An approach for best practice in model validation

An approach for best practice in model validation

By Alexander von Felbert, Head of Risk Management/Authorised Officer, Airbus Bank.

Alexander, can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?

I am a passionate risk manager with more than a decade of experience in mainly market, credit and liquidity risk. I have worked in several financial institutions, whereby I validated various risk and valuation models – from potential future exposure to alternative investment risk models.

I am living with my family in Munich (Germany).

What, for you, are the benefits of attending a conference like the ‘Stress Testing Europe Summit’ and what can attendees expect to learn from your session?

Meeting other experienced risk managers and benefiting from their experience. Another big plus is to get an overview of markets best practice and expanding my network.

Can you briefly describe an approach for best practice in model validation? What validation tools would you recommend?

Model validation is all about reducing uncertainty (e.g. model risk) and increasing the trust in your model outcomes. It is therefore crucial to validate in a structured and governance-compliant way.

A good validation approach starts with understanding for what purpose a model is actually used. Moreover, possible validation activities should be divided into validation areas to further structure the validation process. This ensures that every model will be validated according to a defined process while complying with the in-force model validation governance. For instance, the validation of input data might be sub-divided into the validation of its accuracy, its appropriateness and its completeness.

What are the key considerations that need to be met when taking a structured approach to model validation?

The validator should apply the principle of proportionality and materiality while covering all main sources of uncertainty (e.g. model risk). The validation process should comply with model (validation) governance, for instance, the process and its results should be documented in the so-called validation report.

How do you deal with uncertainty in model validation?

Uncertainty is a situation which involves imperfect or unknown information. Uncertainty is everywhere and you cannot escape from it. Validation is one way to reduce uncertainty by double-checking assumptions, reviewing the methodology while considering the intended model purpose, and complying with model governance rules.

How do you see the risk landscape evolving over the next 6-12 months?

Non-financial risks such as political risks might become even more important. An efficient risk management is therefore required to understand the link between external and internal risk drivers.

You may also be interested in…

Make your free account