2022 ECB climate risk stress test: Let’s keep the ball rolling

The ECB’s supervisory climate risk stress test began on 27 January 2022 and will run through the first half of 2022, with aggregate results due to be published in July. Participating banks have worked against the clock in the last weeks, preparing starting point data and projection templates for the first submission on 7 March 2022. And yet, immediately after, banks will still have significant challenges and uncertainties to resolve before the following submission in late March.

After spending the final months of 2021 in preparation for the climate risk stress test,significant institutions have focused in the first months of 2022 on execution. Banks were also collecting or approximating missing data, calibrating internal approaches for the ECB scenarios and preparing templates for the Advanced Data Collection (ADC) submission on 7 March.

As highlighted in our previous article, this is the first supervisory climate stress test for most banks directly supervised by the ECB. That presents new challenges, especially when it comes to sourcing climate-relevant data and making meaningful credit risk projections. The ECB has described the climate risk stress test as a learning exercise for all parties, but banks will no doubt be keen to compare their readiness and experiences with those of their peers. 

One of the ECB’s main objectives for this exercise is to better understand banks’ data capabilities around climate-related risks. The preparatory phase showed that most banks’ systems do not contain all the data requested by the ECB. Banks therefore had to rely on a combination of publicly available data, third-party providers and proxies to fill the gaps in their internal data.

That is certainly the case for Scope 1, 2 and 3 emissions data for non-SME corporate counterparties, and for energy performance certification (EPC) ratings assigned to mortgage and corporate real estate exposures. For example:

• Discussions with banks revealed that, as of February 2022, the proportion of actual Scope 1 and 2 GHG emissions data collected to date remained below 40% on average, and that the availability of actual EPC ratings was below 30% for half of banks.
• In some countries, banks appear to have been collecting some of the missing emissions data through direct dialogue with clients.
• Following the ECB’s recommendation, banks have developed proxies to overcome data scarcity, mostly relying on the exemplary approaches suggested in the Methodological Note – e.g., using average sector-based emissions and approximating the EPC rating based on a property’s characteristics, such as the year of construction or size. It is also worth noting that the ECB incorporated internal benchmarks on emissions data in its data quality framework and challenged banks data prior to the ADC submission by assigning data quality flags in case of large discrepancies.

In addition to starting point data, some banks are required to provide bottom-up projections for credit risk. The need to include projections already in the ADC submission, almost a month earlier than in the previous EBA stress test, added further pressure to an already challenging exercise.

Before the submission of the credit risk projections, banks had to deal with various challenges. These included:

• How to reflect the impact of carbon (CO₂) pricing in credit risk projections. Discussions with banks suggest many are considering the impact of carbon pricing indirectly via macro-economic variables, rather than using carbon prices as a model risk driver. For example, some banks are using GVA¹ forecasts to derive sector-specific projections, assuming that these implicitly consider the impact of carbon pricing.
• How to produce long-term projections under the dynamic balance sheet assumption. In line with the ECB’s methodological note, banks are expected to consider two components when making balance sheet projections: 1) the bank-specific strategy, and 2) the bank’s business environment linked to the development of macro-economic scenarios.
• Calibration of credit risk parameters to the ECB scenarios: Unlike the previous EBA stress test exercises, the ECB did not provide a probability of default (PD) and loss given default (LGD) path generator as a standard benchmark on the expected impact across the different scenarios. The inclusion of new elements into banks internal models, such as industrial sectors, EPC grades, risk regions, new climate-related risk drivers, and a variety of projection horizons, increased their uncertainties on the appropriateness of their self-produced projections.

The ECB's data quality checks prior to the submission of the ADC confirmed their expectations in terms of accuracy and consistency between other supervisory disclosures and the data included in the climate stress test templates. In addition to the checks against COREP and FINREP, known from previous stress tests, checks against AnaCredit were required for Module 2.

Even after their first submissions, banks need to resolve some uncertainties, especially regarding the adequacy of:

• The approaches adopted to reflect the high level of impairments from the end of 2021 due to the COVID-19 crisis in their start point and projected parameters.
• The calibration of parameters under the various scenarios.

The coming weeks will see intensive activity as banks are expected to receive the official data quality report from the ECB , following the ADC submission. This moment will mark the start of the Quality Assurance process, with two submission cycles — Full Data Collection (FDC) 1 and 2. During this phase it is foreseen a continuous dialogue between banks and the ECB stress test team. The FDC 1 submission is due on 31 March, while FDC 2 is expected by 23 May. The ECB will then publish the exercise’s main conclusions and aggregate, non-bank specific findings on 6 July.

The results of the stress test may not have a quantitative impact on banks’ Pillar 2 Guidance, but the data quality of ADC submissions — such as Module 3 projections — will be qualitatively factored into the Supervisory Review and Evaluation Process (SREP).

Looking further ahead, banks should expect the climate risk stress test to become integrated into yearly supervisory reviews. As Edouard Fernandez-Bollo, Member of the Supervisory Board of the ECB, said recently:

“The climate stress test is just the first step in learning to quantify this risk — it’s a learning exercise to improve the reliability of the figures…. In other words, this stress test isn’t the end of the story — eventually we will have capital requirements”.²