Andrew Harnik/AP Photo
Federal Reserve Board of Governors Vice Chair for Supervision Michael Barr testifies at a House Financial Services Committee hearing on March 29, 2023. Barr was selected to lead the central bank's review of its supervision of Silicon Valley Bank.
The Federal Reserve’s internal review of the collapse of Silicon Valley Bank was released on Friday, along with a number of supervisory materials that the public usually does not get the chance to see. They revealed an operation that’s simply unable to deal with challenges that don’t move at a glacial pace. If you read them closely enough, they’re an indictment of the Fed’s ability to supervise any bank.
I have said all along that everybody did something wrong in the SVB debacle: the bank executives, the auditors, the rating agencies, the stock analysts, the regulators, the Congress that passed the law defanging the regulators, the Federal Reserve’s Board of Governors who spiked interest rates, and even the leading depositors who unthinkingly put hundreds of millions of their dollars into one bank account, for some insane reason. When everyone’s to blame, you usually get an epic amount of finger-pointing, which effectively prevents anyone from really taking responsibility.
There’s a little of that in the Fed report too. Michael Barr, the current vice chair of supervision, stresses the role of his predecessor, Randal Quarles, in defanging the entire process. And the report definitively assigns a role to Congress’s 2018 law easing rules on large regional banks, making clear that the bank would have failed enhanced liquidity tests if those requirements had not been removed.
But while there was good reason to be skeptical that the Fed could be self-critical, the larger picture the report paints quietly condemns the entire project of Federal Reserve bank supervision. The Fed will likely never recommend eliminating that piece of their authority altogether. But any independent analysis would have to consider that conclusion.
For example, at the time of Silicon Valley Bank’s collapse, it “had 31 unaddressed safe and soundness supervisory warnings—triple the average number of peer banks,” the report states. Obviously, SVB didn’t see those warnings as, well, warnings, but merely as helpful suggestions that could be filed away. The fact that the average number of supervisory warnings is more than ten suggests that the level of regulatory neglect applies to the entire industry; a culture of ignoring regulatory concerns can be inferred.
At the same time, the report’s evidence shows that those 31 warnings clearly underplayed what was happening at SVB. To review, SVB grew rapidly after the 2018 law, which raised the thresholds for many forms of enhanced supervision from $100 billion to $250 billion, and after Quarles “tailored” regulations for banks in that size range, essentially by weakening them.
SVB had a large client base of wealthy depositors who routinely exceeded FDIC deposit insurance limits, making them a flight risk in the event of trouble. And the bank invited trouble by throwing a majority of its assets into long-dated securities that would fall in value if interest rates ever rose.
But even though Fed supervisors first identified interest-rate risk deficiencies at SVB in 2020, they issued no formal warnings, only verbal guidance and lower-level “advisories” for the ensuing two years. SVB initially thought it would make money from interest rate increases, and only changed that view in October 2022. After telling supervisors about this, the supervisors issued one formal warning in November 2022, and then planned but never executed a downgrade of the bank’s Sensitivity to Market Risk rating, part of the overall CAMELS score that assesses a bank’s overall health (sensitivity to market risk represents the S in CAMELS).
This dithering seems structural. The report states that, because SVB got a positive score in 2017, when it was a much smaller bank, that set a precedent for how it was perceived years later, even after a new supervisory team entered in 2021. Moreover, a particularly Keystone Kops section of the report explains that the regional Fed banks (in SVB’s case, the San Francisco Fed) had explicit authority to downgrade CAMELS ratings, but in practice they would try to get consensus from the national Fed’s Board of Governors before doing anything. Meanwhile, any enforcement action—say, a sanction for having 31 unaddressed safety and soundness warnings—has to go through the Board of Governors’ staff, slowing things down further.
The Fed Board of Governors has input in the supervisory process, and clearly during Quarles’ tenure their mantra was to lay off the banks. “Staff repeatedly mentioned changes in expectations and practices, including 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 report notes. In other words, the bosses paralyzed the supervisors from reacting to what was right in front of them.
One piece of supplemental material stands out. The Board of Governors was shown a presentation in February 2023 on the impact of rising interest rates on banks. One slide specifically featured Silicon Valley Bank, saying it has “significant interest rate risk.” It notes that SVB’s unrealized losses were massive, highlights an existing loss of depositors even before the March 2023 run, and states that SVB’s internal risk measurement systems were faulty. All this was known well before the collapse. And then, the slide says that supervisors had downgraded SVB, when in fact they hadn’t; that was the planned downgrade that never happened before the bank collapsed.
Maybe in 2036 there will be another report about another collapse where the supervisors knew it was coming but failed to act.
This is the bank regulatory version of Waiting for Godot; everybody knows something is wrong and cannot bring themselves to do anything about it. An organization that has set up such an inert structure should act to dismantle it.
Another structural hurdle is the phase-in period. SVB passed $100 billion in assets in June 2021. Even under Quarles’ weakened tailoring rules, that was supposed to add some additional scrutiny. But there were transition periods in the rules, designed so a bank could gradually meet compliance. The first supervisory stress test wasn’t scheduled until 2024, three years after its assets had crossed that regulatory line. Yet SVB “repeatedly” failed internal liquidity stress tests as soon as it first had to impose them, in July 2022. Yet supervisors gave SVB a satisfactory rating on liquidity just one month following that first internal failure. SVB, for its part, changed its own stress test assumptions, making it easier to pass. Because it was an internal stress test, they could simply game it.
Under the old rules, SVB would have been subject to full liquidity requirements. There has been a lot of chatter, mostly from bank lobbyists, about how the bank would have easily met this requirement. The Fed report shows that to be untrue. According to one table, SVB would have missed the liquidity requirement every month for a full year, from March 2022 to February 2023, in some cases badly. Under the old rules, that would have required the bank to obtain high-quality assets, up to $14 billion by the last report in February 2023. SVB’s capital requirements were also deficient, because the bank was allowed to opt out of recognizing unrealized losses, something it wouldn’t have been able to do if the Fed hadn’t changed the rules in 2019.
So the SVB debacle was not a freak event brought on by a digital bank run. SVB was in a bad position, everyone saw it was in a bad position, and its bank examiners were not only reluctant but, it appears, institutionally incapable of acting. The system, in other words, simply prevents regulation from working. There are way too many veto points, too much internal reluctance and inertia.
One of the more amusing materials in the report is a 2010 report evaluating supervision in the wake of the global financial crisis. Among several problems, that report stated that supervisors were good at identifying deficiencies but not good at demanding corrective action or holding bank executives accountable. That’s exactly what went down in the SVB collapse 13 years later. Maybe in 2036 there will be another report about another collapse where the supervisors knew it was coming but failed to act. Until we see the problem as systemic to the Fed supervision process, nothing will change.