FINMA update for Swiss banks
Swiss banks must calculate a general loss provision for inherent default.
The FINMA published Circular 2020/01 – Accounting-Banks. The circular update contains a revision to the standards regarding the provisioning for counterparty default risks. Banks are now required to create value adjustments for both impaired and non-impaired loans, whereas previously it sufficed to create value adjustments for impaired loans only. The aim of the new regulation is to eliminate weaknesses in the current system, in particular the risk of procyclical effect due to value adjustments being created too late.
Reguirements by Finma
This topic is already in scope of IFRS and US GAAP, the circular therefore does not apply to banks that report under these standards. The requirements by FINMA depend on the category of the banks. Systemically important banks (categories 1 and 2) should apply an expected loss approach.
The medium-sized (category 3) banks which operate primarily in the interest rate business are required to calculate a general loss provision for inherent default risk. Relatively small banks (category 4 and 5) and medium-sized banks that do not operate primarily in the interest rate business can continue to apply value adjustments for latent default risk.
Besides systemically important banks that do not prepare their financial statements under the international accounting standards IFRS or US GAAP, we expect this circular to materially impact the medium-sized banks that operate primarily in the interest rate business. They must switch from performing value adjustments for latent default risk to calculating a general loss provision for inherent default risk. There are several possible approaches to calculate the general provision for inherent default risk. The approach can be historically based and/or model-based.
In the historically based approach, we observe the historical transitions from non-impaired loans to impaired loans together with the losses that have occurred and calculate a provision based on these data. This approach is relatively simple (no parameter estimation involved), and the results are easy to interpret. However, it relies heavily on the dependency of historical data. If the historical time series of the data is not long enough and/or the quality of the data is low, provisions can be volatile and/or unreliable.
In the model-based approach we estimate the probability of default (PD) with a regression model and combine the estimated PD with the Loss Given Default (LGD) and Exposure at Default (EAD) to calculate the general provision for inherent default risk. Depending on the portfolio and availability of historical data, this approach can be more accurate than the historical based approach. However, it is more complex, and the interpretation of the results is less straightforward than those of the historically based approach.
Which approach is preferred depends on the availability of the data. If the financial institution has historical records for at least ten years of which loans have transitioned from non-impaired to impaired, which of those have defaulted and what the realized losses were, a historically based approach is justified. Ideally, the historically based approach is implemented, and the model-based approach is used to benchmark the results.
Get in touch with Zaid Siddiqi for more information about this column.