Appendix A: Model Changes for the 2024 Stress Test
Each year, the Federal Reserve refines both the substance and process of the stress test, including its development and enhancement of independent supervisory models. The supervisory stress test models may be enhanced to reflect advances in modeling techniques; enhancements in response to model validation findings; incorporation of richer and more detailed data; and identification of more stable models or models with improved performance, particularly under stressful economic conditions. Each year, the Federal Reserve also makes relatively minor refinements, if necessary, to models that may include re-estimation with new data, re-specification based on performance testing, and other refinements to the code used to produce supervisory projections.
For the 2024 stress test, the Federal Reserve has made the following notable updates to the supervisory models:113
- The first-lien mortgage model was re-estimated to include more recent time periods. The specification was changed to control for the effects of anomalous macroeconomic data variations and market distortions observed during the COVID-19 pandemic.
- The model for the international bank and charge cards portfolio and the international small-business and corporate cards portfolio, within the other retail loans category, was updated. The updated model has a simplified roll-rate structure and refined macro-economic variables to improve sensitivity to scenarios and model stability.
- The trading IDL model was simplified to remove securitized products and the dependence on the GMS. This simplification recognizes the limited obligor concentration risk presented by securitized products and the immateriality of the GMS dependence.
- The specification of the regression portion of the operational risk model was changed to control for the effects of anomalous macroeconomic data variations observed during the COVID-19 pandemic.
An assessment based on model monitoring will inform whether any other model adjustments are warranted in the 2024 stress test. The Board will provide descriptions of any model adjustments made for the 2024 stress test in the 2024 stress test results disclosure.114
References
113. None of these model changes are material and so will be implemented in full for the 2024 stress test, in accordance with the stress test policy on averaging material model changes. In accordance with the stress test policy on averaging material model changes, certain model updates discussed in the 2023 supervisory stress test methodology will no longer be averaged, including updates to the commercial real estate loss given default model and trading revenue model. Return to text
114. Consistent with past practice, the Federal Reserve will disclose any significant model adjustments made permanent. Return to text