In the wake of recent bank failures, all eyes have turned to liquidity risk — the risk of a bank run should depositors decide to pull their funds. Regulators are increasing the frequency and intensity of their inquiries into each bank's funding mix, focusing on buckets of deposits that have long been categorized, in regulation and supervisory practice, as "hot money" and more likely to flee.
In this moment, greater focus on liquidity risk is obviously critical. The problem is that new data points from a bank that recently failed suggest that longstanding presumptions about the relative stickiness of various buckets of deposits may well be dead wrong. If these data points are accurate, increased supervisory focus, despite best intentions, may inadvertently be pushing banks to increase dependence on deposits that are actually riskier.
Industry and regulatory presumptions about the riskiness of types of deposits are grounded in the savings and loan crisis experience of 40 years ago. Like the recent bank failures, that crisis was triggered by sharp Federal Reserve rate increases to fight inflation. Certain buckets of deposits — such as brokered deposits, time deposits and interest-bearing deposits — were seen as much more likely to run out the door, while other deposits — including transaction accounts, retail deposits generally and noninterest-bearing deposits — were seen as much more stable and desirable.
Evidence from contemporaneous depository failures supported those presumptions, including the relative run experienced by IndyMac Bank between March and September 2008 during which brokered deposits fled at more than twice the rate of retail deposits, time deposits at about three times transaction accounts and interest-bearing deposits at about two-and-a-half times noninterest-bearing accounts.
Based on these painful experiences, regulators have subsequently driven banks to shift funding away from deposit buckets seen as riskier toward those seen as safer. This can be seen explicitly in regulations such as the Liquidity Coverage Ratio (which directly penalizes disfavored types of deposits) and the FDIC's risk-based deposit insurance assessment calculation (in which brokered deposits can drive sharp increases in a bank's costs). More broadly, the supervisory process, including a focus on what are termed potentially volatile funding sources (which includes the buckets historically seen as risky), directly shapes the mix of funding used by banks of all sizes.
A lot has changed in banking since the 1980s, including technology-driven innovation that now permits depositors to manage multiple accounts across multiple providers and to move money on a smartphone. Gone are the historical sources of friction that may have reduced the pace and magnitude of outflows of certain types of deposits in the past: no need to wait in potentially tortuous lines at a physical branch or to concentrate accounts at a single institution. Such fundamental shifts are reshaping long-held presumptions in sectors from retail to media. In that light, it is reasonable to ask whether these changes may be having similarly fundamental impacts on long-held presumptions of the relative stickiness of various types of deposits during this period of banking stress.
Federal Reserve Vice Chair for Supervision Michael Barr said he is overseeing a six-month project to overall supervisory culture, practices, behavior and tools. He said regulatory changes will also be explored.
Policymakers have access to deposit data across the industry. Sadly, most such information is not yet publicly available. For Silicon Valley Bank, as an example, we have the "before" picture from its fourth quarter 2022 financial reports but no "after" snapshot since the bank failed before quarter-end.
By sheer happenstance, First Republic Bank was clearly under stress during the first quarter but was not formally seized until May 1, after it had filed first quarter financials reflecting that stress. So, we have both "before" and "after" snapshots — and what that data shows about how different types of deposits are actually performing in the modern world is strikingly at odds with longstanding industry and regulatory presumptions about how those deposits "should" behave. Brokered and similar deposits, presumed to be the epitome of "hot money," were actually more than eight times stickier than favored retail deposits. Time deposits were more than twice as sticky as transaction accounts. And noninterest-bearing deposits ran out the door 30% faster than the presumably problematic interest-bearing deposits.
The First Republic experience could turn out to be an aberration. But its lessons are reinforced by indicators of the same dramatic shifts in relative stability of different types of funding in the call reports and public filings from the first quarter: Transaction, retail and noninterest-bearing deposits all materially underperformed types of deposits traditionally seen as riskier "hot money."
Banks depend on stable deposit funding. And while regulators are appropriately now intensely focused on pushing banks to increase the stability of their deposits, these recent data points suggest we may well be inadvertently pushing these institutions toward riskier and less stable balance sheets.
At a minimum, this data calls for an intense analysis of deposit behavior based on the far more robust data available to the government; such analyses and data should be shared broadly. If the shifts that seem to be taking place are proven out, it is critical and urgent to revise both industry and regulatory presumptions and rules to reflect the ever-evolving reality of financial risk management for banks.
