Comment on the Preliminary Report into the Failure of Silicon Valley Bank

Probative Inquiry into the Financial Sector Stress Events of 2023
Agency: Starling Advisory Group, retained by US Federal Reserve Board’s Division of Supervision & Regulation
Comment Submitted: October 16, 2026

Comment on the Preliminary Report into the Failure of Silicon Valley Bank

We welcome the Starling Advisory Group’s September 2026 preliminary report commissioned by the Federal Reserve Board’s Vice Chair for Supervision Michelle W. Bowman.¹Its examination of how the Federal Reserve’s supervisory system handled Silicon Valley Bank’s (SVB’s) vulnerabilities offers a valuable basis for improving supervision. We encourage the continuing inquiry to further examine two related issues: the line of authority within the Federal Reserve’s supervisory framework and the use of duration-insensitive capital risk weights in judging the soundness of a banking institution.

The report is clear about the evidence behind its findings. Its interviews, directors’ statement, and independent analysis broaden the record available to earlier official reviews.

The report’s executive summary lists seven preliminary findings:

1. SVB’s failure combined investment losses, unstable deposits, and inadequate emergency liquidity.
2. Supervisors had sufficient warning well before the collapse.
3. Recognizing the risks did not produce effective intervention.
4. The report rejects regulatory tailoring as the explanation for supervisory inaction.
5. Long-standing uncertainty over decision-making authority impeded action.
6. Committee-based supervision separated responsibility from authority and accountability.
7. Twitter did not trigger or meaningfully accelerate the run.

We suggest two further areas of inquiry:

1. Supervisory authority and leadership. The report diagnoses unclear authority and inertia within the Federal Reserve’s supervisory function, but has not fully traced decisive warnings to the executive positions responsible for resolving them. The continuing inquiry should further clarify how warnings were carried through the chain of command, what positions had authority to act, and how decisions were delayed or declined.
2. Duration risk in capital rules. Standardized credit-risk weights did not rise with the interest-rate sensitivity of SVB’s government and agency securities. The report explains unrecognized losses more fully than it does how risk weights affected portfolio choices or how supervisory safeguards were meant to address duration risk.

The initial report develops the failure to address SVB’s exposures, which is a valuable contribution. As the review continues from this point, it would be useful to further develop the policy and regulatory conditions in which those exposures accumulated. The first issue, of authority and leadership, is central to diagnosis of the Fed’s performance in its supervisory role. The second question, about duration risk, concerns macro-prudential policy and bank resiliency.

1. Clarifying Authority and Responsibility

The report’s strongest institutional finding concerns uncertainty about who could make a decision and bring consultation to an end. Staff describe committee layers, “pocket vetoes,” and divided responsibility among Board and Reserve Banks staffs. Risk specialists could identify serious exposures but could not require the supervisory team responsible for SVB to act. This evidence supports the report’s conclusion that inertia reflected the organization of supervision as well as individual judgment.²

The bank’s unfinished Memorandum of Understanding (MOU) illustrates this organizational problem. According to the report’s interviews, legal staff had prepared an approvable draft in late 2022. The San Francisco Fed preferred to await further examinations and pursue a broader action. Later discussions focused on cyber and IT matters. Neither the original nor revised recommendation expressly addressed interest-rate risk. An earlier MOU therefore would not necessarily have required SVB to shorten its portfolio’s duration, restore hedges, or reduce its main financial exposure. The problem involved both delay and the failure to prioritize the right intervention.³

Starling’s preliminary report also distinguishes formal authority from the authority staff understood they could exercise. It says the Division Director had decision-making power, but uncertainty remained about whether staff below that level should act, seek agreement, or escalate. A useful reform would make clear how a material warning becomes a decision assigned to an executive officer, who must document whether the recommended action is accepted or rejected.⁴

The continuing inquiry should reconstruct the critical warnings and decisions in sequence: who received each warning, what action was proposed, what objections arose, how the issue was escalated, and who made the final decision. That account should distinguish disagreements about risk from disagreements about process or authority. It should also examine how the vacancy in the Vice Chair for Supervision role from October 2021 to July 2022 affected escalation, without assuming that the vacancy explains any particular decision. The report notes that frontline testimony and important decisions remain to be fully reconstructed.⁵

SVB director Mary Miller’s account raises a related question about priorities. She reported that SVB received 54 Matters Requiring Attention or Matters Requiring Immediate Attention from 2020 to 2022, and that 40 were closed. Many early findings concerned technology and operations. Her description, however, raises the questions of whether the examiners had made directors aware of the interest rate risk that bank management had assumed and, if so, why the directors failed to address this major weakness. Closing findings show that work was done; the inquiry should establish why the directors failed to prioritize the bank’s performance risks.⁶

Reform could assign clear ownership of material risk decisions, set escalation deadlines, and require written reasons when specialists’ warnings do not lead to action. Consultation would remain part of the process, with a defined point at which an authorized official must decide. The continuing inquiry can test whether such a process would have changed the Fed’s response to precrisis signs of SVB fragility.

2. Examining Duration in Capital Requirements

The report appropriately explains that accounting treatment and regulatory capital measures allowed SVB’s held-to-maturity (HTM) securities to be reported at amortized cost despite market losses. SVB’s Accumulated Other Comprehensive Income opt-out also excluded unrealized losses on available-for-sale (AFS) securities from regulatory capital. The Charles River Associates analysis says that recognizing the AFS losses would have reduced SVB’s reported CET1 ratio from 12.1 percent to 10.4 percent, still above the regulatory minimum. About 85 percent of unrealized securities losses were in the HTM portfolio. These findings explain why reported capital could appear adequate even as the bank’s economic exposure worsened.⁷

A significant issue related to SVB’s resilience is the duration neutrality of standardized credit-risk weights. Under US rules, Treasury exposures receive a 0 percent credit-risk weight. Ordinary non-equity exposures to government-sponsored enterprises receive a 20 percent weight, while directly and unconditionally US-government-guaranteed exposures generally receive 0. Agency securities therefore do not all receive the same weight. But within a given credit category, a security’s risk weight does not rise as its maturity—and its sensitivity to interest rates—increases. A short Treasury bill and a long Treasury bond receive the same credit-risk weight even though a rise in yields affects their market values very differently.⁸

This treatment makes longer-term securities an attractive way to earn more yield without increasing the standardized credit-risk charge. In SVB’s case, the credit-risk denominator did not reflect the duration exposure that later proved so consequential. Loss recognition and risk weighting operate through different channels: One affects measured capital, whereas the other determines risk-weighted assets.

Basel did recognize interest-rate risk in the banking book. Its 2016 standard kept this risk under Pillar 2, the supervisory review process, after considering a standardized Pillar 1 capital charge. The Basel Committee cited the difficulty of developing a standardized measure that was sufficiently accurate and sensitive to risk. The question, then, is how US supervisors applied the framework intended to manage duration risk when credit-risk weights did not capture it.⁹

Implications for the Continuing Inquiry

Starling’s report offers strong evidence of failures in supervisory decision-making and gives a detailed account of how accounting treatment obscured losses. It says less about duration incentives in capital rules, which deserve attention.

We encourage the continuing inquiry to connect its institutional findings with SVB’s contemporaneous portfolio decisions and the incentives surrounding them. That analysis would help guide reforms to both bank risk management and the Federal Reserve’s ability to require timely action when a bank’s decisions threaten its resilience.