Different Types of Credit Risk Model

Algorithmic Trading Models

The term ‘Algorithmic Trading Models’ refers to models that provide estimates/predictions and create
inherent uncertainty of market features used in the context of Trading Algorithms as defined in the
Markets FIM B.11.29, including automated trading decisions, trade execution or both via electronic
exchanges and platforms.

All Algorithmic Trading Models must be assigned to the Model Category “Algorithmic Trading Models”,
and to one of the following subcategories, depending on the asset class each model is used to trade in:

FX and Commodities;
Equities;
Global Debt Markets.

All Algorithmic Trading Models’ use cases must be assigned to one of the following purposes:

“Other – Algorithmic Trading–Client Execution”. This is the use of an Algorithmic Trading
Model to process orders provided by clients, either by internalisation or by orders sent on
exchanges and venues, generally with a view to reach an optimal equilibrium between market
impact and speed of execution, in a context where the Bank does not create market risk from
the execution.
“Other – Algorithmic Trading–Bank Trading”. This is the use of an Algorithmic Trading Model
to optimise the transactions conducted by the Bank itself for its own purposes (i.e. principal
basis)including hedging and market making, or on account of a client if the Bank incursmarket
risk from the execution.
“Other – Algorithmic Trading–Client Execution and Bank Trading”. This is the use of
Algorithmic Trading Model to trade for/by clients where the Bank creates market risk and for trading on Bank’s own behalf (i.e. principal basis), generally with a view to risk-manage the
inventory.

Asset Management Models

Asset Management Group (“AMG”) models are being used primarily in a Fiduciary capacity to support:

Portfolio construction
Risk management
Performance monitoring
Cash flow monitoring
Oversight of the Investment Team
Regulatory and client reporting

AMG models includes:

Models used to assist portfolio manager is their asset allocation and investment decisions,
which are not quant driven. For Example, Bank Analytics, Visualiser.
Models used to provide oversight of the portfolios; where no investment decisions are being
directly made based on the model outputs. For example, FiS APT, BBG LQA.

Each AMG model must be assigned with the Model category as “Asset Management Models”.

Each model use case within this category must have the purpose “Other – Fiduciary Asset
Management” and may also be used for “General Stress Test” where applicable.

Capital Markets Product Pricing Models

Capital Markets Product Pricing models refers to models used in the valuation and risk management of
financial products that can be traded on exchange or Over the Counter (OTC). These are generically
referred to as “capital markets products”. These models may be used by Global Markets to price
products for Bank’s own positions or by Securities Services to price products for clients’ portfolios.

These products refer to tradable assets and liabilities such as:

Shares, bonds and currencies, or to less liquid equivalents such as private equity, debt and
loans.
Listed and OTC Derivative contracts and structured products, for which a concept of Fair
Value is relevant.

In the rest of this document, unless otherwise stated, “pricing model” will be used for “Capital Markets Product Pricing model”.

All pricing models owned by Global Markets must be assigned the model category “Global Markets
Product Pricing Models”. All pricing models owned by Securities Services must be assigned the
category “Securities Services Product Pricing Models”. Models used to price other categories of
products offered by the bank do not belong to this category. No sub-categories are required.

All model use cases for pricing models should be assigned one of the following Purposes for each
region and business:

“Financial Reporting – IFRS13 Fair Values”. This is the front office estimation of accounting
value of a traded product, excluding any Fair Value Adjustment. This purpose also means the
model is used to price products, estimate market risks and design a hedging strategy. Models
that have a MUC with purpose “Financial Reporting – IFRS13 Fair Value” do not include
additional MUCs for Pricing and Market Risk purposes.
“Other – Other Risk Management”. This covers the use of the model for estimating the trading desk’s exposure to market risk factors, without any contribution to valuation. For example,
complex instruments that are directly marked to an observable market price, but for which a
model is needed to project market risks into separate factors.
“Other – Pricing and Valuation”. For models in the Global Markets Product Pricing Models
category this relates to use of the model for estimating a price for a capital markets product actually or potentially traded by the bank. For example, the price at which Bank should
market the product, or market values estimated for collateral purposes. For models in the
Securities Services Product Pricing Models category this relates to the use of valuation
models in the pricing of complex assets in the portfolios of clients of Securities Services. The
valuations and risk sensitivities that are calculated by Securities Services may be utilised by
the client for a number of purposes including but not limited to; collateral management,
portfolio valuations or risk reporting purposes.

Uses of pricing models as inputs for other types of models belonging to other categories should not be
considered as separate use cases. Instead, the corresponding pricing models should be registered as
feeder models.

Compliance Models

Compliance models are a group of models used to mitigate financial crime and regulatory conduct
associated risk at the bank. Models within this group include those designed to combat several different
kinds of financial crime.

All Compliance models must be assigned to one of the following sub-categories:

Anti-Money Laundering Transaction Monitoring (AML TM)	A model which is involved in the monitoring of transactional activity to identify potentially suspicious behaviour for investigation
Risk Assessment Rating	A model which determines the risk probability of a given vector (e.g. Customer Risk, Product Risk, Country Risk etc.).
Sanctions Name Screening	A model which is involved in the detection of named sanctioned individuals.
Sanctions Transaction Screening	A model which is involved in the monitoring of transactional activity for links to sanctioned individuals, countries or entities.
Surveillance	A model which is used to identify suspicious activity, market abuse and financial crime, such as insider trading and market manipulation.
Regulatory Compliance (NonSurveillance)	A model which is used in the identification of mis-selling or conduct risk via: Analytical tooling solutions that deliver connected risk insights across product lifecycles to identify customer segments who may not be receiving fair outcomes, or; An intelligence led investigation service that can be deployed to analyse major internal and external regulatory conduct incidents and associated read-across.

Compliance Model Use Cases are to be assigned one of the following purposes:

Financial Crime – Financial Crime Risk
Financial Crime – Fraud Risk
Financial Crime – Other

Financial Reporting Impairment Test Models

The Financial Reporting Function owns and uses a small number of models for calculating impairment
of goodwill and of interest in certain associates.

Impairment of Group goodwill.
Impairment of interest in Associates.

Model sub-categories are not required.

The Model Owner must establish use cases in line with the balance sheet presentation of the asset
being tested for impairment. Models used by the Corporate Centre for impairment of associates must
be assigned as follows:

BoCom: GBM, CMB and WPB
SABB: CMB

Financial Reporting Impairment Test Models are used for:

Financial Reporting – IAS 36 Value-in-Use
Financial Reporting – Fair Value Less Costs of Disposal

Global Private Banking Models

This Model Category Standards (“MCS”) sets out additional granular minimum control requirements,
where required, for identifying, measuring and managing model risk for models in two categories:

GPB Investment Models
GPB Credit Models

GPB Investment Models

Investment advice and investment management are defined as the business of managing or providing
advice on investment portfolios or individual assets for compensation. GPB Products include: Funds,
ETFs, Fixed Income, Equities, Structured Products, Derivatives, Deposits, Foreign Exchange, Custody,
Thematic Swaps, Customised Baskets, Private Equity, Private Credit, Private Debt, Real Estate Equity,
Hedge Funds, Commodities and investment-linked insurance products.

Investment Models must be sub-categorised as follows:

GPB Product / Portfolio Risk	Models that produce metrics with the purpose of measuring the risk of the security and/or the portfolio of securities for investments purposes. Examples of models under this subcategory are: Volatility, Expected Shortfall, VaR, cVaR or instances where those model outcomes are evaluated according to hypothetical or historical scenarios (Stress Testing)
GPB Asset Allocation	Models that produce portfolios of investments in terms of percentages of allocation to assets according to a target risk/return.
GPB Pricing / Tariff	Models that produce a price or a benchmark price for a loan or the tariff, or a benchmark tariff for an investment service / trade / custody.

All Investment Models must be assigned with the purpose “Other – Fiduciary Investment Management”.

GPB Credit Models

This category contains models that are developed for clients / facilities falling under the GPB scope and are used to measure or manage credit risk.
GPB Credit Models must be assigned to model sub-category of “PD”, “EAD” or “LGD”.
GPB Credit Models must be assigned an appropriate purpose:
- Regulatory Capital – Pillar 1
- Financial Reporting – IFRS9 ECL
- Other – Credit Decisioning
- General Stress Test

Models that are built solely to generate Temporary Model Adjustments (“TMAs”) must be assigned the purpose “Regulatory Capital – Pillar 1 TMA”. Where a replacement model is awaiting external approval but is currently used solely to calculate RWA in order to quantify a TMA to be applied to the approved “In Use” model, then the replacement model must also be assigned the Purpose “Regulatory Capital – Pillar 1 TMA” until the replacement model is approved and used to calculate the RWA.

Insurance Models

The Insurance models category consists of models that are commissioned by the Insurance businesses
and associated functions for:

Manufacturing and underwriting
Reinsuring
Selling and customer servicing of Insurance products
Business management

The Insurance businesses and associated functions consists of

Insurance manufacturing entities;
The Bermuda Captive; and
All areas and functions that form a part of Insurance or specifically support it.

All insurance models must be assigned to one of the following sub-categories:

Reporting	Models used for local regulatory capital and reserving, IFRS reporting, Economic Capital and Stress Testing.
Other	Models not used for Reporting but used for making decisions by Bank or by its customers.

In scope purposes for Insurance model use cases:

Financial Reporting – IFRS 4 Insurance Contracts
Financial Reporting – IFRS 9 ECL
Financial Reporting – IFRS 17 Insurance Contracts
Regulatory Capital – Insurance Pillar 1
Regulatory Capital – Insurance Pillar 2
Regulatory Capital – Basel–Group–Pillar 2A ICAAP
Regulatory Capital – Basel–Local–Pillar 2A ICAAP
Other – Asset-Liability Management
Other – Benefit Illustration
Other – Pricing and Profitability Monitoring
Other – Strategic Planning
Insurance models may also be used for stress testing, as appropriate

Models used for pre-provision net revenue forecasting, including balance sheet and profit and loss items
must follow the PPNR Forecasting MCS. The Insurance Product Risk Rating model must follow the
GPB MCS.

IRRBB Models

The Interest Rate Risk in Banking Book (“IRRBB”) Models category covers models used in the
calculation of:

Economic Value of Equity (“EVE”),
Net Interest Income (“NII”), and
Other measures required for the day to day IRRBB management, including hedging activities
and IRRBB regulatory reporting.

This category includes IRRBB feeder models that are used as key model assumptions, such as:
behaviouralisation assumptions (prepayment assumption models, pass-on assumption models), market
data assumptions, pricing assumptions.

All IRRBB models must be assigned the Category “IRRBB Models” and Model Sub-Category of “IRRBB Metrics”, “Pre-payment/Attrition Risk”, “Pipeline Risk”.

All IRRBB models must be assigned the Purpose “Financial Reporting – IRRBB”.

Liquidity Models

Liquidity models defined as any measurement identified as a model per GMRS that is used as part
of liquidity and funding risk management of the bank.

All liquidity models must be assigned to one of the following sub-categories:

Liquidity– Internal measures	Models developed to measure the liquidity risk of the bank’s portfolios based on internal design and assumptions. This covers Internal Liquidity Metric, Customer Stability Scorecard(CSS) Model, Intraday Liquidity Risk, internal long term funding measurement or any other model calculated to measure liquidity& funding risk based on internal rules and assumptions, including forecasting LCR, NSFR, ILM etc.
Liquidity–Regulatory reporting	Models developed to be used for any regulatory requirement.
Liquidity- Fund Transfer Pricing	Models that calculate internal cost of fund, Liquidity Premium(LP) curve generation and cost allocation across business lines
Liquidity- Fund Transfer Pricing	Models falling outside of the above defined sub-categories, Other measures required for the Liquidity Management activities within Group Treasury or Legal entity ALCM

Liquidity models are used for one or more of the following purposes:

Financial Reporting – Liquidity Regulatory Reporting
Other – Liquidity Risk Management

Operational Risk Models

Operational risk models are used to estimate the risk of loss arising from people, processes and systems.

Operational Risk Models must be sub-classified between “Operational Risk – Capital Models”, covering models used to calculate capital for economic capital and stress testing purposes, and “Operational Risk – Other Models” for all other models.

The following Purposes are assigned to Operational Risk Models:

Regulatory Capital – Basel – Group – Pillar 2A ICAAP
Regulatory Capital – Basel – Local – Pillar 2A ICAAP
Regulatory Capital – Insurance Pillar 2
General Stress Test

There are a number of Operational Risk models in development that will not be used for Capital
Reporting purposes. Standards for these models will be developed and incorporated into future versions
of the Operational Risk Model Category Standards. In the interim, the Model Owner must liaise with the
relevant Model Risk Steward for any additional control requirements.

Other Accounting Models

The Operational Accounting and Financial Control Functions own a small number of models. These
models are used for calculating a number of balances used as part of the period-end close process for
producing Financial Statements. These models support the following processes:

Impairment test of investments in subsidiaries held by the Holding Company Group and other
similar investments across Bank.
Calculation of right-of-use assets and related lease liabilities as required by International Reporting Standard (‘IFRS’) 16.
Determination of IFRS9 Expected Credit Losses for immaterial sites using a simplified approach
from the Group’s consolidation system.
Determination of effective interest rate, and subsequent recognition of interest income and deferral of fees and expenses for credit cards in WPB in the UK, as required by IFRS9.
Calculation of provisions for customer redress, for miss-selling Payment Protection Insurance in
the UK, as required by IAS37.
Accounting for mortgages in the USA, including judgemental reserves as per US GAAP and IFRS,
and valuation of real-estate owned on the balance sheet.
Fair value calculations for the purpose of providing fair value disclosures as required by IFRS.
Calculation of the asset and liability values and associated measures for pension schemes as
required by IAS19

Model sub-categories are not required. Other accounting models are used to generate outputs for use in Financial Statements, other than models covered in other MCS. Use cases for models in this category may have the following purposes:

Financial Reporting – IAS36 Impairment Testing
Financial Reporting – IFRS16 Right of Use Assets
Financial Reporting – IFRS9 ECL
Financial Reporting – IFRS9 EIR
Financial Reporting – IAS37 Customer Redress
Financial Reporting – US GAAP
Financial Reporting – IFRS13 Fair Values
Financial Reporting – IAS19 Pensions
Financial Reporting – Other

Pension Risk Model

The Pension Risk models category covers only risk measurement or quantification models for pension
risk. Pension Risk is defined as the risk:

To Bank caused by its contractual or other liabilities to, or with respect to, a pension scheme
(whether established for its employees or those of a related company or otherwise), and
That Bank will make payments or other contributions to, or with respect to, a pension scheme
because of a moral obligation, or because Bank considers that it needs to do so for some
other reason

The Pension Risk Model Category does not include models operated by third parties appointed by
Bank to produce pension accounting advice or data/reporting under IFRS or other accounting
requirements.

The model sub-categories covered in this MCS are limited to:

Economic Capital Models (EC) – for the purpose of Individual Capital Adequacy
Assessment Process (ICAAP) at group and local entity levels. These models are used for
the purpose of assessing Economic/Pillar2A Capital.
Stress Testing Models – for the purpose of regulatory stress tests (including PRA, CCAR,
DFAST, EBA, HKMA, GIST) consequently these models are also used for the purpose of
assessing regulatory capital; and
Internal risk metric reporting (including management information dashboard for the GPOF
and pension Risk Appetite setting). This use does not support regulatory capital
submissions and the purpose is to provide key pension risk management information.

Pension Risk models are used for the following purposes:

Regulatory Capital – Basel – Group – Pillar 2A ICAAP
Regulatory Capital – Basel – Local – Pillar 2A ICAAP
General Stress Test

PPNR Forecasting

A ‘PPNR Forecasting’ model is one whose primary use is to generate forward-looking estimates (i.e.,
forecasts) of pre-provision net revenue (“PPNR”), which is comprised of net-interest income (NII), noninterest income or revenue (“NIR”), and non-interest expense (“NIX”) according to the general formula PPNR = NII + NIR – NIX. This includes models used to forecast feeder subcomponents, e.g., asset and liability balance models, interest rate/spread models for NII and NIX, and fee rate models for NIR.

The Model Owner must ensure that each PPNR Forecasting model has at least one model use case
for the purpose of:

“Other – Strategic Planning”, which models used for includes Annual Operating Plan (“AOP”)
and Monthly Forecasting, or
“General Stress Testing”.

Product Control Models

The Product Control Model Category includes all the models owned by Product Control. This includes
models used for Fair Value Adjustments, Prudent Value Adjustments, Stress Testing and Hedge
Accounting.
All Product Control models must be assigned to the Category “Product Control Model Category”

Depending on the usage of the Model, each Product Control model and model use case must be
mapped to one of the following sub-categories and purposes, respectively:

Model Sub-Category	Description	Purpose
Fair Value Adjustments	Model use cases related to Fair value adjustments impacting PnL	Financial Reporting – IFRS13 Fair Value Adjustments
Financial Reporting OCS OCI	Model use cases related to the OCS reserve posted as a reserve (other comprehensive income).	Financial Reporting – IFRS9 OCS OCI
Prudent Value Adjustments	Model use cases related to the prudent value adjustments	Regulatory Capital – Prudent Value Adjustments
Stress Testing	Model use cases related to the stress testing usage	General Stress Test
Financial Reporting ECL	Model use cases related to IFRS 9 Expected Credit Loss impairment calculation	Financial Reporting – IFRS9 ECL
Financial Reporting Hedge Accounting	Model use cases related to Hedge Accounting process	Financial Reporting – IAS39 Hedge Accounting
Product Control – Other Models	Minor Model Use cases: PnL attribution, curve construction, FINREP purposes, IFRS 16	Financial Reporting – IFRS16 Leases Financial Reporting – Other Regulatory Capital – Other Other – Other Non-Risk Management

Retail Collections Model

Retail Collections Models are used to classify past due and/or delinquent customers into different risk
segments so that appropriate and timely collections actions can be applied.

Segmentation is generally based on their probability to roll forward.
Collection actions include decisions over contact channels, effort intensity and operational
queues.

Retail Collections Models include:

Customer collections-specific segmentations across WPB Products (assigning risk level and
treatment at customer or account levels)
Customer collections-specific scorecards

Retail Collections models do not include:

Account Management Models such as customer behaviour scorecards, CLI/CLD models, etc.
Behaviour Scores that are used as input to other Retail Collections Models.

All Retail Collections models must be assigned the Category “Retail Collections”. No sub-categories
are required.

All Retail Collections models must be assigned the Purpose “Other – Retail Collections”.

Retail Credit Capital Models

The Retail Credit Capital Models category covers models used in the calculation of pillar 1 capital
requirements for Retail Credit exposure: Probability of Default (“PD”) models, Exposure at Default
(“EAD”) models and Loss Given Default (“LGD”) models. This category does not include any other
models used to feed PD, EAD or LGD estimates such as behaviour scores, or any models used in
regulatory stress tests.

All Retail Credit Capital models must be assigned the Category “Retail Credit Capital” and Model SubCategory of “PD”, “EAD” or “LGD”.

All Retail Credit Capital models must be assigned the Purpose “Regulatory Capital – Pillar 1” except for models that are built solely to generate Temporary Model Adjustments (“TMAs”), for which the Purpose “Regulatory Capital – Pillar 1 TMA” is used. Where a replacement model is awaiting external approval but is currently used solely to calculate RWA for TMA purposes1 , then the replacement model must also be assigned the Purpose “Regulatory Capital – Pillar 1 TMA”. This purpose should be removed when the replacement model is approved and used to calculate the RWA.

Retail Credit Other Models

The Retail Credit Other Models category primarily covers models that provide a prediction of risk at the
point of application or throughout the lifetime of the account. These are referred to as Application
scorecards and Behaviour scorecards.

The Retail Credit Other Models Category does not cover marketing/propensity, pricing, collections or
fraud models that are covered by other model categories.

All Retail Credit Other models must be assigned Model Sub-Category of either “Application”, “Behaviour” or “Other”

All Retail Credit Other models with Model sub-category of either “Application” or “Behaviour” must be assigned the Purpose “Other – Credit Decisioning”.

Retail Expected Credit Loss and Economic Response Models

The Retail ECL and ER models category covers models used:

To calculate pre-Forward Economic Guidance (“FEG”) ECL estimates; and
To stress risk parameters based on economic scenarios for use in post-FEG ECL estimates
or stress tests

All Retail ECL and ER models must be assigned the Category “Retail Expected Credit Loss and
Economic Response” and Model Sub-Category of “IFRS 9”, “CECL”, “ERM” or “Recovery”.

Model use cases must be assigned one of following purposes:

Financial Reporting – IFRS9 ECL
Financial Reporting – CECL
General Stress Testing

Retail Fraud Models

Retail Fraud Models are used to predict the probability of fraud in a banking transaction or an account
opening application. They may also be used to predict the susceptibility of customers to fraud or scams.
These models identify possible predictors of fraud associated with known fraud transactions/applications in the past to detect events that are abnormal. Model outputs are generally
used directly or indirectly (through rule sets) to approve, decline or refer application or transactions on
the basis of suspected fraud. They may also be used to provide appropriate warning messaging to
susceptible customers.

Fraud models include:

Fraud models include:
i. Models that detect potentially fraud card or payment transactions (both 1st party and 3rd party).
ii. Models that detect potentially 3rd party fraud applications.
iii. Models that detect potentially 1st party fraud application either directly or by targeting creditrelated operational losses.
iv. Models that predict susceptibility of customers to potential fraud or scams, for example
customer warning models.
v. Models that are designed to improve the efficiency of fraud operational alerts, for example
false positive ratio reduction models.

Fraud models do not include:

i. Models used for anti-money laundering.
ii. Models used for detecting sanctions activity.
iii. Models used for Know Your Customer activities.
iv. Models used for making credit/limit related decisions.
v. The platform/system where fraud rules/model are coded is not a model by itself.

Retail Fraud Models must be assigned to one of the following sub-categories

1st Party Application Fraud
3rd Party Application Fraud
Card Transaction Fraud
Payment Transaction Fraud
Scams/Susceptibility
Retail Fraud Other

All Retail Fraud models must be assigned the Purpose “Financial Crime – Fraud Risk”.

Scenario Generation and Expansion Models

Scenario Generation and Expansion (“SEG”) Models generate projections for a set of macroeconomic
and financial market variables used for:

Stress testing for capital planning and capital adequacy assessment;
Forward Economic Guidance (“FEG”) for IFRS 9/CECL reserve calculations; and
Financial planning proposals.

Scenario projections reflect an economic narrative over a specified time horizon. The economic
narrative could be characterised by a simultaneous or successive occurrence of a set of expected or
hypothetical economic events. An appropriate scenario could be:

Internally designed based on firm’s vulnerabilities, financial risk assessments, or assumptions
on forward looking economic views.
Expanded from narratives directly mandated by the regulatory agencies.

Scenario models could be used both in design, and expansion of such economic scenario narratives.

There is some form of an assembly characteristic that distinguishes a scenario model from other model
types. Scenario models can have multiple outputs, namely a set of mutually related economic or
financial market variables. Unlike many other models, scenario models do not aim to project a single
output, such as one particular revenue or loss metrics. Rather, scenario models project a set of
economic and financial market variables that best represent a particular economic narrative. Hence,
Scenario models are not balance sheet, market risk or financial projection models although their output
could feed those models.

Operational Risk scenario models are not covered by this category because those scenarios are based
on non-Financial risk events and have limited linkage to macroeconomic and financial market variables.

Examples of a scenario model are the Moody’s Macro model that feeds the FEG and stress testing
process.

SEG Models must be assigned the category ‘Scenario Generation and Expansion Models’. No subcategory is required.

SEG models within traded risk, for example the traded risk factor expansion model, should be classified
as Traded Risk Models.

Scenario models are feeder models and the model outputs are used for one or more of the following
purposes:

Financial Reporting – IFRS9 ECL
Financial Reporting – CECL
General Stress Test

Sherwood Models

Project Sherwood is the framework for making new lending pricing decisions, excluding pricing for
retention or back-book portfolio decisioning, using vintage profitability for all lending products (assets)
offered by Bank’s WPB Business. Sherwood models ensure that new account booking returns:

Are optimised and meet a minimum segment and marginal Return on Tangible Equity
(“ROTE”) hurdle rates;
Have efficient use of capital; and
Credit risks taken are in line with rewards.

All Sherwood Models must be assigned the category “Sherwood Models”. The RBWM B4.4.1 New
Lending Decisioning FIM defines the “in-scope” markets eligible for a Sherwood model. The Model
Owner must also register “Sherwood-like” models developed for non-eligible market portfolios (for
example, smaller portfolios like Sri Lanka Cards, Argentina Auto Loans etc.) in the central model
inventory with “Sherwood Model” as category, with agreement from the relevant Global Product Heads.

Each Sherwood model must be sub-categorised based on the product covered by the model: “Credit
Cards”, “Personal Loans”, “Mortgages”, “Auto Loans” and “Payroll Loans”.

All Sherwood models must be assigned the purpose “Other – Pricing and Valuation”

Traded Risk Model

Traded risk model categories consist of the following types of models:

Traded Risk – Market Risk Models	Models developed to measure the market risk of the bank’s trading positions. This may arise due to changes in market factors such as interest rates, credit spreads, foreign exchange rates, equity or commodity prices.
Traded Risk – Counterparty Credit Risk Models	Models developed to measure the counterparty credit risk exposure of the bank’s trading positions. This may arise due to changes in market factors such as interest rates, credit spreads, foreign exchange rates, equity or commodity prices.
Traded Risk – Treasury Risk Models	Models developed to measure Treasury-related metrics.
Traded Risk – Other Traded Risk Models	Other Traded Risk models falling outside of the above defined categories

Regardless of the model category, all traded risk models must be assigned to one of the following
sub-categories:

Simulation	Scenario generation and risk factor simulation models
Pricing	Models that calculate PnL using pre-generated scenarios and trade data
Aggregation	Models used to aggregate results calculated using pricing models
Simulation Pricing Aggregation	Models that generate scenarios, calculate PnL and aggregate the results.
Closed Form Estimator	Non-statistical models implemented as a formula
Risk Factor Relationship	Statistical models aiming at establishing relationships between risk factors
Other	Models falling outside of the above defined sub-categories

Traded risk models are used for one or more of the following purposes:

Regulatory Capital – Basel – Group – Pillar 1
Regulatory Capital – Basel – Local – Pillar 1
Regulatory Capital – Basel – Group – Pillar 2A ICAAP
Regulatory Capital – Basel – Local – Pillar 2A ICAAP
Other – Pricing and Valuation
Other – Initial Margining
General Stress Test

Wholesale Credit Risk Models

Wholesale Credit Risk Rating Models are models where the output is either:

Probability of Default (“PD”)
Loss Given Default (“LGD”)
Exposure at Default (“EAD”)
Slotting Approach
Expected Credit Loss (“ECL”)

To meet:

Regulatory requirements for calculating risk weighted assets for credit risk, or
Accounting requirements for calculating provisions requirements.

Wholesale Credit Risk Rating Models must be sub-categorised based on the parameter produced by
the model:

PD
LGD
EAD
Slotting
ECL

Wholesale Credit Risk Rating Models are used for the following purposes:

Regulatory Capital – Pillar 1
Financial Reporting – IFRS9 ECL
Financial Reporting – CECL
General Stress Test

Wholesale Credit Non-Risk Rating Models are models where the output is not a PD, LGD, EAD or
Slotting output, for example CoLT. No sub-categories are required.

Wholesale Credit Non-Risk Rating Models are used for the following purposes:

Regulatory Capital – Basel – Group – Pillar 2A ICAAP
Regulatory Capital – Basel – Local – Pillar 2A ICAAP
Other – Credit Decisioning

WPB Customer Selection Models

WPB Customer Selection Models are used for customer targeting by propensity or behaviours,
customer elasticity, product usage, channel and digital engagements, and customer servicing, etc. This
category excludes risk (fraud, compliance, operational, credit, regulatory, etc.), Sherwood, finance, and
credit decisioning models.

Each model within the WPB Customer Selection category must be classified into one of the following
sub-categories:

Cross-sell/Upsell, including upgrade/downgrade proposition,
Activation/Attrition,
Product/Transaction Behaviour, for example balance build, revolver, digital usage, etc.,
Customer Elasticity, for example to pricing, to product, to channel,
Customer Servicing and Communication, for example ATM Cash, Pega, Inbound routing,
channel migration, operational improvements, miscellaneous,
Other WPB Customer Selection

Each model use case within the WPB Customer Selection Models category must have the purpose
“Other – Customer Selection”.