Model Taxonomy

Definition

Model Taxonomy is a classification and documentation system for the (financial / risk) models used by an institution.

A model taxonomy might be the organizing principle behind a Model Inventory system. It is related to be is distinct from the Risk Taxonomy that aims to create a classification system for the risks to which the organization is exposed.

Classification Dimensions

The following are key classification dimensions of a Model Taxonomy

• The Type (quantitative nature) of the Financial / Risk Model (e.g. risk classification, forecast, pricing etc.)
• The nature of Risk Data that are used as Model Inputs, which is linked to the Risk Type (or multiple risk types) that are captured by the model
• The usage of the Model (more precisely Model Outputs) in decision making (The nature of the business model or business activity being supported by the model)
• In addition risk models get naturally grouped by
• The Client (counterparty) type involved (where applicable)
• The financial product type involved (where applicable)

Classification Dimensions

By Type of Financial Model

Despite the huge variety of models, there are only very few major conceptual classes of financial models in use today. Two major and recognized classes are as follows:

• risk models that use empirical data (historical data, market data, expert inputs etc.) to create a projected scenario set for future outcomes for the domain of interest
• valuation models that use some market observables and possibly further assumptions to value current positions, whether liquid or illiquid. This class contains also the sub-class of "no-arbitrage" pricing/valuation models that use assumptions about market behaviour to calculate a unique theoretical for derivative products

These categories are linked primarily to the model use classification (see below). The first class uses mostly historical information about prices or other risks factors to forecast future prices/risk factors. The second class uses future forecasts embedded in current asset prices to construct prices for less liquid or more complex assets. A third class of models can be termed "business support" models. This is a large and diverse collection of quantitative models and tools that may be used to assist with pricing, trading, structuring, marketing or any other business / management purpose not linked to risk management or valuation / financial reporting.

By Model Use Context

• Risk Acceptance
• Pricing of Risk
• Risk Control (use in Risk Limits)
• Hedging of Risk (determining hedge size)
• Portfolio Management of Risk (e.g., Risk Attribution, Portfolio Optimization)
• Model Derived Capital Buffers for risk (and Capital Allocation)
• Valuation / PnL calculation

By Business Activity

Different business activities typically lead to different collections of models and potentially different solutions to (seemingly) similar requirements. The factors that underpin this phenomenon are:

• The prevalent risk type in a given business activity. E.g. trading businesses tend to recognize and treat many risks as "market risks"
• The prevalent Accounting regime (standards) in a given business activity. E.g. Accrual Accounting skews the perceptions of Credit Risk in many commercial and retail banking activities

The following classification of business lines follows the Basel III taxonomy:

• Commercial Banking
• Retail Banking
• Corporate Finance
• Trading and Sales
• Payments and Settlement
• Asset Management
• Retail Brokerage

By Risk Type

Risk types (market, credit, operational etc.) and corresponding risk taxonomies are a natural taxonomy also for risk model classification:

• Market Risk
• Credit Risk
• Operational Risk
• Business Risk
• ALM Risk

These classes can be subdivided further, e.g. by market risk factors, default versus recovery risk , nature of operational event etc.

By Client Type

This segmentation by client type is useful in connection with credit risk and insurance risk:

• Sovereigns (and associated entities)
• Corporates
• SME's
• Individuals (with possible Stratification along levels of wealth, geography etc)

By Product Type

Products tend to be specific to business lines / clients. The factors that drive product specific models are

• Legal context (bilateral contracts versus securities)
• Purpose and optionality / complexity

Linkages with other taxonomies

Each one of the preceding classification dimensions links the model taxonomy with other organizational schemes used by the firm, for example, risk models are naturally linked with the Risk Taxonomy.