The impact of credit risk on 2021 stress tests

Mon 12 Apr 2021

On 13 November 2020, the EBA published the final methodological note for the 2021 EU-wide stress-testing exercise. The aim of the stress tests is to assess the resilience of financial institutions to adverse economic and financial developments, in particular in the event of an increase in credit risk due to the default of the borrower.

Banks are required to measure the credit risk impact of the macroeconomic scenarios on capital available, via impairments (P&L) and the REA (Risk Exposure Amount) for positions exposed to risks stemming from counterparty default.

The stress test is conducted on the basis of data as of end-2019 and 2020, and aims to study movements in the REA and results indicators (P&L) in two scenarios over a three-year period from 2021 to 2023. As in 2018, the exercise imposes a number of constraints. For example, for each scenario, the REA of each asset may not be lower than the value declared in 2020. Additionally, the EBA particularly stresses the assets recorded in bucket 3 for the purposes of IRFS 9, applying strict conditions: a prohibition on carrying out reversals/transfers to another bucket, and mandatory recording in non-performing exposures (NPE).

Exposures

In this context, all counter parties, including sovereigns, institutions, financial firms and securitisation positions, are within scope of credit risk stress tests for the estimation of P&L and REA impacts.  Nevertheless, there are some divergences in terms of exposures. The impact in P&L of positions valued at fair value through equity (FVOCI) or fair value through profit or loss (FVPL) will be subject to the market risk approach in the methodology. However, these positions are taken into account for when estimating the REA in terms of counterparty credit risk.

Static balance sheet

As in the previous exercise, the EBA requires stress tests to be based on a static balance sheet assumption. For the credit risk, assets reaching maturity will be assumed to be replaced with similar assets. Any new exposures in bucket 3 will be transferred to the stock of bucket 3 exposures, reducing bucket 1 and/or bucket 2 exposures correspondingly. This maintains the total exposure at a constant level.

Impact on P&L 

The estimation of credit impairments requires the use of statistical methods and includes the estimation of starting values for point-in-time risk parameters (PD, LGD) such as those used for ECL computation models; the estimation of the impact of the scenarios on these risk parameters;  and the computation of changes in the stock of provisions which will have an impact on P&L.

Banks are required to forecast credit impairments resulting from the materialisation of two separate scenarios, baseline and adverse, on the basis of IFRS 9 unless they are subject to nGAAP. Additionally, banks using an IRB approach must take account of the capital impact of any positive difference between the expected losses (EL) computed using the stressed prudential risk parameters and the ECL determined by the above procedure.

More broadly, both for baseline and projected scenarios, banks must follow a hierarchy of approaches to determine the parameters. For sovereign exposures in particular, banks must compute provisions applying parameters developed by the ECB for a selection of countries. For  countries where the ECB does not provide parameters, banks are required to estimate their own parameters, adopting a conservative position and following the hierarchy of approaches.

Finally, banks with significant foreign currency exposure are required to take into account the altered creditworthiness of their respective obligors, given the FX development under the baseline and adverse scenarios.

Impact on REA

The scope of the REA templates is wider than that taken into account for P&L impacts. The REA scope also extends to securitisation exposures. The REA level at 31 December 2020 serves as a floor for the total REA. This floor is applied separately for the aggregate IRB and STA portfolios. Similarly, for securitisation exposures, the REA at end-2020 will serve as the floor for the total risk exposure and must be applied separately for each calculation method (SEC-IRBA, SEC-SA, SEC-ERBA and SEC-IAA).

The Covid-19 context

In light of the pandemic, information additional to that required in previous stress tests will be gathered in a special template for exposures subject to Covid-19 related moratoria and newly originated exposures under a state-backed guarantee. This information includes exposure values, stocks of provisions, the REA and credit risk parameters per year, scenario, computation method and the main asset classes affected by Covid-19 measures.

However, two important points should be noted. Firstly, banks should make the conservative assumption that Covid-19 moratoria are no longer in place on 1 January 2021 for P&L and REA impacts. This could mean significant credit risk losses during 2021 for the purposes of these stress tests. However, maturing loans falling under a public guarantee scheme (PGS) during the stress test horizon will be replaced and still guaranteed, regardless of whether the PGS is still in place.

In conclusion, the stress tests carried out are in line with past exercises, while also taking account of the impact of the health crisis. An estimation of the potential risks incurred is useful in such an unusual economic context.

For participating banks, the main challenge will be that of properly reflecting the severity of the constraints on bucket 3 assets while at the same time identifying, translating and estimating the impacts of the pandemic.   

While the objective is to compare the different European banks, the results will need to be put into perspective in light of differences in government policies in the European Union countries, notably related to PGSs, and the difficulty of assessing the likely persistence of the risk factors for the different positions.

This article was written by Vincent Iapadre, Consultant and David Dubo, Senior Manager at Mazars.