Vol. 008｜米銀行ストレステスト改革——予測可能性は安全性ではない｜The Fed’s Stress-Test Reset: Predictability Is Not Resilience
What Happened | Beyond the proposal stage On 30 September 2026, the Federal Reserve finalized changes to stress-test transparency and the volatility of related capit
What Happened | Beyond the proposal stage
On 30 September 2026, the Federal Reserve finalized changes to stress-test transparency and the volatility of related capital requirements; Reuters reported the finalization the same day. The 2025 proposal is no longer the latest policy stage.16
The framework includes annual consultation on scenarios and material model changes, and two global market shocks, using the larger loss for each bank. The transparency rule takes effect on 2 November, while key model and scenario changes apply to the 2027 test. Averaging based on the two latest tests begins in 2028. Announcement, legal effectiveness and operational application are different dates.23
The Fed expects roughly 50% less year-to-year capital-requirement volatility without a material aggregate capital effect. That is the regulator’s expectation, not an observed outcome.1
Why It Matters | Capital levels versus the machinery setting them
The SCB is a capital buffer reflecting, among other inputs, the projected decline in a bank’s common equity tier 1 (CET1) ratio under stress. The exercise tests conditional loss absorption; it is not a forecast.3
My interpretation starts with planning predictability. More stable requirements may reduce the discretionary cushion banks hold against regulatory surprises and make lending, dividends and repurchases easier to plan. But management decisions, supervision, other regulations and funding conditions remain constraints.
A 50% reduction in volatility is not a 50% reduction in required capital. Aggregate neutrality does not mean neutrality for every bank. Neither statement supports an indiscriminate bank-equity trade.
Transmission Map | The profit channel and its reversal
These are transmission hypotheses, not established effects of the reform.
- Transparency → better forecasts of requirements → reassessment of discretionary cushions → possible lending or distributions. Every step differs by bank.
- More lending → more interest income only if demand and credit costs cooperate. Available capital does not create creditworthy borrowers.
- Model-sensitive portfolio changes → lower measured losses → potentially less visible real risk. Shared omissions can weaken diversification in a crisis.
For Japan-based investors, the channels include U.S. bank valuations, dollar credit markets and comparisons of banks with U.S. businesses. U.S. reforms cannot simply be translated into capital release for Japanese megabanks; entity-level scope and consolidated rules require separate checks.
Counterpoint | Keep both sides of the transparency argument
Governor Barr dissented, warning about model optimization and slower responses to emerging risks. Yet he supported using the greater bank-specific loss from two market shocks. Calling the whole package simply an easier exam misses that distinction.4
Governor Cook expressed optimism that the framework could remain effective, while stressing continued scrutiny of severity and predictability. She also supported exploratory scenarios not directly tied to capital requirements.5
The counterevidence to my concern would be prompt incorporation of new risks and adequate resilience under independent, out-of-model tests. If that occurs, predictability can represent a genuine improvement rather than a safety trade-off.
Bull / Base / Bear | Conditional paths, not forecasts
- Bull: Transparency and independent challenge coexist; less planning friction permits distributions without weakening capital quality.
- Base: Requirements become steadier, but other constraints remain and benefits differ by bank. Disclosures in 2027 and 2028 provide the test.
- Bear: Optimization and distributions precede resilience, leaving unmeasured concentrations and thin headroom exposed when credit deteriorates.
I assign no probabilities or price targets: bank-level inputs needed to estimate capital release have not been assembled.
What to Watch | Monitoring and questions for 21 October
- CET1 headroom relative to applicable combined requirements and management targets—not the headline ratio alone.
- The 2027 loss and revenue models, both market shocks and explanations for revisions.
- Whether lower requirement volatility and genuine loss absorption coexist around the 2028 averaging start.
- Distributions versus internally generated capital, provisions, delinquencies and credit concentrations.
- Evidence that exploratory and internal scenarios challenge published-model blind spots.
Two questions for discussions before 21 October: Does greater distribution capacity come from changed requirements or a smaller discretionary safety margin? What evidence distinguishes stable test results from real crisis resilience?
One-line Verdict
A more predictable exam can improve capital planning. It cannot make the next crisis predictable.
Sources and scope
Source numbers link directly to the documents listed below the Japanese text. Primary documents establish policy and technical facts; Reuters is separately identified as reporting. No company-specific valuation or trade recommendation is offered.
