EBA stress test 2023

The European Central Bank (ECB) has announced their top three supervisory priorities^[1] from 2023 to 2025. Among them, supervised institutions are requested to strengthen their resilience to immediate macro-financial and geopolitical shocks.

The 2023 EU-wide stress test exercise will support this effort and feed into the SREP cycle. The outcomes will be critical in determining banks’ minimum capital requirements – most notably, in setting the Pillar 2 Guidance (P2G) add-on within the SREP assessment.

On 31 January, the EBA officially launched the 2023 Stress Test exercise^[2]. The announcement included publication of the final methodology, updated templates, as well as the baseline and adverse scenario paths for the forecast period (2023 to 2025).

The adverse scenario considers the heightened political tension, coupled with higher inflation and interest rate projections, which will strongly impact private consumption and investments. The scenario also includes shocks on real gross value added (GVA) by economic sector; this will be crucial in incorporating climate risk into the capital assessment, looking at key sectoral concentrations.   

The methodology is broadly consistent with the 2021 Stress Test. The 2023 exercise is performed in a bottom-up manner, utilising common methodologies, assumptions and templates. The prescribed scenarios have been developed to assess banks’ income tolerance and observe how material risks are affected through the forecast period. This year’s exercise will incorporate a much larger sample of 70 EU banks and 75% of total EU banking assets. The primary objectives are to:

• Provide all participants with consistent methodologies, framework and scenarios to assess their capital adequacy and financial resilience to macroeconomic shocks; and
• Challenge the current capital position of EU banks, to inform the SREP assessment and to facilitate greater market discipline.

The key milestone dates for the exercise are outlined below:

While the methodology is consistent with prior years, there are some notable differences – primarily in the scenario expansion. From an Irish perspective, the graphs below illustrate the key macroeconomic variable paths and compares with the corresponding forecasts for the 2021 exercise – for both baseline and adverse scenarios.  

Some key observations are as follows:

• The uncertainty around the impact of Covid-19 has reduced. This is reflected in the observed parameters, with higher initial values for Gross Domestic Product (GDP) (12.9% vs 0.5%) and lower Unemployment rate (4.7% vs 6.2%).
• The severity of the adverse scenario has increased. This is evident in GDP, Unemployment rate, and Residential Real Estate (RRE) projections, with much higher cumulative changes from the starting position. There is also a much greater “delta” between baseline and adverse pathways. 
• There are significant uplifts in both inflation (HICP) and interest rates variables in both starting positions and projections. This is likely to pose significant challenge for credit risk work streams, as neither variable is (typically) a powerful predictor of modelled losses.

These changes reflect the scenario’s downturn of geopolitical developments, along with an increase in commodity prices, higher inflation and economic contraction. Given the above changes, the capital impact of the stress test is likely to be very different from the previous exercise – with likely increases in the capital impact in the adverse scenario.

In general, the key elements of the stress test methodology remain aligned with the 2021 exercise. Banks are required to estimate the impact of credit, market, counterparty, and operational risks under baseline and adverse scenarios using a common methodology. Several of the key constraints – such as the static balance sheet assumption and the RWA floor – remain in place.

However, there are some updates from the 2021 exercise, which include:

• There is a new sector-specific reporting template. Banks are required to provide historical and projected credit risk information on exposure, LTV ratio, funded collateral, provisions and credit risk parameters broken down by year, scenario, country of exposure and NACE sector (at the 2-digit granularity level).
• The COVID-19 moratoria section has been removed in the 2023 methodology and template.
• The Stage 1 ECL coverage ratio constraint is no longer in place; Stage 1 ECL coverage – can be allowed to reduce over the time horizon of the exercise.
• The credit risk starting point parameters are to be determined from the baseline scenario – rather than assuming flat macroeconomic variables.
• In the 2021 exercise, the methodology for Counterparty Credit Risk required consideration of both an internal and an external Probability of Default (PD). However, the 2023 exercise requires banks to consider only an external PD, with the only exception to use internal PD is when no external PD is available.

In May 2022, the EBA also published its final report on Benchmarking of Internal Models^[3] in which a number of changes were recommended regarding credit and market risk.

For credit risk, there were 2 minor changes to the portfolio’s definition suggested. These were clarifications on:

1. Data collection, regarding a change in the definition of default (DoD); and
2. Further portfolio breakdown regarding FINREP sectors and unfunded credit protection.

For market risk, a number of instruments are set to be extended to provide additional information and analysis insights. There is also a list of new instruments to be introduced including equity, FX and interest rates.

For the upcoming 2023 stress test exercise, the new methodological elements introduce some new challenges for participating banks – particularly in the credit risk space. These include:

• Accurately incorporating the current interest rate and inflationary pressures within the credit risk projections. Typically, inflation and interest rates are not explanatory variables within stress testing models; therefore, additional sensitivity analysis may be required to determine final loss estimates.
• The EBA has stated that banks should rely on their sectoral models to project sector-specific risk parameters. Portfolio segmentation for stress testing models does not produce reliable models at a specific sectoral level. Therefore, alternative solutions – including sectoral sensitivities or loss distribution approaches – may be required to complete this element of the template.
• The new sectoral template utilises NACE sectors – both at the single-digit and double-digit level. This aligns the results with those from the previous Climate Risk exercise^[4], which took place in 2022. The double-digit breakdown requirement is likely to present data challenges in ensuring that the sectoral information is captured accurately.

With these updates in scenario and sectoral analysis, banks will need to take necessary steps to make the necessary adjustments to their existing processes and systems. Ensuring a strong governance framework around adjustments, modelling decisions or deviation from internal methodologies will be a critical part of the final submission.