account of the actual rules. Readers not interested in why
the Basel Committee sought such a fundamental change to the standardised
approach may skip the next section.

The first consultative document

11.9.3 As already mentioned, this was published in May 2012. The new standardised
approach is stated to serve two purposes: firstly, as a method for calculating
capital requirements for banks whose business models does not require a
sophisticated measure of market risk measurement and, secondly, as a fall-
back in case a trading desk is not approved for the internal models-based
approach, or it no longer qualifies for that approach 1497. The Basel Committee
set out the following five design principles for the new standardised approach:

improved risk sensitivity. The approach should give prudent recognition

of genuine hedging and diversification, and reduce the sensitivity gap
between the standardised and models-based approaches;

calibration. The standardised approach should be calibrated via clear,

logical and specified processes that reflect differences in asset volatility
and instrument type;

simplicity. Capital requirements should be based on a simple and

transparent methodology, placing a limited burden on banks and
regulators;

limited model reliance; and

being a credible fall-back. The capital charges must be reasonable in

terms of magnitude compared to those produced by a well-specified
internal model1498.

11.9.4 Whether the new standardised approach achieves all these objectives will be
a matter for readers of the full Basel III standard to decide.

11.9.5 Essentially, the Committee sought to address the risk that the threat of model
revocation was not credible under the 1996 Market Risk Amendment as the
capital charges under the standardised approach were so much higher that a
bank risked breaching its minimum capital requirement if model approval was
withdrawn.

11.9.6 The 2012 consultative document put forward two alternatives for
consideration: a partial risk factor approach and a fuller risk factor approach.
As the latter was subsequently discarded only the former will be considered
briefly in seeking to understand, on an intuitive basis, the new framework
which is highly mathematical and formula-driven in its final iteration as the
2019 Basel III standard. Under the partial risk factor approach there are three
steps:

 Back to contents 242

 MARKET RISK

assign all instruments in scope to prescribed “asset buckets” unless they

require decomposition;

a calculation of each “bucket’s” capital charge using supervisor-

determined risk weights and correlations; and

the aggregation of the “buckets” using a supervisory method to

determine the capital requirement 1499.

The second consultative document

11.9.7 This was published in October 2013 and confirmed adoption of the partial risk
factor approach1500. This document describes the standard ultimately adopted
by the Basel Committee as follows:

“the first step is to decompose instruments into ‘notional positions’. The

proposed Accord text sets out principles that should be followed by banks, as
well as detailed decompositions of commonly traded instruments. Notional
positions will then be allocated to risk buckets. It is expected that a bank
should be able to decompose all instruments for the purposes of the
standardised approach. However, if this is not possible, then supervisors will
have discretion to impose a prudent percentage of either the market value or
the notional of the position.

A set of risk buckets has been defined for each asset class. These risk buckets,
which have been designed based on a combination of statistical analysis and
expert judgment, group positions with similar risk characteristics together.
Notional positions are assigned to risk buckets according to certain categorical
variables, such as industry sector or credit quality. These buckets have been
defined following a statistical procedure combined with a judgmental overlay.
The maximum buckets in an asset class is 12. … If a notional position cannot
be allocated to any of the rick buckets in an asset class (for example, because
data on categorical variables is not available), it will be allocated to a
‘residual’ bucket for that asset class. Hedging and diversification benefits
between the residual bucket and other buckets in the asset class will not be
recognised …”1501.

11.9.8 The document explains further that “[a] single risk weight will apply to all
notional positions assigned to a risk bucket. In addition, at least two
regulatory-specified correlation parameters will be defined for each risk
bucket. One correlation parameter will be used where positions have the
same sign, to recognise diversification, and the other correlation parameter
will be used where positions have different signs, to recognise hedging. The
correlation parameter for positions with the same sign will be higher than that
where positions have different signs. This approach prudently captures the
risk to perceived hedging and diversification benefits that arises due to the
unstable and time-varying nature of correlation parameters, particularly in
times of stress”1502.

11.9.9 For general interest rate risk and credit spread risk cash flows were originally
proposed to be used as a starting point. Instruments would be decomposed
into their constituent cash flows, discounted, and then assigned on a
proportional basis to nearby maturity points. This proposal was abandoned in

 Back to contents 243

 MARKET RISK

the third consultative document (see below). Having determined the risk
measure for each bucket or currency, offsetting and diversification would then
be recognised across different buckets to determine the standardised capital
requirement for a given asset class using a cross-bucket aggregation
formula1503.

11.9.10 The approach for non-linear financial instruments (e.g. options) involves the
allocation of a delta equivalent position in the underlying instrument, which
is equal to the delta of the financial instrument multiplied by the underlying
notional amount1504. Non-delta risks (i.e. risks that are not driven by a change
in the price of the underlying, but the risk characteristics of the instrument)
are captured through a delta-stripped scenario matrix approach (vega risk)1505.

11.9.11 The trading book standardised capital requirement is the sum of standardised
requirements across all asset classes, with no recognition of diversification or
offsetting between asset classes1506.

The third consultative document

11.9.12 This document was published in December 2014. Banks objected to the
proposed cash flow model outlined above, due to a lack of stored data and
the technical requirement to calculate separate discount curves for each
instrument1507. As a result, the Basel Committee settled on a sensitivity-based
approach which requires banks to use price and rate sensitivities, thereby
reducing the implementation cost for banks1508.

11.9.13 According to the third consultative document (which is the foundation of the
new standardised approach):

“a sensitivity-based approach entails reliance on the pricing model of firms.

This may improve the risk sensitivity of the standardised approach, but comes
at a cost to simplicity and consistency. However, recognising the industry
concerns above to be valid, the Committee views the cash flow-based method
as even more complex and impractical to design or compute by comparison,
without offering any clear advantages in terms of risk capture or
comparability. Specifically, the technical challenges involved in prescribing
the cash flow of complex instruments were prohibitive in terms of enhancing
the proposed standardised approach as a functional fallback to the internal
models-based approach”1509.

11.9.14 However, the basic design features of the second consultative document were
retained. The framework would capture both delta and non-delta risks for
both non-securitisation and securitisation positions (securitisations are
excluded from the internal models-based approach), as well as equity risk,
foreign exchange risk and commodity risk. Capital charges are computed at
an asset class level, with no recognition of diversification effects across
different asset classes1510. For options (and positions featuring optionality)
both vega risk (the sensitivity of the value of an option to a change in
volatility) and curvature risk (which measures the rate of change of delta risk)
are required to be modelled1511. The Committee also decided to capture “basis
risk” within the standardised approach i.e. the risk that the relationship
between the prices of correlated instruments weakens over time 1512, through

 Back to contents 244

 MARKET RISK

a “correlation method” as more sensitive to interest rate and foreign

exchange risk1513.

The January 2016 market risk standard

11.9.15 This document published the Basel Committee’s original market risk standard
(although it was replaced in 2019 following further consultation and input
from the banking industry). This set out a tripartite capital charge for market
risk measured under the standardised approach. This may be summarised as
follows1514:

a sensitivities-based method for capital charges based on delta, vega and

curvature risk within a prescribed set of risk classes;

a default risk capital charge with three elements: non-securitisation,

securitisation and the correlation trading portfolio within securitisation
exposures; and

a residual add-on applied to the notional amounts of instruments that are

non-linear1515.

11.9.16 The standard explained that the default risk charge “is calibrated to the credit
risk treatment in the banking book to reduce the potential discrepancy in
capital requirements for similar risk exposures across the banking book and
the trading book”1516. This might be seen as a “belt and braces” approach
given the prohibition on recognition of regulatory capital benefits from
switching positions between both books under Basel III, but is consistent with
the Basel 2.5 reforms. Whether it is necessary to impose the same capital
treatment on banking and trading book positions given their different
characteristics for default risk may be questioned as default risk should
usually be small in trading book positions, as the firm should be able to trade
out of the position, or hedge it, or if not the position should not be held in
the trading book.

11.9.17 The residual add-on was “introduced to capture any other risks beyond the
main risk factors already captured in the sensitivities-based method and the
Default Risk Charge. It provides for a simple and conservative capital
treatment for the universe of more sophisticated trading book instruments for
which the Committee has refrained from detailed speculation under the
standardised approach, so as to limit excessive risk-taking and regulatory
arbitrage incentives”1517. Sed quaere.

The fourth consultative document

11.9.18 This was published in March 2018 and followed dissatisfaction with the
ostensibly definitive 2016 standard. This document states that the
sensitivities-based method specifies:

the risk weights that should be applied to the sensitivities for each of the
prescribed list of risk factors. Banks are required to multiply their
sensitivities to risk factors by supervisory risk weights to estimate the
change, on a risk factor by risk factor basis, the value of their trading
book portfolios; and

 Back to contents 245

 MARKET RISK

capital requirements for risk factors are aggregated using a set of

formulae using defined correlation assumptions specifying diversification
benefits across risk factors1518.

11.9.19 Based on evidence provided by banks in the quantitative impact assessment,

the Basel Committee identified deficiencies in the following elements of the
2016 standard:

the approach to determine foreign exchange pairs that are liquid;

the correlation scenarios applied in the capital calculations; and

the treatment of non-linear instruments such as options1519.

11.9.20 The changes proposed were as follows:

allowing combinations of currency pairs to benefit from a lower capital

requirement;

adjustments to the capital charge for risk factors judged to by highly

correlated; and

a request for further information from the banking industry on possible

double counting of curvature risk in respect of foreign exchange risk 1520.

11.9.21 Also, based on bank data, the Basel Committee concluded that the capital
charges for certain segments of the standardised approach were too high, and
reduced them for general interest rate risk by 20-40%, and for equity and
foreign exchange risk by 25-50%1521. Amendments were also proposed for
multi-underlying options and index instruments1522.

The 2019 market risk standard

11.9.22 The final results of the Committee’s deliberations on the new Market Risk
Standard were published in 2019. According to the 2019 Explanatory Note
published by the Committee:

“The framework specifies:

 A set of risk factors which are considered to be the main market variables
that affect the value of banks’ trading portfolios. Similar risk factors are
grouped together into “buckets” (eg for equities, buckets are defined by
industrial sector). Banks calculate the sensitivity of their trading book
portfolio to movements in the value of each of the risk factors.

 Risk weights to be applied to those risk factors. Risk weights have been
calibrated to stressed market conditions to ensure a calibration aligned
with the internal models approach. Banks must scale up their
‘sensitivities’ to each risk factor based on the prescribed risk weight to
estimate how much value the portfolio would lose if a shock was to
happen to the risk factor.

 Back to contents 246

 MARKET RISK

 A methodology for aggregating the losses calculated for each risk factor
shock to determine the loss for the scenario at the portfolio level. In
order to ensure a level of risk sensitivity, the aggregation method
recognises a degree of diversification benefit between risk factor-level
losses (applying different levels of assumed correlation between shocks
applied to risk factors in the same buckets and those in different
buckets)”1523.

11.9.23 The Committee adds:

“The above steps are applied separately for three different types of risk and
added as a simple sum to calculate the sensitivities-based method capital
requirement:

 Delta risk – the potential loss due to a small change in price of an equity
or commodity, or a small change in an interest rate, credit spread, or FX
rate.

 Vega risk – the potential loss due to a change in the implied volatility of
an option (for instruments that feature optionality).

 Curvature risk – the potential incremental loss beyond delta risk when
large movements occur in risk factors of instruments that feature
optionality”1524.

11.9.24 In terms of the matters consulted on in 2018 the final determination of the
Basel Committee was as follows:

 “Under the FX risk class, the scope of currency pairs that are considered
liquid, and are therefore subject to lower risk weights, has been
broadened. The overall approach to FX risk has also been amended so
that banks, subject to supervisory approval, may calculate FX risk with
respect to the currency in which they manage their trading business
(their “base currency”) rather than with respect to their reporting
currency. …

 The equity risk and the credit spread risk classes have been enhanced,
with new ‘index’ buckets for equity and credit spread risks introduced to
provide a simple approach that does not require the identification of
each underlying position in an index to calculate the capital requirements
for equity and credit indices.

 The calculation of curvature risk capital requirements for options has

been modified to (i) apply consistent shocks to similar risk factors; (ii)
address double-counting of FX risk for certain instruments; and (iii)
remove a potential cliff effect in the aggregation formula for capital
requirements. The scope of the curvature risk calculation has been
broadened to allow banks to include bonds and other instruments without
optionality when curvature risk is managed holistically across options and
other instruments.

 With regard to the aggregation of risk sensitivities, the ‘low correlations’

scenario has been modified to ensure it does not produce unrealistically

 Back to contents 247

 MARKET RISK

low correlations for risk factors that are considered to be highly

correlated in stressed market conditions”1525.

11.9.25 With this introduction we proceed to describe the new standardised approach
to market risk.

11.10 Basel III - Final Standardised Approach to Market Risk

11.10.1 All banks, except those allowed to use the simplified standardised approach,
must calculate their capital requirements under the standardised approach 1526.
This includes banks with model approval for one or more trading desks 1527.

11.10.2 All banks must also use the standardised approach for securitisation positions
and equity investments in funds that cannot be looked through to the
underlying1528.

11.10.3 Under the final Basel III standard the standardised approach is the sum of
three components: the capital requirement calculated under the sensitivities-
based approach, the default risk capital requirement and a residual add-on1529.
As foreshadowed above, the sensitivities-based capital charge is the sum of
three distinct elements: delta risk (measuring sensitivities to regulatory delta
risk factors), vega risk (based on sensitivities to regulatory vega risk factors)
and curvature risk (which captures the incremental risk not covered by the
delta measure for price changes in an option) 1530. The three former risk
measurements are aggregated using specified correlation parameters to
address diversification benefits between risk factors. Three sensitivities-
based risk scenarios must be calculated to address the risk that correlations
may increase or decrease in periods of stress1531.

11.10.4 The default risk capital requirement is designed to capture the jump-to-
default risk for instruments subject to credit risk, calibrated on the capital
charge in the banking book to reduce the risk of regulatory arbitrage between
the two books. Some hedging is allowed1532.

11.10.5 The add-on is “introduced to ensure sufficient coverage of market risk for
instruments with an exotic underlying and other instruments bearing residual
risks”1533.

11.10.6 As has been seen, the sensitivities-based method consists of a prescribed list
of risk factors in respect of which a bank is required to determine the delta,
vega and curvature risk capital requirements. These are then aggregated,
firstly within risk buckets (risk factors with common characteristics) and then
across risk buckets within the same risk class as defined below 1534.

11.10.7 For the purposes of applying the standardised approach the following
definitions apply:

A “risk factor” is a variable that affects the value of a trading book

instrument (e.g. the price of an equity, or the interest rate on a bond)1535.
This is described in more detail below.

The “delta risk” is the linear estimate of the change in the value of a
financial instrument due to the movement in a risk factor. The risk factor

 Back to contents 248

 MARKET RISK

could be the price of an equity or commodity, or a change in an interest

rate, credit spread or foreign exchange rate 1536.

The “vega risk” represents the potential loss resulting from the change
in value of a derivative due to a change in the value of a derivative as a
result of a change in the implied volatility of its underlying1537.

“Curvature risk” is the additional potential loss beyond delta risk as a

result of a change in a risk factor for financial instruments with
optionality (which is not restricted to options). This is calculated based
on two stress scenarios involving an upward shock and a downward shock
to each regulatory risk factor1538. In this respect the curvature risk
calculation differs from delta and vega risk.

A “risk bucket” is a set of risk factors that are grouped together by

common characteristics (e.g. all interest rates in a common currency) 1539.
The definitions of each risk bucket is set out in the text of Basel III and
is examined briefly below.

There are seven risk classes under the standardised approach:

general interest-rate risk (GIRR);

credit spread risk (CSR) for non-securitisation positions;

credit spread risk (CSR) for securitisation positions;

credit spread risk (CSR) for securitisation positions included in the

correlation trading portfolio (see below);

equity risk;

commodity risk; and

foreign exchange (FX) risk1540.

A “risk position” is the portion of the risk of an instrument that relates

to a risk factor. The standardised approach sets out how to calculate the
delta, vega and curvature risk1541.

The “risk capital requirement” is the amount of capital a bank is required

to hold under the standardised approach calculated as an aggregation of
risk positions, first at the bucket level and then across risk buckets within
a risk class as set out in the Basel III text1542.

The correlation trading portfolio

11.10.8 The correlation trading portfolio has been mentioned above and is a subset of
securitisation positions held in a bank’s trading book that meet certain
requirements and benefit from a more lenient capital treatment. To fall
within the securitisation correlation portfolio a position must satisfy either
one of the two following requirements (denoted under (1) and (2)):

 Back to contents 249

 MARKET RISK

1(a) the securitisation position is not a re-securitisation position, nor a

derivative of a securitisation position that does not provide a pro rata
share in the proceeds of a securitisation tranche;

1(b) all reference entities are single-name products, including single-name

credit derivatives, for which a liquid two-way market exists, including
traded indices on these reference entities;

1(c) the instrument does not reference an underlying that is treated as a

retail exposure, a residential mortgage exposure, or a commercial
exposure; and

1(d) the instrument does not reference a claim on a special purpose entity;
or

(2) the instrument is a non-securitisation hedge to a position as described

above1543.

11.10.9 A two-way market is deemed to exist where there are bona fide offers to buy
and sell so that a price reasonably related to the last sales price or to a current
bona fide competitive quote exists1544.

Application of the sensitivities-based method

11.10.10 All instruments subject to the sensitivities-based method (i.e. all trading book
positions other than exotic derivatives) are subject to delta risk1545. Exotic
derivatives include longevity swaps, weather derivatives, derivatives based
on market volatilities, etc.1546.

11.10.11 The following instruments are additionally subject to vega risk and curvature
risk:

any instrument with optionality. Optionality obviously includes all

options. In addition, instruments with an embedded option, such as
convertibility into another instrument, or rate-dependent prepayment is
covered. This includes calls, puts, caps, floors, swaptions, barrier
options and exotic options;

any instrument with an embedded prepayment option. This is a debt

instrument that grants the debtor the right to prepay part or all of the
principal amount before the contractual maturity date without being
obliged to compensate the lender/holder for any foregone interest
(which is quite common). Such options are subject to vega and curvature
risk in respect of interest rate risk and CSR (either non-securitisation or
securitisation, as the case may be). If the prepayment option is, in
addition, a behavioural option, then there may be a regulatory risk add-
on (RRAO). This is considered further below;

instruments whose cash flows cannot be written as a linear function of

the underlying notional instrument. An example is a plain-vanilla option
where the cash flows are not linear as they are the maximum of the spot
and strike price1547.

 Back to contents 250

 MARKET RISK

11.10.12 Instruments that do not have optionality, but which the bank manages
holistically with instruments with optionality may be subjected to curvature
risk at the bank’s discretion. However, in this case, the bank must use this
approach consistently over time1548.

Calculation of the delta and vega capital requirements for each risk class

11.10.13 For each risk class a bank must determine its positions to a set of prescribed
risk factors, risk weight those sensitivities, and aggregate the resulting risk
weights separately for delta and vega risk using the following step-by-step
approach:

for each risk factor the Basel III text specifies a sensitivity;

sensitivities to each risk factor must be netted to give a net sensitivity

across all instruments in the portfolio. Positions in opposite directions
are offset;

the weighted sensitivity is the product of the net sensitivity and the risk
weight defined in the Basel III text;

when determining aggregation within each risk bucket the risk position
for both delta and vega risk is determined by aggregating the weighted
sensitivities to risk factors within that bucket using a prescribed formula;
and

in determining across risk bucket positions, the delta and vega risk
capital requirement is determined by aggregating positions for delta and
vega risk using a different formula 1549.

Curvature risk

11.10.14 For each risk factor curvature risk capital requirements are calculated based
on an upward shock and a downward shock to each prescribed risk factor
based on a calculation of incremental loss for instruments sensitive to that
risk factor beyond that captured by the delta risk capital requirement. The
size of the shock is specified in the Basel III text. If the price of an instrument
depends on several risk factors, then curvature risk must be determined
separately for each risk factor. The net curvature risk capital requirement is
determined through a prescribed formula. For within risk bucket aggregation
a correlation formula applies. Curvature risk positions must then be
aggregated across all buckets in each risk class. Again, a formula applies 1550.

Aggregation of risks

11.10.15 To address the risk that correlations may increase or decrease during a period
of financial stress the aggregation of bucket-level capital requirements, and
risk class capital requirements for delta, vega and curvature risk have to be
repeated according to three different scenarios for the correlation between
risk factors within a risk bucket and the correlation across risk buckets within
a risk class. These are:

the “medium” correlations scenario;

 Back to contents 251

 MARKET RISK

the “high” correlations scenario; and

the “low” correlations scenario1551.

11.10.16 Unsurprisingly, the general prescribed correlation parameters are modified to

reflect the possible performance of the correlations in a situation of financial
stress1552.

11.10.17 The total capital requirement under the sensitivities-based method is

aggregated as follows:

for each of the three correlation scenarios (see preceding paragraph) the
bank sums up the delta, vega and curvature capital requirements for that
scenario; and

the actual capital requirement is the largest of the three scenario capital
requirements1553.

11.10.18 We will now proceed to consider the relevant risk factors.

Risk factors for delta, vega and curvature risk

11.10.19 The following risk factors are relevant: general interest rate risk (GIRR), credit
sensitivity risk (non-securitisation), credit sensitivity risk (securitisation),
credit sensitivity risk (securitisation: correlation trading portfolio), equity
risk, commodity risk and foreign exchange risk. These will now be examined
in a little more detail.

GIRR. The delta risk factors are defined along two dimensions: (1) a risk-
free1554 yield curve for each currency and (2) the following
tenors/maturities of the debt instrument: 0.25 years, 0.5 years, one year,
two years, three years, five years, ten years, 15 years, 20 years and 30
years, to which each risk factor must be assigned. The delta risk factors
include a flat curve of market-implied inflation rates with the term
structure of interest rates not included as a risk factor. All inflation risks
must be aggregated for a single currency to one number via a simple sum.
The GIRR delta risk factors also include one of two possible cross-
currency basis risk factors for each currency (each GIRR bucket), with
the term-structure of interest rates not recognised. The two recognised
currencies are the USD and EUR. Other cross-currency bases are
computed on the basis of the USD or EUR (but not both) 1555.

The vega GIRR for each currency is the implied volatility of options
defined using two dimensions:

the maturity of the option mapped to one or several of the

following: 0.5 years, one year, three years, five years and ten
years1556; and

the residual maturity of the underlying of the option at the expiry

date of the option for the same set of tenors1557.

 Back to contents 252

 MARKET RISK

Curvature GIRR risk factors are defined along one dimension: the
constructed risk-free yield curve per currency disregarding the term
structure in interest rates. Inflation and cross-currency basis risk are
ignored1558.

CSR (non securitisation). Delta risk factors are defined along two
dimensions:

The issuer credit spread curves (bonds and CDS); and

Maturities of 0.5 years, one year, three years, five years and ten
years1559.

The vega risk factors are the implied volatilities of options that reference
the relevant issuer names as the underlying (bonds, credit default swaps)
based on the maturity of the option. This is defined in Basel III as the
implied volatility of the option defined at the same maturities as for
delta1560.

Curvature risk factors are defined by the relevant issuer credit spread
curves1561. According to Hull1562:

“The curvature risk charge is a capital charge for a bank’s gamma risk
exposure under the standardized approach. Consider the exposure of
a portfolio to the ith risk factor. Banks are required to test the effect
of increasing and decreasing the risk factor by its risk weight Wi. If the
portfolio is linearly dependent on the risk factor, the impact of an
increase in Wi in the risk factor is Wiδi. Similarly, the impact of a
decrease in Wi in the risk factor is –δiWi. To evaluate the curvature net
of the delta effect, the standardized therefore calculates

1. Wiδi minus the impact of a increase of Wi in the risk factor; and

2. – Wiδi minus the impact of a decrease in the risk factor of Wi.

The curvature risk charge for the risk factor is the greater of these
two”.

And:

“The curvature risk charges for different risk factors are combined to
determine a total curvature risk charge. When diversification benefits
are allowed, aggregation formulas broadly similar to those used for
deltas are used with correlations specified by the Basel Committee” 1563.

CSR (securitisation). Delta risk factors are calculated with respect to

the spread of the tranche rather the underlying instruments. The
following maturities are relevant: 0.5 years, one year, three years, five
years and ten years1564.

Vega risk factors are implied volatilities of options that represent credit
spreads as underlyings (bond and CDS) based on the same maturities as
for the delta risk factors1565.

 Back to contents 253

 MARKET RISK

The curvature risk factors are defined by the relevant tranche credit
spread curves1566.

CSR (securitisation: correlation trading portfolio). Delta risk factors

are computed with respect to names underlying the securitisation or nth-
to-default1567. Delta risk factors are defined by reference to the
underlying credit spread curve and specified maturities (the same as for
other securitisation positions)1568.

Vega risk factors are the implied volatilities of options that reference
correlation trading portfolios credit spreads as underlyings based on the
maturity of the option with the same maturities 1569.

The curvature risk factors for this portfolio are the underlying credit rate
spreads1570.

Equity risk factors. For delta risk the risk factors are the spot and repo
rates1571.

For vega risk, the risk factors are the implied volatilities of options
referencing the equity spot price based on the maturity of the option
over maturities of 0.5 years, one year, three years, five years and ten
years. There is no vega risk capital requirement for equity repos 1572.

The equity curvature risk factors are all the equity spot prices, with no
curvature risk capital requirement for repos 1573.

Commodity risk factors. The commodity delta risk factors are generally
commodity spot prices (although sometimes for certain commodities the
forward price can be used). Two dimensions are defined: the legal terms
on place of delivery of the commodity1574, and the time to maturity of the
traded instrument calculated based on 0 years (spot), 0.25 years, 0.5
years, one year, two years, three years, five years, ten years, 15 years,
20 years and 30 years1575.

Vega risk factors are the implied volatilities of options referencing

commodity spot prices mapped to the following maturities: 0.5 years,
one year, three years, five years and ten years 1576.

Curvature risk factors are defined based on the constructed curve

(without a term structure) of commodity spot prices 1577.

FX risk factors. The delta FX risk factors are all the exchange rates
between the bank’s reporting currency and the currency in which the
exposure is denominated, with the possibility (with supervisory approval)
of relying on a base currency instead of the reporting currency) 1578.

Vega FX risk factors are the implied volatilities of options referencing

exchange rates between currency pairs based on specified maturities 1579.

Curvature FX risk factors are all the exchange rates between the currency
in which an instrument is denominated and the reporting currency,

 Back to contents 254

 MARKET RISK

although, with supervisory approval, a base currency may be used as

described above in relation to delta risk1580.

No distinction is required for all FX risk factors between onshore and

offshore variants of a currency 1581, although there may be significant
differences in market price1582.

Calculating the capital charge

11.10.20 For each risk factor (as defined above) sensitivities are calculated on the basis
of a change in the market value of the instrument as a result of applying a
specified shift to each risk factor, assuming all other relevant risk factors are
held at the current level1583.

11.10.21 Banks are therefore required to use a pricing model to apply the sensitivities-
based approach. If they cannot then the fall-back simplified standardised
approach may be available (at supervisory discretion). The pricing model
must be used by an independent risk control unit to report to senior
management.

11.10.22 For each asset class, the Basel III test sets out how to calculate delta risk 1584.
The precise formulae will not be described here. The basic idea is to assess
the effect of a small shift in the spot price on the market value of the
instrument. For instruments subject to vega risk (e.g. options) the vega is
multiplied by the implied volatility of the option or other instrument 1585. There
is a specific treatment for options without a maturity, options without a strike
or barrier, options with multiple strikes and barriers, and correlation trading
securitisation positions which do not have an implied volatility 1586.

11.10.23 Index instruments, and multiple underlying options should generally be

subjected to a “look through” approach1587. The same, generally, applies to
funds1588. Positions that cannot be “looked through”, or where the bank does
not have access to daily price quotes and knowledge of the mandate, are
excluded from the trading book, and instead attract a 100% capital
requirement1589 (equivalent to a deduction from capital).

11.10.24 The Basel III text sets out at length the supervisory risk buckets, risk weights
and correlation factors banks are required to use in respect of delta risk, vega
risk and the calculation of curvature risk, including many formulae 1590. It is
outside the scope of this chapter to go into the detailed requirements.

Default risk capital requirement (DRC)

11.10.25 As was mentioned at the outset, when considering the standardised approach
to market risk there are three components: the sensitivities-based approach
(summarised above), the default risk capital requirement and the residual risk
add-on or RRAO. The latter two will now be considered.

11.10.26 The DRC capital charge is intended to capture the jump-to-default (JTD) risk
that may not be captured by credit spread shocks under the sensitivities-based
method explained above. The DRC capital charges provide some limited
recognition of hedging 1591.

 Back to contents 255

 MARKET RISK

11.10.27 The DRC is calculated by reference to all trading book instruments that are
subject to default risk. This constitutes:

non-securitisation portfolios;

securitisation portfolios; and

securitisation positions held within the correlation trading portfolio 1592.

11.10.28 In calculating the DRC under the standardised approach the following steps
must be taken:

the gross jump-to-default (JTD) risk is calculated separately for each

exposure;

for each exposure to each obligor long and short positions are netted to
produce a single net long or short position;

each net JTD risk position is then allocated to a supervisory bucket;

within each bucket a hedge-benefit ratio is calculated using net long and
net short JTD positions. This is a discount factor that reduces the amount
of net short positions that can be netted against net long positions within
each risk bucket. Once this is done, a prescribed risk weight is applied;
and

the DRC is the sum of all DRC capital requirements across all risk
buckets1593.

11.10.29 No diversification benefit is recognised between the DRC calculation for: (1)
non-securitisations, (2) securitisations and (3) securitisations recognised
within the correlation trading portfolio1594.

11.10.30 At national discretion, claims on sovereigns, public sector entities that

present sovereign risk and multilateral development banks may be excluded
from the DRC calculation. The criteria are the same as under the standardised
approach for credit risk discussed in chapter 4.

11.10.31 We will now examine the DRC framework in more detail.

Gross jump-to-default (JTD) positions

11.10.32 This is calculated on an exposure-by-exposure basis. So a long exposure to

IBM shares and a short exposure to the same issuer are separate exposures 1595.
A long position is a position that exposes the bank to loss in the event of a
default. For derivatives, again the question is whether the bank will suffer a
loss in the event of a default. Thus, a sold put option, or CDS, is treated as a
long position1596. The gross JTD calculation is a function of the loss given
default of the position and the cumulative P&L figure for that exposure 1597.
This is entirely logical, as otherwise losses already incurred through the P&L
would not be taken into account when calculating the JTD, thereby
exaggerating the maximum potential loss should the obligor default. For

 Back to contents 256

 MARKET RISK

instruments that have incurred a total loss through the P&L there is no
separate DCR as the instrument has been written off as a total loss.

11.10.33 The loss-given-default under the standardised approach is based on

supervisory parameters. These are as follows:

equities and subordinated debt instruments: 100%;

senior debt: 75%; and

covered bonds: 25%1598.

11.10.34 The notional amount is used to determine the loss of principal at default, and
the mark-to-market figure to determine the net loss (to avoid double
counting, see above). The following specifications apply. For a bond, the
notional is the face value. For a credit derivative or put option it is the
notional amount of the contract. For a call option the JTD is zero as the call
option would not be exercised in a default, and the loss will instead be
recognised through the P&L account 1599.

11.10.35 As the JTD is calculated over a one year horizon, exposures with a shorter
maturity are scaled down to a fraction of a year. Exposures over one year are
not scaled up1600. For exposures with a maturity of under three months, there
is a three month floor1601.

11.10.36 Once the gross positions have been determined, netting applies. The
following rules are applicable. Long and short positions are netted provided
the short position has the same or lower seniority to the long position. Thus
a short position in an equity can be netted with a long position in a bond, but
not vice versa. For guarantees, the Basel II, framework applies (quaere if this
is a mistake). In cases of maturity mismatches, where the exposure is longer
than one year the mismatch is ignored. In other cases a time-weighted
approach applies1602.

11.10.37 For the DCR capital charge for all non-securitisation positions there are three
risk buckets: (1) corporates, (2) sovereigns and (3) local governments and
municipalities1603. Hedging within buckets is recognised as follows. The
hedging benefit is the ratio of net long JTD positions to net short JTD
positions, and the absolute value of net short JTD positions (based on actual
and not risk-weighted values)1604.

11.10.38 In calculating the risk-weighted JTD, default risk weights are set based on the
credit quality of the obligor. The following risk weights apply:

Default risk weights for non-securitisations by credit quality category

Credit quality Risk weight
AAA 0.5%
AA 2%
A 3%
BBB 6%

 Back to contents 257

 MARKET RISK

Default risk weights for non-securitisations by credit quality category

Credit quality Risk weight
BB 15%
B 30%
CCC 50%
Unrated 15%
Default 100%

11.10.39 The capital requirement for each bucket is determined as the sum of the risk-
weighted net long JTD, the hedge benefit ratio (see above) and the sum of
risk-weighted short net JTD positions, calculated across the above credit
weighting bands1605. No alternative is provided for jurisdictions that do not
allow the use of credit rating agency ratings, unlike the standardised approach
to credit risk.

11.10.40 No hedging is allowed between different buckets, and the DRC capital
requirement for non-securitisations is the simple sum of the bucket level
capital requirements.

11.10.41 The default capital requirement for securitisations is determined as follows.

When calculating the gross JTD on securitisation positions, exactly the same
procedure is followed as set out above, except no loss-given-default (LGD)
figure is applied. The reason is that the LGD is already captured in the
default-risk weights, so this is necessary to avoid double counting. The JTD
is therefore simply the market value of the tranche1606. When it comes to
netting and hedging, the underlying names, or a non-tranched index position,
may be decomposed into the equivalent tranches that cover the entire
tranche structure1607.

11.10.42 The net JTD position is restricted to specific securitisation exposures (i.e.
tranches with the same underlying pool of exposures). No netting is permitted
between different tranches even if they have the same attachment and
detachment points. Instead of relying on credit quality (as applies to non-
securitisation positions) the following treatment is mandated. The default
risk weights are based on the banking book rules for securitisation exposures,
with an assumed maturity of one year to avoid double counting as credit
migration risk in the trading book is already captured in the credit spread
capital requirement. Individual cash securitisation positions can be capped
at the fair value of the transaction 1608. No hedging is recognised between
different buckets1609.

11.10.43 For securitisation positions held within the correlation trading portfolio the
following rules apply. The gross JTD figure is the same as for all other
securitisation positions1610. For single-name and index hedges the gross JTD is
the market value1611. A specific treatment is provided for nth-to-default credit
derivatives based on their attachment and detachment positions 1612.

11.10.44 When calculating the net JTD, exposures, otherwise identical (except for
maturity), may be netted. The same rules apply for non-securitisation

 Back to contents 258

 MARKET RISK

positions. Long and short positions that are perfect replications through
decomposition may be netted. Different tranches may not be netted 1613.

11.10.45 In calculating default risk on securitisation positions each index is defined as

a bucket of its own1614. Bespoke securitisation tranches should be allocated to
the index bucket they are a bespoke tranche of 1615. The default risk weights
for securitisations applied to tranches are based on the corresponding risk
weight for banking book instruments, subject to a one year maturity, as
elsewhere in the trading book1616. Securitisations in the correlation trading
portfolio are subject to a similar treatment for non-securitisation positions,
with a modified hedge benefit ratio, and the possibility of negative DRC
requirements at the bucket level1617.

11.10.46 The total DRC capital requirement for correlation trading portfolio
securitisations is determined by aggregating bucket-level capital
requirements1618.

The residual risk add-on (RRAO)

11.10.47 Because the Basel Committee considered that the sensitivities-based model,
and the DRC. did not exhaust all risks for banks applying the standardised
approach to credit risk, the Committee decided to impose an add-on for other
risks. The RRAO “is to be calculated for all instruments bearing residual risk
separately in addition to other components of the capital requirement under
the standardised approach”1619. In January 2019 the Committee explained:

“The final component of the revised standardised approach is the residual risk
add-on. This provides a simple, conservative capital requirement for any
other risks not addressed by the main risk factors included in the sensitivities-
based method or standardised DRC requirement. The residual risk add-on is
the simple sum of gross notional amounts of instruments with residual risks,
multiplied by a risk weight of 1.0% for instruments with an exotic underlying
(eg weather derivatives) or 0.1% for instruments with other residual risks (eg
complex derivatives such as barrier options)” 1620.

11.10.48 The instruments subject to the RRAO are instruments with an “exotic”
underlying and instruments bearing “other residual risks”1621. An “exotic
underlying” consists of “trading book instruments with an underlying exposure
that is not within the scope of delta, vega or curvature risk treatment in any
risk class under the sensitivities-based method or default risk capital (DRC)
requirements in the standardised approach”1622. A footnote states “[e]xamples
of exotic underlying exposures include: longevity risk, weather, natural
disasters, future realised volatility (as an underlying exposure for a swap)”1623.
It is clearly a residual category, that will in the first instance be left to banks
to determine, but subject to supervisory guidance and review. The note is
clearly non-exhaustive, as the capacity of banks to create new derivative
instruments is essentially limitless. For example, terrorism derivatives (based
on the likelihood and severity of a future terrorist attack) would be covered.

11.10.49 MAR states that instruments bearing other residual risks are the following:

instruments subject to vega or curvature risk capital requirements in the

trading book with pay-offs that cannot be written or perfectly replicated

 Back to contents 259

 MARKET RISK

as a finite linear combination on vanilla options with a single equity price,

commodity price, exchange rate, bond price, credit default swap price
or interest rate swap [i.e. most complex derivatives]; or

instruments which fall under the definition of the correlation trading

portfolio, except for those instruments that are recognised in the market
risk framework as eligible hedges for correlation trading exposures 1624.

11.10.50 The problem with this definition is that it is ambiguous as to which instruments
are subject to the RRAO, and differences in judgment between banks and
national regulators may be expected.

11.10.51 MAR sets out a “non-exhaustive” list of other residual risks “that may fall
within the criteria” (but need not) including:

gap risk, i.e. a significant change in vega parameters in options due to

small movements in the underlying, resulting in “hedge slipping”.
According to the Basel III text this includes all path-dependent options,
such as barrier options, Asian options1625 and all digital options1626;

correlation risk. This is the risk of a correlation parameter necessary for

the determination of the value of an instrument with multiple
underlyings, such as basket options1627, best-of-options, spread options1628,
Bermudan options1629 and quanto options1630. (Bermudan options need not
have multiple underlyings and are a hybrid between American and
European options); and

behavioural risk. This is the risk of a change in exercise or pre-payment

outcomes, such as fixed-rate mortgage products where retail clients may
take decisions other than motivated by financial gain (the text refers to
demographic factors or other social factors) 1631. What these may be is
left unexplained1632.

11.10.52 The Basel III text states that the following criteria will not “cause” an
instrument to be subject to the RRAO (although they may in specific cases):

risk from a cheapest-to-deliver option1633;

smile risk. This is defined as the risk of a change in an implied volatility

parameter necessary for determining the value of an instrument with
optionality relative to the implied volatility of other instruments’
volatility with the same underlying and maturity, but different
moneyness1634;

correlation risk arising from multi-underlying European or American plain

vanilla options, and from options that can be written as a linear
combinations of such options (including index options);

dividend risk arising from a derivative instrument whose underlying does

not consist solely of dividend payments; and

index instruments and multi-underlying options that satisfy specified

criteria1635.

 Back to contents 260

 MARKET RISK

Calculation of the RRAO

11.10.53 The RRAO is additional to any capital charge under the standardised approach
as follows:

the scope of instruments must not have an impact in increasing or

decreasing the delta, vega or curvature risk under the (sensitivities-
based) standardised approach1636; and

the RRAO is the sum of gross notional amounts of the instruments deemed
to bear a residual risk multiplied by a risk weight as follows: (1) if the
underlying is “exotic” then the risk weight is 1%; and (2) if the instrument
bears other residual risks is 0.1%1637.

11.10.54 It should be noted that these are absolute figures and not risk weights, so a
1% capital charge is 1% of the gross notional amount and not a 1% risk weight.

11.10.55 If a bank cannot satisfy its supervisor that the RRAO provides a sufficiently
prudent capital charge, then the supervisor may impose an additional capital
charge under Pillar 21638. This reflects the rule that the Basel standards are
only minimum requirements.

11.11 Basel III - The Simplified Standardised Approach

11.11.1 The original 2016 Market Risk Standard contained only versions of the models-
based approach and the standardised approach. However, in a consultative
document published in June 2017 the Committee observed:

“As a standardised approach, however, the complexity of the SbM

[sensitivities-based method] may pose challenges for some banks (eg banks
with a low concentration of trading book activity and smaller banks that
typically do not have sufficient infrastructure for computing the SbM).
Moreover, in some jurisdictions, large banks face less complex risks. Broadly
speaking, the current requirements of the SbM make the revised market risk
framework’s standardised approach difficult or not necessarily appropriate to
implement and adopt across all Basel Committee on Banking Supervision
member jurisdictions and non-member jurisdictions”1639.

11.11.2 The consultative document proposed a reduced sensitivities-based method,

including removal of capital calculations for vega and curvature risk, with, as
an alternative, a recalibrated version of the Basel II standard. In a further
consultative document published in March 2018, the Committee concluded
that an amended version of the latter was appropriate. Basically, the change
from the 1996 standard was to apply a multiplier to the capital calculation
based on the calculation for interest rate risk, equity risk, commodity risk and
foreign exchange risk. According to the March 2018 consultative document:

“The recalibration proposed is intended to make the Basel II standardised

approach’s calibration comparable with, but slightly more conservative than,
the revised ‘full’ standardised approach. Given its relatively more
conservative calibration, the Committee does not propose to specify eligibility
requirements for banks that may use this approach. Nevertheless, the
Committee notes that the simplicity of the approach means that it may not

 Back to contents 261

 MARKET RISK

be appropriate for banks that (i) are globally systemically important banks (G-
SIBs); (ii) use internal models for determining the market risk capital
requirement for part of their trading book; or (iii) maintain correlation trading
portfolios”1640.

11.11.3 The 2019 Market Risk Standard states that “[s]upervisors may allow banks that
maintain smaller or simpler trading books to use the simplified alternative to
the standardised approach”1641. Supervisors may wish to consider the criteria
listed immediately above and “[t]he use of the simplified alternative is
subject to supervisory approval and oversight. Supervisors can mandate that
banks with relatively complex or sizeable risks in particular risk classes apply
the full standardised approach” even if the indicative criteria are not met 1642.
The position seems to be that there is a strong presumption that G-SIBs, banks
using the internal-models approach for any of their trading desks or banks
with a correlation trading portfolio may not use the simplified standardised
approach, although national supervisors have discretion whether or not to
permit it. However, given the reasons that the Basel Committee articulated
for adopting the simplified approach, it ought in principle to be available for
banks with smaller or simpler trading books. If a bank cannot determine the
full standardised approach capital charge, or only could at excessive cost,
then we expect regulators will allow its use, particularly given the intended
calibration which is designed to produce higher capital charges than under the
standardised approach.

11.11.4 When publishing the 2019 Market Risk Standard the Committee stated that
the existing Basel 2.5 “approach will be retained as a simplified alternative
to the revised standardised approach, subject to the application of specified
scalars to ensure a sufficiently conservative calibration of capital
requirements for these banks. The scalars per risk class are set at: 1.3 for
interest rate risk; 3.5 for equity risk; 1.9 for commodity risk; and 1.2 for FX
risk. As the scalars are multiplied by the capital requirement calculated under
the Basel 2.5 framework, the scalar of 1.3 for the interest rate risk means a
30% increase in capital requirements”1643.

11.11.5 MAR 40 sets out the simplified standardised approach including the rules for
determining capital requirements for interest rate, equity, commodity and
foreign exchange risk. The capital requirement is the sum of the capital
charges for these four risk classes1644.

Interest rate risk

11.11.6 Interest rate risk covers the risk of holding or taking positions in debt
securities and other interest rate related instruments held in the trading book.
This includes fixed and floating rate securities, as well as non-convertible
preference shares. Convertible bonds are treated as debt securities if they
behave like debt securities and as equities if they behave like equities 1645.
Traded mortgage securities and derivatives on mortgages are stated to possess
unique characteristics because of the risk of prepayment by the mortgagor.
No common treatment applies, and the capital charge is left to national
supervisors to determine1646.

11.11.7 The capital charge is sub-divided into “specific risk” and “general market
risk”. Specific risk is the idiosyncratic risk applicable to the issuer, whereas

 Back to contents 262

 MARKET RISK

general market risk is the risk of loss as a result of general changes in the
interest rate (e.g. a shift in the Bank of England base rate or FedFunds rate).

11.11.8 The capital charge for specific risk on interest rate securities is set out in the
following table1647:

Specific risk capital requirement for issuer risk

Categories External rating Capital requirement
Government AAA to AA- 0%
A+ to BBB- 0.25% if residual
maturity is six months
or less
1% if the residual
maturity is over six
months and up to 24
months (2 years)
1.6% if residual
maturity is greater than
24 months
BB+ to B- 8%
Below B- 12%
Unrated 8%
Qualifying 0.25% if residual
maturity is six months
or less
1% if residual maturity
is greater than six
months and up to 24
months (2 years)
1.6% if residual
maturity exceeds 24
months
Other BB+ to BB- 8%

Below BB- 12%

Unrated 8%

11.11.9 If government paper is denominated in the domestic currency and funded by

the bank in the same currency a lower risk weight may be applied 1648.

11.11.10 Important is the definition of “qualifying” items as they benefit from a lower
capital requirement. This class comprises:

securities issued by public sector entities and multi-lateral development

banks that do not qualify for sovereign treatment under the credit risk
framework;

 Back to contents 263

 MARKET RISK

other (corporate) securities rated investment grade by at least two credit

rating agencies specified by the national authority;

other (corporate) securities rated investment grade by one credit rating

agency and not less than investment grade by any other rating agency
specified by the national authority; and

subject to supervisory approval, unrated securities deemed to be

investment grade by the bank, provided the issuer has securities (debt or
equity) listed on a recognised stock exchange 1649. If the bank uses the IRB
approach to credit risk, then the unrated securities must be treated as
investment grade under the bank’s approved credit risk model1650.

Securitisation positions held in the trading book attract a specific risk capital
charge the same as banking book positions, as set out in chapter 7 on
Securitisation1651.

11.11.11 Banks may limit the capital charge for securitisation positions and credit
derivatives to the maximum potential loss 1652.

11.11.12 There are specific rules relating to the treatment of hedges for specific risk 1653,
as well as nth-to-default credit derivatives1654, and for the securitisation
correlation trading portfolio1655.

11.11.13 The capital charge for general market risk is “designed to capture the risk of
loss arising from changes in market interest rates”1656. As under the Market
Risk Amendment, two approaches are allowed: a maturity method or a
duration method1657. In each method the capital requirement is the sum of
four components:

the net short or long position in the whole trading book;

a small proportion of the matched positions in each time band;

a larger proportion of the matched positions across different time bands;

and

a net charge for option positions1658.

11.11.14 Separate maturity ladders must be used for each currency with the capital
charge calculated on a currency-by-currency basis, and then summed with no
offsetting between long and short positions, with a possible different
treatment for those currencies in which business is insignificant1659.

11.11.15 Under the maturity method, long or short positions in debt instruments (and
derivatives) are slotted into 13 time bands (15 for low-coupon instruments).
Fixed-rate instruments are slotted according to their residual maturity, and
floating-rate instruments until the next repricing date1660. The first step in the
calculation is to weigh the positions in each time band by a factor designed
to reflect the price sensitivity of those positions to assumed changes in
interest rates. The following table sets out the requirements 1661:

 Back to contents 264

 MARKET RISK

Maturity method: time bands and weights

Coupon 3% or Coupon less than Risk weight Assumed change
more 3% in yield
1 month or 1 month or less 0% 1
less
1 to 3 months 1 to 3 months 0.2% 1
3 to 6 months 3 to 6 months 0.4% 1
6 to 12 months 6 to 12 months 0.7% 1
1 to 2 years 1 to 1.9 years 1.25% 0.9
2 to 3 years 1.9 to 2.8 years 1.75% 0.8
3 to 4 years 2.8 to 3.6 years 2.25% 0.75
4 to 5 years 3.6 to 4.3 years 2.75% 0.75
5 to 7 years 4.3 to 5.7 years 3.25% 0.7
7 to 10 years 5.7 to 7.3 years 3.75% 0.65
10 to 15 years 7.3 to 9.3 years 4.5% 0.6
15 to 20 years 9.3 to 10.6 years 5.25% 0.6
Over 20 years 10.6 to 12 years 6% 0.6
12 to 20 years 8% 0.6
Over 20 years 12.5% 0.6

11.11.16 The second step is to offset the weighted long and short positions in each time
band, resulting in a single short or long position for each time band. However,
as each time band includes different instruments and different maturities, a
10% capital requirement applies to the smaller of the long or short position,
to reflect basis risk and gap risk. For example if the weighted long positions
in a time band are $100 million and the weighted short positions are $90
million, then a capital charge of 10% of $90 million applies, i.e. $9 million1662.

11.11.17 The result of the above calculations is to produce two different sets of
weighted positions: the net long or net short position in each time band and
the additive capital charge based on the smaller of the net long or short
position across all positions in that time band, as mentioned above 1663. Once
these calculations have been performed two further sets of off-setting are
required:

between the net positions in each of three supervisory zones; and

then between the net positions in each zone1664.

11.11.18 The zones are designated as follows:

time bands on 0 to one year: Zone 1;

time bands of one to four years: Zone 2; and

time banks of four years or over: Zone 3 1665.

 Back to contents 265

 MARKET RISK

11.11.19 A different calculation of the time bands applies for coupons of less than 3
years1666.

11.11.20 Matching takes place as follows: in the first instance between weighted long
and short positions in each time band; and in the second, residual unmatched
positions in one time band against positions in another time band. The rules
are set out in the table below:

Offset between zones

Zones Time band Within the Between Between
zone adjacent zone 1 and 3
zones
1 0 to 12 40% 40% 100%
months
2 1 to 5 years 30% 40% 100%
3 5 years plus 30% 40% 100%

11.11.21 The way this works is that matched positions within each zone are subject to
the capital charge in the third column. Remaining unmatched positions that
can be matched against an adjacent zone are subject to a 40% capital charge,
and remaining positions that can be matched only against zones 1 and 3 are
subject to a 100% capital charge. This is referred to in the Basel III text as a
disallowance factor1667.

11.11.22 The alternative to the maturity-based method referred to above for

calculating general interest rate risk is the duration-based approach. This is
dependent on supervisory approval. This is considered a more accurate means
of measuring market risk. The basis of the duration-based approach is as
follows:

banks must calculate the price sensitivity of each instrument in terms of

a change of interest rates of between 0.6% and 1% depending on the
maturity of the instrument;

the resulting sensitivity measures are slotted into a duration-based

ladder with 15 time bands;

long and short positions in each time band are subject to a 5%

disallowance factor (i.e. a capital charge); and

the net positions are carried forward each time for horizontal
offsetting1668.

Duration method: time bands and assumed change in yield

Zone 1
1 month or less 1
1 to 3 months 1
3 to 6 months 1
6 to 12 months 1

 Back to contents 266

 MARKET RISK

Duration method: time bands and assumed change in yield

Zone 2
1 to 1.9 years 0.9
1.9 to 2.8 years 0.8
2.8 to 3.6 years 0.75
Zone 3
3.6 to 4.3 years 0.75
4.3 to 5.7 years 0.7
5.7 to 7.3 years 0.65
7.3 to 9.3 years 0.6
9.3 to 10.6 years 0.6
10.6 to 12 years 0.6
12 years to 20 years 0.6
Over 20 years 0.6

11.11.23 The capital charges for offsets across different zones are the same as for the
maturity approach considered above1669.

11.11.24 The treatment of interest-rate derivatives is complex. Derivatives must be

converted into positions in the underlying instrument and subject to specific
and general interest rate risk1670. The amounts reported must be the principal
amount of the underlying or notional underlying under the Basel II standard1671.
Futures and forwards are treated as a combination of a long and a short
position in a notional government security1672. Swaps are treated as two
notional positions in government securities with relevant maturities. Thus a
swap under which a bank receives floating rate interest payments is treated
as having a long position in a floating rate instrument of a maturity until the
next interest rate fixing period, and a short position in a fixed-rate instrument
with a maturity equal to the residual maturity of the swap 1673. Equity swaps
are treated under the equity framework 1674. Banks can exclude from the
market risk framework identical long and short positions in the same
instruments1675. This makes sense as there is no market risk on such positions.
Certain other derivative positions may be treated as matched and fully
offset1676. Alternative rules may apply to banks with large swap books1677, and
certain swaps are not subject to a specific risk capital charge 1678.

11.11.25 The treatment of interest-rate derivatives is summarised in the following

table1679:

Interest rate derivatives

Instrument Specific risk General interest
charge rate market
charge
Exchange- Government debt Yes Yes, as two
traded future positions

 Back to contents 267

 MARKET RISK

Interest rate derivatives

Instrument Specific risk General interest
charge rate market
charge
Corporate debt Yes Yes, as two
positions
Index on interest No Yes, as two
rates (e.g. LIBOR positions
or RFR1680)
OTC forwards Government debt Yes Yes, as two
positions
Corporate debt Yes Yes, as two
positions
Index on interest No Yes, as two
rates positions
FRAs and No Yes, as two
swaps positions
Forward FX No Yes, as one
position in each
currency
Options Government debt Yes Either (i) carve
out or (ii)
general interest
risk charge under
the delta-plus
method with
additional capital
charges for
gamma1681 and
vega risk
Corporate debt Yes ditto
Index on interest No ditto
rates
FRAs and swaps No ditto

11.11.26 The specific risk capital charge for government securities only applies where
the securities are rated below AA-1682. No alternative framework exists for
jurisdictions that do not allow the use of external ratings.

Equity risk

11.11.27 Equity risk covers the risk “of holding or taking positions in equities in the
trading book. It applies to long and short positions in all instruments that
exhibit market behaviour similar to equities, but not to non-convertible
preference shares (which are covered by the interest rate risk
requirements)”1683. Long and short positions in the same equity are netted1684.

11.11.28 As with interest rate risk, the capital charge is the aggregation of specific and
general equity risk. Specific risk is “the bank’s gross equity positions (ie the

 Back to contents 268

 MARKET RISK

sum of all long equity positions and short equity positions”. General market
risk is “the difference between the sum of the longs and the sum of the shorts
(ie the overall net position in an equity market”. The long or short position
is calculated on a market-by-market basis1685.

11.11.29 Unlike the position for interest rate risk, the capital charge for specific and
general equity risk are both 8%1686. These are additive requirements.

11.11.30 For equity derivatives the following rules apply. Apart from options, equity
derivatives and off-balance sheet positions affected by changes in equity
prices are subject to these requirements, including swaps and futures 1687.
Derivatives are converted into notional equity positions. Futures and forwards
are reported at current market prices. Future index positions are reported as
the mark-to-market value of the notional underlying portfolio1688. Equity
swaps are treated as two notional positions1689. Matched positions are
disregarded1690.

11.11.31 The treatment of options is set out in the table below.

11.11.32 In addition to the capital treatment described above for general market risk
(8%), a 2% additional capital requirement applies to the net long or short
position in an index contract comprising a diversified portfolio of equities 1691.
Carve-outs apply to specified arbitrage strategies 1692. In such cases the
minimum capital requirement is 4% (i.e. 2% of the positions on each side) to
reflect divergence and execution risks).

11.11.33 The capital treatment for equity derivatives is summarised below 1693:

Treatment of equity derivatives

Exchange-traded or OTC future Specific risk General market
risk
Individual equity Yes Yes, as underlying
Index 2% Yes, as underlying
Options Individual equity Yes (a) carve out; or
(b) general market
risk charge
based on the
delta-plus
method (with
separate
capital
requirements
for gamma and
vega risk)
Index 2% Ditto

Foreign exchange risk

11.11.34 Under the simplified standardised approach gold is treated as a currency 1694
and not as a commodity, unlike elsewhere in the Basel III framework. This is
a hold-over from the Basel II standard.

 Back to contents 269

 MARKET RISK

11.11.35 FX market risk is based on two processes:

measuring the exposure in a single currency position; and

measuring the risks inherent in a bank’s mix of long and short positions
in different currencies1695.

11.11.36 A bank’s net open position in any given currency is the sum of: (1) the net
spot position (all assets less liabilities); (2) the net forward position (i.e. all
amounts to be received less amounts to be paid under forward FX
transactions); (3) guarantees and similar instruments certain to be called and
likely to be irrecoverable; (4) net future income/expenses not yet accrued
but fully hedged; (5) any other item reporting a profit or loss in foreign
currencies; and (6) the net delta-based equivalent of the total book of foreign
currency options1696.

11.11.37 Banks that are not permitted to use an internal model by its supervisory
authority must use a shorthand model that treats all currencies (including
gold) equally1697. This involves converting each FX position at spot rates into
the reporting currency1698. The overall position is calculated by aggregating
(1) the greater of the sum of net short or long positions (without netting); and
(2) the net position in gold1699.

11.11.38 The capital charge is 8% of the overall higher net position (long or short) 1700.

Commodities risk

11.11.39 The commodity risk weight applies to “a physical product which is or can be
traded on a secondary market e.g. agricultural products, minerals (including
oil) and precious metals”1701. Unlike the rest of the Basel III framework gold is
treated as a currency1702. The Committee observes that “[t]he price risk in
commodities is often more complex and volatile than that associated with
currencies and interest rates. Commodity markets may also be less liquid
than those for interest rates and currencies and, as a result, changes in supply
and demand can have a more dramatic effect on price and volatility”1703.

11.11.40 The Basel Committee identifies the following risks with commodity positions:

for spot or physical trading, the directional risk arising from the change
in the spot price;

banks using portfolio strategies involving forward or derivative contracts

may be exposed to additional risks including basis risk, interest rate risk
and forward gap risk (the risk that prices may change for reasons other
than a change in interest rates)1704; and

OTC counterparty credit risk is stated to be captured by Annex IV of Basel

II (not described here)1705.

11.11.41 There are two separate ways of capturing commodities risk under the
simplified standardised approach: (1) the maturity ladder approach and (2)
the simplified approach. According to the Committee, “[b]oth the maturity
ladder approach and the simplified approach are appropriate only for banks

 Back to contents 270

 MARKET RISK

that, in relative terms, conduct only a limited amount of commodities

business”1706. This seems inconsistent with the Basel Committee’s statement
that some non-G-SIBs may use the simplified market risk framework, and may
be a holdover from the Basel II standard. In any case, as supervisors will
decide whether a bank may use the simplified standardised approach they will
be in a position to determine what a “limited amount” means in their
jurisdiction.

11.11.42 Under both the maturity ladder and simplified approaches, long and short
positions in each commodity may be reported on a net basis to calculate open
positions1707. Positions in different commodities are generally not off-settable,
although national supervisors may allow netting against different sub-
categories of the same commodity deliverable against each other, or as close
substitutes, subject to national approval and meeting a quantitative 0.9
correlation1708.

11.11.43 Under the maturity ladder approach banks are required to express each
commodity position in the standard unit of measurement, converting the net
position into the notional (assumed, the reporting) currency at the spot
rate1709. To capture forward gap and interest rate risk within a time band,
matched short and long positions in each time band carry a capital
requirement. Positions in separate commodities are first entered into a
maturity ladder, while physical stocks are allocated to the first time band. A
separate maturity ladder is used for each commodity, and for each time band
the sum of long and short positions that are matched are multiplied first by
the spot price for the commodity, and then by a spread of 1.5% 1710. A table in
MAR 40.69 sets out the 1.5% spread rate for all maturities1711 determining the
relevant time bands.

11.11.44 The residual net positions from “nearer” time bands are carried forward to
off-set exposures in time bands that are further out, with a surcharge of 0.6%
of the net position carried forward in respect of each time band than the net
position carried forward. The capital requirement for matched amounts by
carrying forward net positions is 1.5%. For unmatched positions, the capital
charge is 15%1712. Specified rules apply to commodity derivatives1713.

11.11.45 The alternative to the maturity ladder is the simplified approach. “In
calculating the capital requirement for directional risk under the simplified
approach, the same procedure will be adopted as in the maturity ladder
approach described above … Once again all commodity derivatives and off-
balance sheet positions that are affected by changes in commodity prices
should be included. The capital requirement will equal 15% of the net
position, long or short, in each commodity”1714. To protect the bank from basis
risk, interest rate risk and forward gap risk, the capital requirement for each
commodity is subject to an additional (additive) capital requirement of 3% of
the bank’s gross positions, long plus short, in each particular commodity.
Derivatives are included at the current spot price1715.

Options

11.11.46 Two alternative approaches are permitted at national discretion:

 Back to contents 271

 MARKET RISK

banks that only purchase options may use the simplified approach
(described below, and distinct from other simplified approaches); but

banks that write options must use the delta-plus method or the scenario
approach. The more significant a bank’s trading activity is “the more the
bank will be expected to use a sophisticated approach, and a bank with
highly significant trading activity is expected to use the standardised
approach or the internal models approach”1716.

Simplified methodology for options

11.11.47 Banks “that handle a limited range of purchased options can use the simplified
approach … for particular trades”1717. In the simplified approach to options,
the positions for the options, and the associated cash, underlying or forward
are subject to capital requirements that incorporate both specific risk and
general market risk. The risk numbers are added to the category e.g. interest
rate risk, equity risk, FX risk, etc. This is set out in the table below 1718:

Capital requirements under the simplified approach

Position Treatment
Long cash and long put The capital requirement is the
market value of the underlying
or
security multiplied by the sum, of
Short cash and long call specific and general market risk
charges for the underlying less the
amount the option is in the money
(if at all)
Long call or long put The lesser of: (1) the market value
of the underlying security
multiplied by the sum of specific
and general market risk charges for
the underlying; and (2) the market
value of the option.

The delta-plus method for options

11.11.48 This is the primary method under the simplified standardised approach. “The
delta-plus method uses the sensitivity parameters of Greek letters associated
with options to measure their market risk and capital requirements. Under
this method, the delta-equivalent position of each option becomes part of the
simplified standardised approach … Separate capital requirements are then
applied to the gamma and vega risk of the option positions. The scenario
approach uses simulation techniques to calculate changes in the level and
volatility of its associated underlyings. Under this approach, the general
market risk charge is determined by the scenario grid (ie the specified
combination of underlying and volatility changes) that produce the largest
loss. For the delta-plus method and the scenario approach, the specific risk
capital requirements are determined separately by multiplying the delta-
equivalent of each option by the specific risk weights”1719.

11.11.49 Banks that write options are allowed to include delta-weighted options within
the simplified standardised approach. Such options must be reported as a

 Back to contents 272

 MARKET RISK

position equal to the market value of the underlying multiplied by the delta.
However, as delta is regarded as insufficient to capture the risks arising from
options, the simplified standardised approach requires calculation of gamma
and vega risk. Gamma risk measures the rate of change in delta, while vega
risk measures the sensitivity in the value of an option to a change in
volatility1720. These sensitivities are calculated according to a bank’s approved
exchange model, or proprietary options pricing model, subject to oversight by
the relevant national supervisor1721.

11.11.50 Delta-weighted positions are slotted into the interest rate time bands
considered earlier. A two-legged approach applies, with one leg based on the
time the underlying contract takes effect and a second based on the
derivative’s maturity. An example given by the Basel Committee is of a bought
call option in April on a June three month interest rate future which must be
treated as a long position with a five month maturity and a short position with
a two month maturity. Further examples are provided of floating rate
instruments with caps or floors1722.

11.11.51 The capital treatment of options with equities as the underlying is also based
on the delta-weighted position. The treatment of options on commodities is
based on the simplified or maturity ladder approach has been discussed
above.

Gamma and vega risk

11.11.52 The need to calculate capital charges for these risks has been mentioned
above in the context of options under the delta-plus method.

11.11.53 Gamma risk is calculated as one half of the gamma risk multiplied by the
squared value of VU. VU is determined based on the type of option and varies
from 0% to 15% depending on the underlying 1723. Vega risk (unlike under the
sensitivities-based approach) is based on the sum of all vega risks for all
options based on the same underlying multiplied by a proportional shift in
volatility of +/- 25%1724. The total capital requirement for vega risk is the sum
of the absolute value of all individual capital requirements for vega risk 1725.

Scenario approach

11.11.54 According to the Basel Committee, “[m]ore sophisticated banks may opt to
base the market risk capital requirement for options portfolios and associated
hedging positions on scenario matrix analysis. This will be accomplished by
specifying a fixed range of changes in the option portfolio’s risk factors and
calculating changes in the value of the option portfolio at various points along
this grid. For the purpose of calculating the capital requirement, the bank
will revalue the option portfolio using matrices for simultaneous changes in
the option’s underlying rate or price and in the volatility of that rate or price.
A different matrix is set up for each individual underlying”1726. As an
alternative, at the discretion of the national supervisor, banks that are
significant traders in options are, for interest rate options, permitted to base
the calculation based on a minimum of six time bands. Limitations on time
bands apply1727. Under the scenario approach banks must use the highest of
the assumed changes in yield applicable to the time band. For other portfolios

 Back to contents 273

 MARKET RISK

an 8% figure (plus or minus) applies to equities, gold, foreign exchange and

15% for commodities1728.

11.11.55 Under the scenario approach there is a second dimension based on a change
in the volatility of the underlying rate or price. The Committee states that a
single change in the volatility of the underlying rate or price equal to a shift
of volatility of plus or minus 25% “is expected to be sufficient in most
cases”1729.

11.11.56 After calculating the matrix, each cell contains the profit or loss of the option,
and any underlying hedge instrument. The capital requirement is the largest
loss contained in the matrix1730. After calculating the matrix, the capital
requirement is the largest loss in the matrix1731.

11.11.57 In the final 2019 standard the Basel Committee states that other risks
associated with options including rho (the rate of change of the value of the
option with respect to interest rates) and theta (the rate of change of the
value of the option with respect to time are also relevant: “While not
proposing a measurement system for those risks at present, it [the
Committee] expects banks undertaking significant options business at the very
least to monitor such risks closely. Additionally, banks will be permitted to
incorporate rho into their capital calculations for interest rate risk, if they
wish to do so”1732. Whether this exhortation will be heeded, given the
alternatives of the sensitivities-based approach and the internal models-based
approach discussed earlier in this chapter in unclear.

 Back to contents 274

 CREDIT VALUATION ADJUSTMENT

12. CREDIT VALUATION ADJUSTMENT

12.1 Introduction

12.1.1 A credit valuation adjustment is a change in the market value of derivative

instruments to take account of credit risk. It therefore represents the
discount in the market value of a derivative to take account of the possibility
of the default of the counterparty. It is similar to counterparty credit risk,
but was excluded from CCR, as it is not a pure credit risk but is a function of
the effect of the risk of default on market prices. It therefore has more in
common with market risk, but is treated separately in the Basel framework
from the market risks discussed in the preceding chapter, as it has its own
specific treatment that differs from that for market risk. According to
Akkizidis and Kalyvas 1733 “[t]he CVA analysis is a critical element in pricing
over-the-counter (OTC) derivatives. Since the changes in CVA are due to the
market pricing of counterparty risk, the variability of the counterparty risk
over time could be potentially far more significant than the credit risk of the
underlying position. Hence, the fair value of a financial derivative depends
on the counterparty credit risk … of the traded derivative”1734.

12.1.2 In its 2015 Review of the Credit Valuation Adjustment Risk Framework , the
1735

Basel Committee defined CVA as:

“an adjustment to the fair value (or price) of derivative instruments to

account for counterparty credit risk (CCR). Thus, CVA is commonly viewed as
the price of CCR. This price depends on counterparty credit spreads as well
as on the market risk factors that drive derivatives’ values and, therefore,
exposure. The purpose of the Basel III CVA capital charge is to capitalise the
risk of future changes in CVA”1736.

12.1.3 The Committee explained:

“During the financial crisis, banks suffered significant counterparty credit risk
(CCR) losses on their OTC derivatives portfolios. The majority of these losses
came not from counterparty defaults but from fair value adjustments on
derivatives. The value of outstanding derivative assets was written down as it
became apparent that counterparties were less likely than expected to meet
their obligations”1737.

12.1.4 As a result, Basel 2.5 adopted an express CVA capital charge:

“The current CVA framework sets forth two approaches for calculating the
CVA capital charge, namely the ‘Advanced CVA risk capital charge’ method
(the current Advanced Approach) and the ‘Standardised CVA risk capital
charge’ method (the current Standardised Approach). Both approaches aim
at capturing the variability of regulatory CVA that arises solely due to changes
in credit spreads without taking into account exposure variability driven by
daily changes of market risk factors”1738.

 Back to contents 275

 CREDIT VALUATION ADJUSTMENT

12.1.5 The 2015 review identified two major deficiencies with the Basel 2.5
approach:

a failure to capture all CVA risks i.e. CVA sensitivity to variability in

underlying market risk factors; and

the Basel 2.5 framework does not cover the exposure component of CVA
risk, and, consequently, does not recognise the hedges that banks put in
place to target the exposure component of CVA variability1739.

12.1.6 Additionally, the Committee sought to align the CVA capital charge with
existing accounting practices and the Basel III framework for market risk.

12.1.7 Accordingly, the two Basel 2.5 approaches are withdrawn and two new
approaches are introduced. These are:

the basic approach (BA-CVA); and

the standardised approach (SA-CVA)1740.

12.1.8 Unlike other elements of the Basel III framework, banks require supervisory
approval to use the standardised approach. If this is not forthcoming, then
banks must use the basic approach1741. No internal model approach is
permitted for CVA. For this reason we will describe first the basic approach
before proceeding to consider the standardised approach.

12.1.9 The regulatory CVA treatment is intended to be aligned with accounting

standards, but may differ because:

regulatory CVA excludes the effect of a bank’s own default. This is an

obvious prudential adjustment as a bank that is more likely to fail should
not be able to reduce its capital requirements as a result; and

several aspects of best practice in accounting standards are expressly

incorporated into the regulatory CVA1742

12.1.10 CVA is defined in the Basel III standard as:

“the adjustment of default risk-free prices of derivatives and securities

financing transactions (SFTs) due to a potential default of the
counterparty”1743.

12.1.11 As with the market risk capital charge, and for the same reason, the risk-
weighted assets for CVA are multiplied by 12.5 1744.

Scope of the CVA Charge

12.1.12 CVA must be calculated for all instruments within scope (i.e. derivatives and
repo-style transactions) across both the banking and trading book. This is
another reason for treating it as a market risk. The following exceptions from
the CVA capital charge apply:

 Back to contents 276

 CREDIT VALUATION ADJUSTMENT

all derivatives transacted directly with a “qualifying” central

counterparty;

derivatives giving rise to a central counterparty exposure where the bank

is a client of a clearing member that meet specified criteria (the most
important being the bank has no exposure to the default of its clearing
member); and

securities financing transactions (SFTs) fair-valued by a bank for

accounting purposes where the supervisor is satisfied that the bank’s CVA
loss exposures are “immaterial”. The bank must justify the assessment
with relevant documentation1745.

12.1.13 Capital requirements apply to the totality of the CVA portfolio on covered
transactions across its entire portfolio (including eligible hedges). It is
therefore an additional capital charge1746.

12.1.14 There is a carve-out for banks below a “materiality” threshold. Any bank
whose aggregate notional amount of non-centrally cleared derivatives is less
than or equal to €100 billion may elect not to use the BA-CVA or SA-CVA
methods “and instead choose an alternative treatment”1747. This is to set its
CVA capital requirement at 100% of the bank’s capital requirement for
counterparty credit risk (CCR). Under this approach, hedges are not
recognised, and the treatment must be applied to the entire portfolio (i.e.
banks cannot mix and match between this approach and the other approaches
for different sub-portfolios). The national supervisor may also revoke this
option “if it determines that CVA risk resulting from the bank’s derivative
positions materially contributes to the bank’s overall risk”1748.

12.2 Hedging

12.2.1 Hedging benefits may be recognised under both the basic and standardised
approaches, although this varies between the approaches.

12.2.2 Hedging instruments may be either external (i.e. with an external

counterparty) or internal (with one of the bank’s trading desks)1749.

12.2.3 All external hedges (including both eligible and ineligible CVA hedges) must
be included in the CVA calculation of the counterparty providing the hedge.

12.2.4 All eligible external CVA hedges are excluded from the market risk capital
charge described in the previous chapter.

12.2.5 Ineligible external CVA hedges are instead treated as trading book instruments
and subject to the relevant treatment for market risk and not CVA.

12.2.6 Internal hedges are only recognised under the CVA framework if they involve
two perfectly offsetting positions: one on the CVA desk and another on the
trading desk. If such an internal hedge is ineligible then both positions are
allocated to the trading book where (being perfectly matched) they offset
each other leading to no impact on either the CVA portfolio or the trading
book.

 Back to contents 277

 CREDIT VALUATION ADJUSTMENT

12.2.7 If the CVA hedge is eligible then the CVA desk position forms part of the CVA
portfolio and is capitalised as described in this chapter. The trading desk
position, is instead capitalised under the applicable trading book treatment
for market risk.

12.2.8 If an internal CVA hedge involves an instrument that is subject to either

curvature risk, default risk or the residual add-on (e.g. instruments with
optionality, or exotic options) then it is only eligible as an internal hedge if
the trading desk that is the CVA desk’s internal counterparty executes a
transaction with an external counterparty that exactly offsets the trading
desk position with the CVA desk i.e. the risk must be transferred to an external
third party1750.

12.3 Banks using the IRB Approaches for Credit Risk

12.3.1 Banks using either the basic or standardised approach to CVA risk may cap the
maturity factor (a scalar) at 1 for all netting sets contributing to a CVA capital
charge when they calculate their CCR capital charge under the internal
ratings-based approach1751.

12.4 The Basic Approach to CVA

12.4.1 This is the default approach for banks that do not have supervisory permission
to use the standardised approach. There are two sub-approaches: the full
approach and the reduced approach. Banks can choose which to apply,
although the full approach is intended for banks that hedge CVA risk, while
the reduced approach is intended to simplify the calculation for banks that
do not hedge CVA risk1752. Hedges are simply not recognised under the reduced
version. All banks that use the full version must, in addition, calculate the
reduced version disregarding any hedges1753.

12.4.2 Akkizidis and Kalyvas summarise the basic approach as follows:

“In both reduced and the full version of BA-CVA approaches, the RWs assigned
to the exposures are based on the classification of the counterparty sector
and credit quality. In the reduced BA-CVA the bank has to compute the stand-
alone CVA capital. The capital charge resulting from the reduced BA-CVA is
also included in the capital charge for the full BA-CVA. The latter also
comprises systematic and idiosyncratic components and the quantity …
reflecting the indirect hedges that are not fully aligned with counterparties’
credit spreads”1754.

The reduced version

12.4.3 The reduced version is based on a square root calculation aggregating two
elements. The first element aggregates the systemic components of CVA risk
and the second element the idiosyncratic components of CVA risk. Both
elements in the formula are derived from the CVA capital requirement that
each counterparty of the bank would receive if considered on a stand-alone
basis multiplied by specified supervisory correlation factors. The purpose of

 Back to contents 278

 CREDIT VALUATION ADJUSTMENT

these factors recognises that the actual CVA risk faced by a bank is less than
the sum of each stand-alone CVA factor1755.

12.4.4 The stand-alone calculation for CVA risk for each counterparty is calculated
across each netting set with that individual counterparty. It is based on a
number of components, although the formula will be omitted here. Relevant
factors are the volatility of the credit spread of the counterparty, the
effective maturity of each netting set, the exposure at default (calculated in
the same way as for counterparty credit risk) and two supervisory factors (a
supervisory discount factor and a multiplier to convert effective expected
positive exposure into exposure at default) 1756. The volatility of the credit
spread is calculated based on a supervisory table depending on the sector of
the counterparty, whether the obligor is investment grade, high yield or not
rated1757.

The full version

12.4.5 As mentioned above, the full version allows recognition of hedges. Only
transactions used for the purposes of mitigating the counterparty credit
spread component of CVA risk, and managed as such, are eligible hedges 1758.
The only CVA hedges that are recognised are: single-name CDS, single-name
contingent CDS and index CDS1759. A single-name CDS must either: (1)
reference the counterparty directly, (2) reference an entity legally related to
the entity (i.e. a member of its corporate group); or (3) reference an entity
that belongs to the same sector and region as the counterparty 1760. The
concept of “single-name” is therefore quite wide.

12.4.6 All banks that use the full version must also calculate the reduced version as
well1761.

12.4.7 Recognition of eligible hedges is based on the following elements:

the stand-alone and the correlation parameter are defined in exactly the
same way as above for the reduced version;

a quantity that gives recognition to single-name hedges of credit spread

risk;

a quantity giving recognition to the reduction of CVA risk across all

counterparties from the use of index hedges; and

a quantity characterising hedge misalignment, designed to limit the

extent to which indirect hedges can reduce capital requirements given
that they will not fully offset movements in a counterparty’s credit
spread1762.

12.4.8 The latter is determined in accordance with a formula with three main terms:
(1) the systematic components of CVA risk arising from the bank’s
counterparties, single name hedges and index hedges; (2) the idiosyncratic
components of CVA risk arising from the bank’s counterparties and the single-
named hedges; and (3) the aggregation of the components of indirect hedges
that are not aligned with counterparties’ credit spreads1763.

 Back to contents 279

 CREDIT VALUATION ADJUSTMENT

12.5 The Standardised Approach for CVA

12.5.1 As mentioned above, supervisory consent is required to use this calculation.

It is an adaptation of the standardised approach for market risk (with different
parameters). The main differences are as follows:

there is a reduced granularity of market risk factors; and

default risk and curvature risk are ignored 1764.

12.5.2 Akkizidis and Kalyvas summarise the approach in the following words:

“The SA uses the sensitivities-based method aligned with the FRTB framework
of the trading book. However, SA-CVA reduces the granularity of risk factors
and the complexity of the related methods used in sensitivity analysis, the
estimation of the SA-CVA capital charges relies on a given set of formulae and
estimated sensitivities provided by the new framework. Under the new CVA
banks estimate the sensitivity to each risk factor of the aggregate CVA and of
the market value of all eligible hedging instruments in the CVA portfolio” 1765.

12.5.3 They add that the exclusion from the SC-CVA standardised approach of default
risk and a less demanding set of risk factors, compared with the standardised
approach to market risk “may explain why risk aggregation under the SA-CVA
is more conservative than under the [market risk] framework. Hence, the SA-
CVA framework does not recognise any diversification effects between delta
and vega risks nor any potential perfect correlation between CVA risks and
their corresponding hedges. For each risk factor, the proposed framework
does not allow any correlation between the CVA exposures and hedges, if the
exposures are not perfectly hedged. The fact that correlations of imperfect
hedges are not eligible impacts the diversification and results in higher capital
requirements”1766.

12.5.4 Banks are required to calculate capital requirements and report them to
supervisors at the same frequency as the market risk standardised approach
figures. In addition, banks must be able to produce the standardised approach
capital calculation at the request of supervisors 1767.

12.5.5 The standardised approach uses as inputs the sensitivities of regulatory CVA
to counterparty credit spreads and market risk factors driving the value of
covered transactions1768. The following minimum criteria apply to use of the
standardised approach:

the bank must be able to model exposure and calculate, at least on a

monthly basis, CVA and CVA sensitivities to specified market risk factors;
and

the bank must have a CVA desk or similar dedicated function responsible
for risk management and hedging.

12.5.6 The bank must calculate regulatory CVA for each of its counterparties with
which it has a position relevant to the CVA calculation 1769. The following
principles (amongst others) must be adhered to:

 Back to contents 280

 CREDIT VALUATION ADJUSTMENT

regulatory CVA is the expectation of future losses resulting from the

default of the counterparty assuming the bank has no default risk;

the calculation is based on: (1) the term structure of market-implied

probability of default (PD); (2) market-consensus expected loss given
default (ELGD); and (3) simulated paths of discounted future exposure;

the term structure of market-implied PD is estimated from credit spreads

observed in the market, or proxies where there is not actively traded
credit;

the ELGD has to be the same as the one used to calculate risk-neutral
PDs from credit spreads;

the simulated paths of discounted future exposure are derived by pricing

all derivative transactions with the counterparty along simulated paths
of relevant risk factors, and discounting prices by the risk-free interest
rate;

all market risk factors material for transactions with a counterparty must
be simulated as stochastic processes1770 for an appropriate number of
paths;

dependence between transactions must be taken into account where a

significant level of dependence exists between exposures and the
counterparty’s credit quality; and

specific rules apply for margined transactions1771.

12.5.7 The following qualitative requirements also apply:

the bank must have a CVA risk management framework that includes the
identification, measurement, management, approval and internal
reporting of CVA risk;

senior management must be actively involved in the risk control process

and regard CVA risk as an essential aspect of the business;

the bank has a process for ensuring compliance with a documented set
of internal policies, controls and procedures;

there is an independent control unit at the bank responsible for effective

initial and on-going validation of exposure models;

documentation requirements are met;

pricing models must be tested against appropriate independent

benchmarks for a wide range of market states;

an independent review of the overall CVA management process should be

carried out regularly by internal audit; and

 Back to contents 281

 CREDIT VALUATION ADJUSTMENT

data integrity standards are adhered to.1772

Hedging under the standardised approach

12.5.8 Only whole transactions that are used to mitigate CVA risk are eligible. A
hedge cannot be split into different risks1773. Eligible hedges can include:

instruments hedging the variability of the counterparty’s credit spread;

and

instruments that hedge the variability of the exposure component of CVA

risk1774.

12.5.9 Hedges excluded from the internal models-based approach to market risk
(e.g. tranched credit derivatives) are ineligible 1775.

Capital calculation

12.5.10 The standardised approach capital calculation for CVA is the sum of the capital
requirements for delta and vega risk across the entire CVA portfolio, including
eligible hedges1776. This may be scaled up by national supervisors “if the
supervisor determines that the bank’s CVA model warrants it (eg if the level
of model risk for the calculation of CVA sensitivities is too high or the
dependence between the bank’s exposure to a counterparty and the
counterparty’s credit quality is not appropriately taken into account in its CVA
calculations)”1777.

12.5.11 The capital calculation for delta risk is the sum of the delta capital
requirement for the following risk classes:

interest rate risk;

foreign exchange (FX) risk;

counterparty credit spread risk;

reference credit spread risk (credit spreads that drive the CVS exposure
component);

equity risk; and

commodity risk1778.

12.5.12 The capital requirement for vega risk is the sum of the vega capital
requirement for the following risk classes:

interest rate risk;

foreign exchange (FX) risk;

reference credit spread risk;

 Back to contents 282

 CREDIT VALUATION ADJUSTMENT

equity risk; and

commodity risk1779.

12.5.13 Counterparty credit spread risk is not relevant to the vega risk calculation.

12.5.14 Unlike under the market risk standardised approach, “CVA sensitivities for
vega risk are always material and must be calculated regardless of whether
or not the portfolio includes options”1780. In calculating vega risk the model
must include both (1) volatilities used for generating risk factor paths and (2)
volatilities using for pricing options1781.

12.5.15 For each risk class subject to a delta or vega capital charge the sensitivity of
the aggregate CVA and the sensitivity of all hedging instruments must be
calculated1782.

12.5.16 The weighted sensitivities are then aggregated into a capital requirement
within each bucket specified in the Basel III text, including a hedging
disallowance factor to prevent recognition of perfect hedging. Bucket-level
capital requirements are then aggregated across buckets within each risk
class. The correlation parameters are also specified in the Basel III text, and
differ from those applicable under the standardised approach to market
risk1783. The remainder of the text of MAR 50 sets out the supervisory risk
buckets, sensitivities, risk weights and correlations. As these vary between
delta risk and vega risk, and across the defined risk factors, it would not be
helpful to set them out here and the reader is referred to the Basel III text1784.

 Back to contents 283