In late 2019 issuers calculating the probability of default in their loans for the next 12 months would have put together a range of scenarios. Even their worst-case scenario, however, is unlikely to have included a global pandemic, economic slowdown, market-distorting government lending, and creditors unable to pay either the interest or the principal on their debt.
Issuers are now faced with loans on their books that look a lot more risky. But this is the world we live in today, and they must find a way to reflect the impact of Covid-19 on credit risk over the expected life of their financial assets.
KPMG has looked at actual expected credit losses reported for the period and the change in the total charge in profit or loss for expected credit losses of a sample of 11 European banks reporting between January 2020 to June 2020, comparing this with the same data from 2019. Q1 2020 saw an average increase in the expected credit loss charge of 600% compared with the same period in 2019. For Q2 it was 400%, but by Q3 2020 the average increase had fallen back to 40% compared with the same period in 2019. So the story seems to be that while the initial impact was powerful, issuers reacted fast to get things under control.
These large jumps were driven by the transition of so-called stage 1 or ‘good book’ loans to stage 2 loans – assets with more credit risk, where the expected credit loss for all events of default up to and beyond 12 months to the term of the loan must be recognised. The main challenge for issuers, therefore, is to understand the probability of default now that their borrowers’ circumstances have changed so radically.
Keep it simple
Thankfully, IFRS 9 is not prescriptive, and regulators have indicated that they consider it flexible enough to respond to the ‘specific circumstances of Covid-19’. But this is both a blessing and a curse.
Flexibility allows a spectrum of models, but this can lead to issuers using highly complex models with ‘spurious accuracy’ – a pointless level of precision. The best models are ‘simple, and theoretically sound’.
A good model should use past data and other company data to construct transition matrices that describe the possibility of a loan moving from one state to another. That would allow the issuer to create curves for probability of default, and to model conditional and marginal probability of default.
Issuers should also include the macroeconomic context in their modelling. This is best done by establishing an empirical relationship between the historical time series data of portfolio default rates and macroeconomic variables, typically by using regression techniques.
The regulators’ take
Regulators have also chipped in with advice. They have indicated that issuers should assess the impact of ‘support and relief measures’, including where such measures result in modification of the asset, and modification leading to derecognition. The view of the European Securities and Markets Authority (ESMA) is that if the measures provide temporary relief to debtors affected by the outbreak, and the net economic value of the loan is not significantly affected, then the modification is unlikely to be substantial.
Nor does ESMA regard a moratorium as immediate grounds for a significant increase in credit risk. It has asked issuers to distinguish between forbearance given because of an increase in credit risk and forbearance given because the borrower has a temporary liquidity constraint with no significant increase in credit risk. Above all, ESMA wants to see clear disclosures.
Four problems
Ultimately, issuers are facing four problems: an increase in probability of default, more exposure to stage 2 loans, collateral that is falling in value, and an increase in exposure at default because borrowers are using credit facilities to their limit. They have additional challenges in modelling because historical data is less helpful in a unique event, and because some model variables do not react well when outliers become inputs.
Additionally, models need altering to reflect the fact that the economic outlook of 2021 is good only by virtue of comparison with a dismal 2020, and the 2021 outlook itself is subject to new strains of the virus and the unknown efficacy of vaccines. Issuers can face these difficulties by putting more weight on negative scenarios, performing more out-of-model adjustments with additional data, and recalibrating their model more often.
It isn’t simple, but regulators and leading practitioners consider IFRS 9, Financial Instruments, as flexible enough to cope, even now, in this most testing of times.
