New research suggests bankers, regulators and policymakers are learning the wrong lessons from the large bank failures of 2023.
In a
Instead, they say the real driving force for the collapses of Silicon Valley Bank, Signature Bank and First Republic Bank was their respective business models, which were faltering in the face of macroeconomic headwinds.
"We describe the crisis as a reaction to bank business models that focused on providing banking services to certain economic sectors, crypto-asset firms and venture capital, that had come under economic pressure during the preceding year," the authors write. "We argue this view of the crisis provides a more precise explanation of which banks were affected compared to an explanation focused solely on banks' balance sheet metrics."
The authors — Steven Kelly, associate director of research for Yale's financial stability program, and Jonathan Rose, the Fed's in-house historian and a senior economist at the Federal Reserve Bank of Chicago — say recasting the episode as one driven by distinct choices made by banks, rather than generalized mismanagement of risks, could have a meaningful impact on regulatory and industry responses.
"With this set of facts in mind, we draw very different policy conclusions than the standard account," they write, offering distinct proposals for new liquidity requirements and supervisory practices.
The report comes two years after the failure of Silicon Valley Bank, which most accounts point to as the instigating event for the banking volatility of 2023. The reverberations of that moment are still impacting policy considerations today.
In a speech
"The supervisory failures at the heart of the 2023 banking crisis under President Joe Biden should have been a wake-up call," Bessent said. "As the Fed's review noted, its supervisors did not fully appreciate Silicon Valley Bank's vulnerabilities as a group, in size and complexity. When risks were identified, they did not take sufficient steps to ensure that SVB fixed those problems quickly. The result was the third largest bank failure in United States history. It was a supervisory failure."
Like other conservative policymakers, Bessent argues that supervisors should have acted more forcefully to get SVB to address its interest rate risk exposures — specifically its unhedged holdings of long-dated securities that had incurred paper losses after the Fed increased interest rates. This argument underscores a belief that the driving force behind the run on deposits was a shock of customer concern after the bank was forced to sell assets at a loss to cover drawdown requests.
But Kelly and Rose say that is an oversimplification of what transpired and leaves out key details.
The paper argues that unrealized losses were prevalent throughout the banking sector at the time, but few banks were in jeopardy of failure. Aside from the practical challenges of supervisors cracking down on hundreds — if not thousands — of banks with paper losses, Kelly and Rose note that there was not a strong rationale to do so. They found no historical evidence that such losses are a reliable indicator of imminent distress, even when paired with the other often-cited factor in SVB's demise: high levels of uninsured deposits.
"These metrics would have flagged only some of the banks that did experience runs in 2023, while also implicating several that did not," they write. "Moreover, neither banking history (inclusive of 2023) nor economic reasoning suggests that banks need to have SVB-like levels of uninsured deposits or asset losses to experience market discipline."
While the failure of Silicon Valley Bank did create a contagion effect throughout other large regional banks — in the form of declining stock values and, in some cases, increased deposit outflows — the paper identifies only six banks acutely impacted by the crisis: SVB, Signature, First Republic, Pacific West Bank, Western Alliance Bank and Silvergate Bank.
According to year-end financial disclosures from 2022 compiled for the paper, Kelly and Rose found that SVB and First Republic were the only banks in that cohort with unrealized losses and uninsured deposits meaningfully above industry averages. Western Alliance was slightly above the industry in both categories while the others only exceeded one of the two. They also cite broader analysis that of the 22 banks that experienced runs in March 2023, no more than three were in the bottom 10% of the industry in terms of both unrealized losses and uninsured deposits.
A clearer commonality between the six banks, the paper argues, is their exposure to risky depositor bases that other banks shied away from: crypto firms and venture capitalists. Kelly and Rose note that both sectors had been in decline throughout 2022. Crypto investment had been waning since its peak Nov. 2021, and VC firms were struggling to both raise fresh capital and exit from investments. These conditions forced firms to draw on bank deposits to cover operating costs.
This development played out relatively quietly, the paper notes, but did not go unnoticed by analysts, investors or the financial press. The issues hit a critical moment when SVB — the leading VC-focused bank in the country — announced
Postmortem reports about the failures of SVB, Signature and First Republic found that most of their deposit outflows came from their largest depositors, executed via wire transfers. Kelly and Rose argue that this cuts against the narrative that the runs were, in any meaningful way, the result of retail investors requesting withdrawals after seeing social media posts.
"It's internally inconsistent to suggest the 2023 crisis was both a run of uninsured depositors — that is, those with cash balances above $250,000 — and centered around retail apps and social media," they write. "Bank runs are institutional and led by large corporate depositors, necessarily so since deposit holdings are concentrated in their hands."
Rather than changing liquidity requirements in anticipation of widespread retail shifts, Kelly and Rose argue that a more appropriate shift would be to treat high-net worth customers more like institutional ones, noting their abilities to move large volumes of deposits between institutions relatively quickly.
Overall, they say, regulators would be wise to craft specific policies for dealing with business models that are highly exposed to cyclicalities. This could be done through business model-specific capital charges — in line with so-called Pillar 2 provisions that are common in Europe and elsewhere — or through precise supervisory guidelines.
"Supervision can perhaps be a more precise tool in this space than a mechanical capital requirement," they write. "With more engaging supervision of business models — and the macroeconomics thereof — supervision can develop a greater understanding of which banks have strong enough franchises to bear unrealized losses and which do not."
