Starling Advisory Group, hired by the Federal Reserve to investigate Silicon Valley Bank’s failure and the Fed’s preceding supervision of the bank, released a preliminary report of its findings this week.
Those findings were:
- SVB failed in 2023 because it suffered massive losses when interest rates rose, its depositors ran, and it had insufficient contingent-liquidity readiness.
- By June 2022 at the latest, and perhaps as early as March, Federal Reserve supervisory staff knew, or should have known, that SVB was highly vulnerable to massive losses and a potentially ruinous run.
- Despite that awareness, Board S&R and FRBSF supervisors did not require SVB to reduce its interest-rate risk or exposure to runs by uninsured depositors.
- Their failure to do so did not result from the 2018 tailoring statute (EGRRCPA) or a less assertive supervisory culture promoted by the implementation of related policy priorities.
- Board S&R has featured a culture of uncertainty regarding decision-making authority that dates back decades, and which is directly implicated in its failure to take decisive action at SVB.
- Emphasis on collegial decision-making through committees fosters inaction due to a divorcing of responsibility, authority, and accountability across the Federal Reserve supervisory system.
- Social media did not fuel the run at SVB: if “social media” is understood to imply platforms like Twitter, chatter regarding the run at SVB began after it was well underway; and if “fueled” is meant to imply either ‘triggering’ or ‘accelerating’ the run, then close analysis of time-stamped Twitter data disconfirms either interpretation.
The Fed commissioned Starling to conduct an independent review of “any lapses in supervision that may have contributed to SVB’s failure” and to recommend reforms in supervisory practices.
The report depicts an unfocused supervisory regime that issued a flurry of MRAs and MRIAs on a wide range of perceived shortcomings yet failed to properly identify or mitigate the liquidity and interest-rate risks that brought down the bank. The impression that emerges from the report – which is expected to be followed by subsequent reports – is one of a distracted, disjointed supervisory system that missed the forest for the trees. That impression is familiar – it’s also reflected in BPI’s postmortem analysis of the supervision of SVB from May 2023. (Worth noting: This initial report does not reflect interviews with the frontline examiners from the Federal Reserve Bank of San Francisco; those perspectives will likely be included in subsequent reports.)
- The Fed issued 54 MRAs or MRIAs on SVB between 2020-2022, according to Mary Miller, the former chair of the Audit Committee of SVB’s board of directors. “None of these MRAs or MRIAs directed SVB to reduce the duration of its bond portfolio or its exposure to uninsured depositors. Instead, most of them related to technology and operational matters, including BSA/AML, cyber risk, and identity-access management,” the Starling report says, citing an interview with Miller.
- The report describes a regime that prioritized process and minutiae over material risk. Citing interviews with SVB’s independent directors, Starling states: “The Directors acknowledged the rapid accumulation of supervisory findings but emphasized that most concerned technology, operational resilience, and process, rather than deficiencies relating to the level of SVB’s liquidity or interest-rate risk. Of the twelve MRIAs issued between 2019 and 2022, the Statement identifies three as relating to SVB’s process for managing liquidity risk, but not its level of liquidity risk. That is, the MRIAs were focused on remediating perceived weaknesses in processes for managing financial risk at SVB rather than remediating the financial risk itself. Other findings concerned vulnerability remediation, technology-risk management, IT asset management, vendor management, identity and access management, data governance and information security.” The directors also asserted that when SVB grew large enough to be subject to more stringent supervision, the bank was inundated with directives about the organization of board committees, “IT asset management” and other peripheral or minor matters.
- Examiners failed to prioritize existential financial risks above governance deficiencies. The report offers the perspective of Mark Watson, who was hired by SVB to help remediate its problems. “Watson’s account speaks to the Barr Report’s finding that supervisors did not fully appreciate SVB’s vulnerabilities. On Watson’s account, supervisors did not lack information about SVB’s governance and remediation weaknesses. Those matters were the subject of MRAs, MRIAs, program-level discussions, workstream meetings and continuous examinations. What appears to have been missing was the synthesis and prioritization necessary to distinguish deficiencies requiring long-term institutional improvement from a financial vulnerability capable of destroying SVB before that improvement could be completed.”
- Expectations and communication from examiners to the bank were muddled and inconsistent, according to SVB directors. “To the Directors, ‘it felt like supervisors were moving the goalposts.’ Whether that was the product of shifting supervisory expectations, the discovery of deeper deficiencies, or the inadequacy of SVB’s initial response cannot be resolved from the Directors’ Statement alone. But the sequence provides material context missing from an account that treats an open supervisory finding as proof that the board had failed to act upon it.”
- Process and procedure took priority over action. Starling cites an interview with a Fed staffer who noted that “even the exam findings at SVB that had focused on interest-rate and liquidity risk were nevertheless not aimed at addressing those risks in a manner calculated to reduce them. Supervisors were not, he said, directing SVB to hedge substantially more of the duration risk in its securities portfolio or to reduce the portfolio’s duration. Instead, ‘It was a lot of process and procedure and governance and controls criticism that may or may not have helped.’” Elsewhere, an enforcement attorney suggests that there was a failure to prioritize the risks that mattered most. “[Board Staff 5] therefore rejected the proposition that SVB’s unfinished [memorandum of understanding] should be understood primarily as a failure of the enforcement machinery. ‘Again, I would say that the issue is not an enforcement process issue in so much as it’s the inability or the failure to prioritize material financial risk,’ he said. ‘The back and forth is not about interest-rate risk,’ [Board Staff 5] continued. ‘It’s not about liquidity risk with the exception that they want to do another target exam in two months. It’s about how to describe and how to address cyber/IT issues.’ … Supervisors were not inactive. They were examining SVB, producing findings, considering enforcement, and attempting to construct a comprehensive response. But the concentration of supervisory and legal effort did not correspond to the hierarchy of risks capable of causing SVB to fail.”
- S&R senior staff echoed this sentiment: “The supervisory actions that did emerge from this process, moreover, may also have been directed too often toward improving risk-management processes rather than reducing risk directly. [Board Staff 10] observed that supervisors frequently wrote MRAs focused on process improvement. And even where supervisors identified a financial risk, she said, they might require the development of better models rather than prescribe an immediate change in the institution’s financial position. She emphasized that an MRA should identify its anticipated impact on the institution; if supervisors could not explain that impact, she questioned why the MRA was being written. [Board Staff 13] similarly described supervisors as emphasizing stronger risk-management frameworks rather than prescribing concrete risk-reduction measures. This distinction bears directly upon the experience at SVB, as outlined by the Independent Directors in the Statement they provided Starling….”
