How severe are European regulatory stress test scenarios? A probabilistic calibration for the euro area

This paper applies the Growth-at-Risk (GaR) framework to assess downside risks to euro area GDP growth, and it examines its usefulness for stress testing. Supervisory stress-test scenarios are often criticized for being either too mild or implausibly severe, raising questions about calibration. This is consequential for banks' business plans as well as for systemic financial stability. We conduct a pseudo real-time evaluation of European Banking Authority (EBA) adverse scenarios published in the last decade, establishing a probabilistic benchmark against which their scenario severity can be evaluated. We find that, except for the 2021 and 2023 rounds, EBA adverse scenarios consistently lie below the 10th percentile threshold. After the pandemic shock, scenario severity and model-implied risks have moved in opposite directions, with GaR estimates pointing to declining downside risks and EBA scenarios becoming increasingly severe. However, these still fall within the models' probability distributions and therefore represent plausible-if extreme-realizations of downside risk. Supporting exercises attribute downside risks primarily to financial stress in the short-term, while medium-term horizons are shaped by term structure and housing or credit channels more. Taken together, these results suggest that GaR can serve as a transparent, data-driven complement to expert judgment in stress-test scenario design-helping to balance severity with plausibility and enhancing scenarios' credibility for financial stability assessments.