On October 24, the Federal Reserve unveiled a landmark proposal aimed at enhancing the transparency and accountability of its annual supervisory stress tests – addressing long-standing industry concerns about opacity and unpredictability. At a time when banks and regulators face mounting pressure from evolving risk dynamics, the proposal seeks greater predictability so banks can plan capital and lending strategies with a clearer understanding of how stress-test mechanics work.
Historically, the stress test framework has been seen less as a discovery tool and more as a compliance requirement. A more constructive model would identify potential capital deficiencies early, using consistent data and transparent benchmarks rather than relying on opaque assumptions and post-hoc adjustments. This can be accomplished based on an analytical foundation shared by banks and regulators, allowing them to focus on interpreting risk, not debating methodology.
Recent work shows how this can be done using FDIC-based comparative models of regional banks. While such models may not predict failure directly, they can highlight sharp divergences in projections for peer institutions – potentially prompting earlier regulatory or investor scrutiny. Transparent, data-driven benchmarking could thus become the foundation of a more open and proactive stress testing framework –one capable of identifying anomalies before they evolve into crises.
Alla Gil
A key objective of the October proposal is to open the “black box” of supervisory stress testing by disclosing models, scenario design frameworks, and even the upcoming hypothetical 2026 scenarios for public comment. According to Vice Chair for Supervision Michelle W. Bowman, the goal is to create clearly articulated, transparent rules for setting capital requirements.
This too can be achieved with a more robust benchmarking approach – one in which banks, supervisors, and investors can systematically compare institutions side-by-side and improve clarity around capital planning. Defining comparable inputs – such as loss rates, revenue shocks, and macroeconomic drivers – enables banks to map their performance relative to peers. This comparability not only exposes hidden assumptions but also empowers regulators, investors, and rating agencies to ask sharper questions when a bank’s projected capital buffer diverges from its peer group.
Historic Step
As a first step, the Fed has published the structure, variable selection, and modeling criteria for stress test scenarios in advance, allowing public and industry feedback before finalization. This marks a historic move toward transparency and collaborative risk management.
The proposed 2026 Severely Adverse Scenario envisions a severe global recession triggered by a sharp decline in risk appetite – steep losses in equities and corporate bonds, plunging interest rates, surging unemployment, and double-digit declines in GDP and real estate values.
Some critics worry the proposal could make stress tests less stringent. Fed Governor Michael S. Barr, the former vice chair for supervision, warned the reforms might render tests “less conservative” and more open to gaming.
To safeguard against that, it’s crucial to develop a wider set of systemic stress scenarios, including reverse stress testing across multiple shock combinations linked to banks’ balance sheets.
A major technical concern is the continued reliance on standard regression models for projecting balance sheet and income statement variables. In real-world stress conditions, regression assumptions – especially stationarity – break down. This forces management to apply judgmental overlays to correct unrealistic outputs, creating large discrepancies between banks’ internal models and the Fed’s supervisory results.
Solution for Complexity
In a typical CCAR bank, as many as 40% of the thousand equations used in stress test expansions may require manual adjustments. This complexity undermines comparability and transparency. A more robust approach would use machine-learning–based regression with regularization, which doesn’t rely on stationarity or a-priori hypotheses, fits outliers without overfitting, and identifies the most critical drivers through exhaustive cross-validation.
Fortunately, much of the data required for such expansion modeling already exists in public or semi-public datasets. FDIC call reports, FFIEC disclosures, and historical series for credit losses across loan categories provide rich input for model calibration. Market risk exposures, trading losses, and even proxy indices for private equity and credit can be leveraged for broader analysis. Where counterparty data is limited, banks and supervisors can align on high-level assumptions to maintain comparability.
The key suggestion is that transparency and scenario discovery don’t automatically lead to better predictions – they lead to better questions. And asking better questions is what makes the system more resilient.
Parting Thoughts
The Fed’s proposal marks a turning point, aiming to make stress testing more realistic, transparent, and efficient while maintaining its rigor. A standardized benchmarking framework shared by banks and regulators could close communication gaps and turn stress testing into a more constructive, forward-looking process.
Real progress lies in building a common analytical language – one that reveals hidden vulnerabilities, fosters trust, and strengthens resilience. The era of black-box stress tests is ending; the next generation should not just measure resilience but build it.
Alla Gil is co-founder and CEO of Straterix, which provides unique scenario tools for strategic planning and risk management. Prior to forming Straterix, Gil was the global head of Strategic Advisory at Goldman Sachs, Citigroup, and Nomura, where she advised financial institutions and corporations on stress testing, economic capital, ALM, long-term risk projections and optimal capital allocation.
Topics: Stress Testing & Scenario Analysis, Regulation & Compliance