A Matter of Prioritization

The latest Silicon Valley Bank post-mortem by the Starling Advisory Group places considerable emphasis on the Federal Reserve’s failure to prioritize the most material threats to SVB’s solvency. But a detailed review of the report itself points to a decided lack of prioritization on Starling’s part. Starling devoted large sections of its report to matters that provide little value added. Meanwhile, the report glossed over other items that might provide greater insight into strengths and weaknesses in the Fed’s supervision. It’s a matter of prioritization.

The Starling Report uses the words “prioritize” or “prioritization” 21 times. As with any work, bank supervisors need to prioritize. Did SVB’s supervisors fail to prioritize? Perhaps. Interest rate risk, for example, clearly seemed to have received too little attention. The emphasis on prioritization also fits neatly into Vice Chair for Supervision (VCS) Bowman’s narrative that the real problem was that SVB had too many matters requiring attention (MRAs), which distracted the bank and its supervisors from focusing on more material matters.

Physician, Heal Thyself!

Now, let’s look at Starling’s report. It runs a whopping 365 pages, albeit with a lot of padding. And that’s only Phase 1, or what Starling calls the Probative Inquiry Preliminary Report. Starling expects completion of a Probative Inquiry Interim Report in late 2026, a Probative Inquiry Final Report in early 2027, and an Independent Expert Review & Recommendations for Reform by Q2 2027. In other words, seventeen months to review the failure of a single bank. Meanwhile, the VCS had already implemented significant changes before the review even started. It’s hard to see the relevance of the Recommendations for Reform other than to justify decisions made years earlier.

What gets too much priority?

The report took eight months and covered 365 pages and it’s still only in Phase 1. Some fat that could be trimmed or cut out entirely, as discussed below.

Investigating the role of social media. The report devotes considerable space investigating the role of social media in SVB’s collapse. Starling commissioned a review by Charles River Associates (CRA) to see if social media played a decisive role in SVB’s collapse. CRA concluded that it did not, at least “social media” defined as Twitter. Social networking, in the form of WhatsApp, emails, and other private channels clearly did. This may make an interesting academic observation but it’s hard to see what role the finding would have in bank supervision.

Interviews with SVB’s outside directors. Starling devotes 45 pages to interviews with former SVB officials and to a “statement” from SVB’s lawyered-up outside directors. The SVB officials especially dispute the statement in the Fed’s initial post-mortem (Barr Report) that bank management primarily viewed resolution of supervision as a “compliance exercise.” Their defense, however, emphasized workstream counts and use of high-priced consultants, which sure sounds like a compliance exercise. Starling noted the statements from SVB officials might be self-serving, but failed to note the FDIC’s current lawsuit against officers and directors of SVB for gross negligence.

Defending the former VCS. The Barr Report discussed changes to supervisory approach under former VCS Randal Quarles. It pointed to “pressure to reduce burden on firms, meet a higher burden of proof for a supervisory conclusion, and demonstrate due process when considering supervisory actions.” The Barr Report acknowledged, however, the difficulty in judging “how these collective changes in policy affected the oversight of SVBFG.” These tepid statements in the Barr Report generated what seems to be a disproportionate response from Starling.

Quarles role is well worth exploring further, since SVB’s rapid growth occurred under his watch. There are many ways to evaluate his role, but Starling chose to concentrate on the touchy-feely stuff, which included staff interviews and an extensive interview with Quarles himself. The Executive Summary asserts that “The failure of … supervisors to require SVB to reduce its interest-rate risk or liquidity risk did not result from a less assertive supervisory culture promoted by related supervisory policy priorities.” What Starling really found is an admittedly hard-to-prove suggestion in the Barr Report was in fact hard to prove.

What Gets Glossed Over?

Guidance on guidance. In 2018, the banking agencies issued the Interagency Statement Clarifying the Role of Supervisory Guidance (aka Guidance on Guidance). The agencies turned this statement into a formal regulation in early 2021. The statement and regulation emphasized that guidance does not create enforceable, binding legal obligations. The agencies also sought to discourage the use of numerical bright lines. These limitations are especially important when approaching IRR. The Fed does not have a regulation that addresses IRR. It only has an Interagency Advisory (guidance) from 2010 that identifies supervisory expectations for IRR management.

Starling makes passing reference to Guidance on Guidance but does not explore its impact on the Fed’s supervision of Starling. The new rules didn’t prevent examiners from issuing IRR-related MRAs (I issued a few myself during this period) but may have discouraged their use. Suppose, for example, that examiners did find SVB’s IRR to be excessive. What practical impediments would they face in taking action? Would Legal approve?

Systemic risk determinations. Tailoring legislation and the Fed’s implementation justified lighter touch supervision on the grounds that banks in the $50-250 billion asset category posed little systemic risk. Regulators reversed course in March 2023 by granting the systemic risk exception to both SVB and Signature Bank. I have previously described this decision as “an educated guess by three sleep-deprived septuagenarians.” But there were earlier signs of SVB’s systemic risk. FR Y-15 data as of year-end 2022 placed SVB tenth among US-domiciled banks, ahead of US Bank, PNC, and Truist. This raises the question of whether the FRB had any process in place to consider whether banks below specified asset thresholds posed elevated systemic risk and required more intense supervision. Starling did not address this question.

Use of horizontals. Starling mentions horizontal reviews several times in the report. These reviews look at exposures and practices across a range of banks. The report describes these reviews in a generally favorable light. For example, supervisors were “positioned to compare SVB with peer institutions through the Federal Reserve’s horizontal view of the banking system.” One SVB director claimed that “had supervisors clearly communicated that SVB was an outlier, the bank would have acted.” Nevertheless, VCS Bowman has substantially curtailed the use of horizontal reviews. The motivation for this action is unclear, except that the Bank Policy Institute complained. Did Starling consider whether elimination of horizontals might further weaken bank supervision?

The role of state regulators. SVB was a state-chartered bank, as was Signature and First Republic. One of VCS Bowmans key initiatives is to increase the Fed’s reliance on state regulators. The Updated Supervisory Operating Principles state that “examiners should rely to the extent possible on the state banking agency’s examination work and conclusions in the alternating years.” However, Starling never discusses the role of state supervisors or whether they strengthened or impeded effective supervision. The report’s Artifact Review includes a link to an internal review by state regulators but leaves it otherwise unaddressed.

The written record. Barr’s review helpfully included exam reports and supervisory letters. This allowed readers to assess the tone and content of the Fed’s supervisory actions. Such documents are missing from the Starling Report. The Starling review appears to rely almost entirely on interviews, supplemented by some public speeches and commentaries. The report doesn’t even discuss supervisory documents in any detail.

What’s missing? The report touches on matters requiring attention (MRAs) multiple times but provides little in the way of specifics. At some points Starling dismisses liquidity-related MRAs as “process-related” while later pointing out that these MRAs identified “weaknesses in the framework by which it intended to manage that profile under stress.” That’s not the only contradiction in the report. Starling correctly points out that the Fed never issued an MRA directing SVB to reduce the duration of its bond portfolio. Later in the report, however, it pointed out the roadblocks to either hedging or selling a held-to-maturity portfolio. In other words, Starling criticized supervisors for not taking actions that may have been impractical.

The Barr Report provided and discussed supervisory letters and other documents but not the bank’s responses. We can infer that bank management and board were unwilling or unable to address its problems in a timely fashion. But it is only an inference. Starling gives the SVB board an opportunity to provide its side of the story, which consists more of spin than specifics. They could address the issue more directly by reviewing responses to supervisory letters, plus whatever progress reports SVB provided to supervisors.

Starling uses the phrase “should have known” seven times and makes much of the Fed’s access to confidential supervisory data (CSI). But we don’t know what bank source data the Fed was receiving and Starling doesn’t tell us. From my own experience, banks supply regulators with only high-level IRR data and provide granular details only if we press them for it. Differences in key assumptions can also make cross-firm IRR comparisons especially challenging. The high-level data might still provide some meaningful information, such as SVB’s breach of its EVE limits. We should understand exactly what data the Fed was receiving from SVB before making broad characterizations of what supervisors knew or should have known.

Limitations on access and disclosure of CSI may have constrained Starling’s review in this area. If that is the case, Starling should make those limitations clear up front.

Getting the facts straight. Immediately after the Opening Letter, the Starling Report includes a “Summary Timeline.” The first item in the timeline, under “Rapid Growth,” states that “in 2021-22, SVB’s deposits grew from $63bn to $191bn and assets from $70bn to $207bn.” Well not exactly. Based on Call Report data, SVB’s total assets grew sharply from 2019-2021 but flattened out in 2022. Deposit levels dropped sharply in 2022.

At other points in the report, Starling more accurately attributes the growth to 2020-2021. But the focus on (non-existent) 2022 growth at the start of the report affected the framing of some issues. Current VCS Bowman defends Randy Quarles tenure by pointing out that he had “stepped down in October 2021, well before 2022, when SVB’s vulnerabilities became most apparent.” In fact, nearly all SVB’s growth occurred under Quarles’ watch.

Starling makes the curious statement that “the reason for these losses (on the bond portfolio) is that between year-end 2021 and March 10th 2023, the Federal Reserve raised the federal funds rate from near zero to over 4.5%.” But the bond portfolio consisted primarily of long-term mortgage-backed securities, where the fed funds rate plays little direct role. MBS rates started to rise more than a year prior to the increase in fed funds. An “expert” report that focuses largely on the supervision of interest rate risk should demonstrate a better understanding of how interest rates work.

Consider information that contradicts the preferred narrative. Several of the interviews point to problems with SVB’s transition into large bank supervision, with one staff member describing it as “about as bad as we could have done it.” It’s hard to see how this doesn’t reflect poorly on the Fed’s implementation of tailoring rules or Quarles’ stewardship. Yet problems with the transition never make it into the report’s executive summary. Similarly, a look at staffing levels might contradict Starling’s prioritization narrative. Examiner hours devoted to SVB were less in 2021 than in 2017, even though the firm’s assets quadrupled during this period. Starling reports SVB’s staffing levels, but not those for the FRB.

The Selection Process for this Review. This is more a question for VCS Bowman than for Starling, but nonetheless a relevant one. Did the selection follow a normal procurement process? How much will Starling receive for this report and does the compensation include the next three phases of the project? Sen. Elizabeth Warren has already raised questions in this regard and has requested a review by the Fed’s Inspector General. Some of Warren’s accusations look a little over-the-top but she does raise some serious issues, especially in light of VCS Bowman’s supposed commitment to transparency.

One item that remains unclear to me is the level of expertise that Starling brings to the review. The report notes Starling’s experience “curating a global dialogue aimed at advancing culture risk governance and supervision,” whatever the hell that means. What seems to be missing is any experience in bank supervision, either directly or through a regulated bank or regulatory consulting firm.

We should rightly be wary of appeals to authority and impressive resumes don’t necessarily translate to a quality product. Some of the interviews make interesting reading and Starling identifies some of the Fed’s cultural impediments to taking decisive supervisory actions. But hard-headed, evidence-based analysis seems lacking. Starling’s delegated much of the analytical heavy lifting to CRA. That includes not just the analysis of the role of social media, but also a financial assessment of SVB’s vulnerabilities. It’s unclear why Starling needed another outside expert to conduct the financial analysis, since such analysis should be central to a failed bank post-mortem. That delegation, along a deemphasis on the written documents point to the limitations of the review.