https://www.bdo.global/getattachment...spx?lang=en-GB
From p13:
-----------------
DETERMINING SIGNIFICANT INCREASES IN CREDIT RISK
"The transition from recognising 12-month expected credit losses (i.e. Stage 1) to lifetime expected credit losses (i.e. Stage 2) in IFRS 9 (2014) Financial Instruments is based on the notion of a significant increase in credit risk over the remaining life of the instrument.
The focus is on the changes in the risk of a default, and not the changes in the amount of expected credit losses. For example, for highly collateralised financial assets such as real estate backed loans when a borrower is expected to be affected by the downturn in its local economy with a consequent increase in credit risk, that loan would move to Stage 2, even though the actual loss suffered may be small because the lender can recover most of the amount due by selling the collateral."
I read that as saying that just because a loan moves up the risk scale , that says nothing necessarily about the likelihood of recovering the debt. For example those legacy sharemillker loans that Heartland has may be listed as impaired even though there is a ready market for cow meat if the cows could not be sold as milk producers. Shoot the cows and consequently the loan can be fully repaid.
-----------------
Now we move to p19
----------------
"Various sources of data can be used to estimate expected credit losses. Entities should consider both borrower specific factors, macroeconomic conditions, and internal and external information such as internal historical credit loss experience, internal ratings, and external reports and statistics. Entities should also take into account both the current and future forecast direction of conditions at the reporting date.An entity is not required to incorporate forecasts of future conditions over the entire remaining life of a financial instrument.
For long dated instruments, the standard does not require a detailed estimate for periods that are far in the future – for such periods, an entity may extrapolate projections from available, detailed information"
------------------
That last sentence is interesting. How I read it is that if you can make a reasonable estimation of medium term losses, say one to two years out, you can then extrapolate those losses forwards. You don't have to worry about guessing what will happen much further out than twelve months, as there will be no comeback if your longer term write off guesses are wrong.