IFRS 9 - Lifetime Expected Loss - time slices below one year
IFRS 9 - Lifetime Expected Loss - time slices below one year
If one wants to calculate the Lifetime Expected Loss, the (introductory) examples often show a breakdown into several years with one value each for PD, LGD and EAD per year. What is the experience with finer time slices (e.g. per quarter or month)? When might these be appropriate? How can a PD be derived for such finer time slices? Thanks for your thoughts, comments, experience.
Re: IFRS 9 - Lifetime Expected Loss - time slices below one year
It may be appropriate to use shorter term PDs when the financial assets mature within 1 year. You can simply scale down proportionally (or using any other technique) the 1-year PDs of the counterparties to the term you need (if the outcome is material).
Re: IFRS 9 - Lifetime Expected Loss - time slices below one year
Perhaps when the distinction between annual and more frequent would be more material?
- JakobLavrod
- Trusted Expert
- Posts: 190
- Joined: 15 Apr 2022, 17:11
- Location: Stockholm
- Contact:
Re: IFRS 9 - Lifetime Expected Loss - time slices below one year
Hi finmase!
There are many methods for turning a 1-year PD into a monthly PD, so I will illustrate a simple one (which only requires the PD12 value) and a more complex one (which requires access to additional information).
1: Simple method
Assume that the hazard of default is constant over the period. The probability of survival for one month is 1-PD1. The probability of surviving 12m then becomes (1-PD1)^12 = 1-PD12. Solving for PD1 gives PD1 = 1 - (1 - PD12)^(1/12).
2: Complex method.
Such a method requires that you have access to the timing of default events within the year. You can fit a survival model (see, for example, https://en.wikipedia.org/wiki/Survival_analysis). If you have a portfolio of assets with homogenous risk, you could, for example, use a Kaplan-Meier estimator (https://en.wikipedia.org/wiki/Kaplan%E2 ... _estimator)
As pointed out by JRSB, how an advanced method is needed depends on the materiality.
Best of luck!
There are many methods for turning a 1-year PD into a monthly PD, so I will illustrate a simple one (which only requires the PD12 value) and a more complex one (which requires access to additional information).
1: Simple method
Assume that the hazard of default is constant over the period. The probability of survival for one month is 1-PD1. The probability of surviving 12m then becomes (1-PD1)^12 = 1-PD12. Solving for PD1 gives PD1 = 1 - (1 - PD12)^(1/12).
2: Complex method.
Such a method requires that you have access to the timing of default events within the year. You can fit a survival model (see, for example, https://en.wikipedia.org/wiki/Survival_analysis). If you have a portfolio of assets with homogenous risk, you could, for example, use a Kaplan-Meier estimator (https://en.wikipedia.org/wiki/Kaplan%E2 ... _estimator)
As pointed out by JRSB, how an advanced method is needed depends on the materiality.
Best of luck!
IFRS 9 Impairment Specialist
Risk Control at Svenska Handelsbanken
Risk Control at Svenska Handelsbanken