Practical Data Needs for IFRS 9 Probability of Default Estimates

Summary

The document concerns estimating a company’s probability of default for an expected loss calculation under IFRS 9. It notes that physical default probabilities can be estimated in many ways, with approaches differing in complexity and data requirements. It does not set out a particular estimation procedure or provide implementation steps.

For work that must meet IFRS 9 requirements, the answer points readers to specialist practical references on the impairment model and credit risk modelling and validation. It emphasizes that implementation requires substantial historical data. The exchange gives no examples, comparison of methods, or guidance on choosing an approach for a specific company, so it serves mainly as a pointer to further study rather than a standalone modelling recipe.

Key ideas

• Physical probabilities of default can be estimated using methods with different levels of complexity and data needs.
• IFRS 9 implementation may require specialist guidance on credit risk modelling and validation.
• Historical data is a significant requirement for building an implementation.

Tags

# Estimation of the probability of default for the expected loss model (IFRS9)


# Estimation of the probability of default for the expected loss model (IFRS9)












Hey guys I have to do a calculation for my BA. More precisely, I have to determine the expected loss of a company.

For this I need the probability of default. What options do I have to determine this myself. I have read many papers that simply don't tell me how to implement something like this in practice.

## Answer by Dimitri Vulis (score 0, accepted)

https://quant.stackexchange.com/a/59566

There are many ways to get physical PD, some much simpler and less data-intensive than others.

If you are really required to jump through all the hoops in IFRS 9, then the books The New Impairment Model Under IFRS 9 and CECL by Jing Zhang (Moody's) and IFRS 9 and CECL Credit Risk Modelling and Validation: A Practical Guide with Examples Worked in R and SAS by Tiziano Bellini probably have all the details one need to implement it. You will also need a lot of historical data.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.