By Katie Brown
ABA DataBank
Discover more in-depth research, dashboards and webinars from the Office of the Chief Economist by exploring ABA’s Economic Research and Insights website.
The spring 2023 failures of Silicon Valley Bank (SVB), Signature Bank (SBNY) and First Republic Bank (FRB) were the second, third and fourth-largest bank failures in U.S. history — and each one was precipitated by a damaging bank run. The run at SVB was the largest in U.S. history, and the runs were novel in that it was the first time clients had run in a time of advanced digital banking and money transfer solutions.
Most of the commentary at the time and since has focused on the balance sheet management, risk management and supervisory decisions that contributed to each bank’s distress. Less attention has been paid to the runs that accelerated as the banks struggled and that ultimately led to the failures. Earlier this year, the FDIC released a staff report analyzing data collected during and after three major bank failures in spring 2023. Based on access to a unique set of FDIC data, the report provides exceptional insight into the failures of SVB, SBNY and FRB.
The analysis focuses on data from March 6 to March 24, 2023, a period that encompasses the runs at all three banks, the failures of SVB and SBNY, and the establishment and operation of Silicon Valley Bridge Bank and Signature Bridge Bank. This article highlights six key findings, each of which can inform banks’ deposit strategy and provide directions for future research.
1. Deposit bases were concentrated and heavily uninsured
Before the runs, each bank held a different mix of deposit types, based on the kind of customers it served. Despite these differences, business deposits (accounts where a business was the beneficial owner of the funds) were the largest category of deposits at all three banks. SVB had the most concentrated amount of business deposits, at more than 90% of its total deposits, followed by SBNY and FRB, where business deposits made up about half of total deposits.
At SBNY, escrow deposits were the next largest deposit type. Passive escrow deposits (money a business holds on behalf of others, like mortgage payments or apartment security deposits, that the actual owners can’t withdraw themselves) made up 11-15% of total deposits. Active escrow deposits, consisting primarily of customer funds from investment companies and banking-as-a-service fintechs, made up 13-15%. At FRB, the next three largest deposit types behind business deposits were consumer deposits, trust deposits and active escrow deposits, a mix consistent with the bank’s focus on high-net-worth individuals.
Estimated deposit insurance coverage varied by deposit type. The vast majority of business deposits were uninsured at all three banks, and a significant share of consumer and trust deposits were uninsured at FRB and SBNY. Virtually all active escrow deposits at SBNY and FRB were uninsured. The study assumed these escrow deposits and brokered CDs were fully insured, since FDIC coverage passes through to the underlying account owners. Consistent with publicly reported figures, SVB had the highest share of uninsured deposits at 94% of domestic deposits prior to the run, compared with 72% each at FRB and SBNY. Prior to the run, 34% of the people and businesses with accounts at SVB held at least some uninsured funds, compared with 13% at FRB and 19 percent at SBNY.
A small group of top depositors, the top 0.5%, held a large share of deposits at all three banks. SBNY had the highest concentration: fewer than 600 top depositors held 62% of total deposits. At SVB, fewer than 500 top depositors held 38% of total domestic deposits. At FRB, 3,000 top depositors held 50% of total deposits.
2. The deposit runs were unprecedented in speed and magnitude
Between March 7 and March 17, the period when all three banks experienced the most stress, deposit outflows were substantial, and the vast majority of top depositors ran. The FDIC compared each day’s withdrawals to the bank’s deposit total on March 6 and counted a depositor as having run if they pulled out 75% or more of what they held that day.
Each bank had one day during the stress period when net deposit outflows reached 20% or more of March 6 deposits and 20% or more of top depositors ran. For SVB, that day was March 9, the day before it failed where 23% of depositors ran, and the bank lost about 20% of its March 6 deposits, a net outflow of $30.2 billion. At FRB, this level of outflow and top-depositor flight occurred on March 13, after the failures of SVB and SBNY.
Wire transfers (mainly via Fedwire or SWIFT) accounted for about 80% or more of net deposit outflows at all three banks. Looking at where depositors sent funds, at SVB about one in five depositors exclusively wired deposits to accounts they’d used in the past six months. At SBNY and FRB, about one in three did the same. At all three banks, about half of depositors wired to accounts with no prior six-month history. This suggests further research is needed on the role of clients’ other banking relationships on run behavior.
The deposit runs were unprecedented in speed and magnitude
3. Runs continued after the establishment of bridge banks and the systemic risk exception
SVB and SBNY each had two of their sharpest deposit outflows happen after their bridge banks opened and after the government announced it would guarantee all deposits at both banks via the systemic risk exception. On its first day in operation, SVB’s bridge bank lost 14.4% of its March 6 deposits, an outflow of $21.5 billion, and 22% of their top depositors fled. Similarly, SBNY’s bridge bank experienced its largest outflow on the first day of its operations and lost 22.4% of its March 6 deposits, an outflow of $20.2 billion, with 23% of its top depositors running.
Wire transfer data confirmed the majority of these deposit outflows were the result of new transfer requests made after the establishment of the bridge banks and the systemic risk exception and were not the result of a delay in processing from the week before.
4. Large and uninsured depositors drove the runs
The biggest depositors at all three banks were substantially more likely and quicker to run than the average depositor, a pattern that held even after accounting for the fact that they also tended to have more uninsured money at stake. At all three banks, the likelihood of running rose with a depositor’s March 6 balance, with sharp increases at the highest percentiles. SVB saw a sharp increase in run rate at the top percentile of depositors while SBNY saw sharp increases at the 75th, 90th and 99th percentiles, and FRB saw similar increases at the 90th, 95th and 99th percentiles. Since many of the largest depositors were financial companies — which the FDIC defines as including depository institutions and their affiliates, insurers, registered investment advisers, hedge funds and private equity funds — the FDIC compared how often financial companies pulled their money out versus how often nonfinancial companies did. Among the top 1% of depositors, financial companies ran at higher rates than nonfinancial companies. Both groups, however, ran at higher rates than depositors who ranked just outside the top 1%, in the next percentile band down.
Two things mattered most in predicting whether a business pulled its money out: how much of its deposits were uninsured, and whether it was one of the bank’s biggest depositors. At SVB and SBNY, being a top depositor increased the chance of running even when a company’s share of uninsured deposits was factored in separately.
How long a business had banked there, how many accounts it held, and how its deposits were structured also mattered, but less than size did. Businesses that had banked there longer were less likely to run: their chances of running were about a third lower than the average business’s chance of running. Customers with four or more open accounts were half as likely to run as customers with only one. Businesses tied to the crypto and digital asset industry were significantly more likely to run at SVB and SBNY.
The results show that large and uninsured depositors drove a large share of the outflows, and they withdrew funds faster and more completely than in previous bank runs.
Consumers were somewhat less likely to run than businesses, but the same two factors, deposit size and uninsured share, remained the strongest predictors for them too. In fact, consumers were even more sensitive to their insurance status than businesses were. How long consumers had banked there, how many accounts they held, and other relationship factors mattered much less for predicting their behavior. The study notes that these factors don’t fully explain run behavior, which may also vary based on each bank’s specific business model and mix of depositors.
These results suggest that depositor incentives, especially among large uninsured account holders, can overwhelm traditional relationship-based sources of funding stability during periods of stress.
5. Most major deposit categories experienced substantial outflows
Business deposits, the largest type of deposit at all three banks, dropped sharply between March 7 and March 17. SVB and SBNY each lost about 60% of their business deposit balances from March 6. FRB lost 36%, but that figure includes a $30 billion deposit that a group of large U.S. banks made into First Republic in mid-March to help stabilize it. Excluding that one-time infusion, FRB’s business deposits actually fell by 70%.
Escrow deposits at SBNY also fell sharply. Passive escrow deposits dropped 57% and active escrow deposits 88%. Active escrow accounts at FRB fell 52% over the same period. Consumer deposits, typically a stable funding source, also fell as SBNY lost more than half its consumer deposits, while FRB lost about a third.
6. Deposit insurance helped stabilize fully insured retail deposits
Fully insured retail deposits, meaning everyday consumer accounts fully covered by deposit insurance, did not run. These deposits held roughly steady at SBNY and actually grew at SVB and FRB, supporting the conclusion that deposit insurance helps keep deposits stable.
By contrast, when you look at deposits based on whether they were insured or not under normal FDIC limits (separate from the fact that the government later guaranteed every deposit at SVB and SBNY), the uninsured deposits left at very high rates. Between March 7 and March 17, SVB and SBNY saw uninsured outflows of 62% and 68% of their March 6 balances. FRB’s uninsured deposits fell 47%, or 71% excluding the $30 billion consortium deposit described above.
Insured deposits also declined, but less steeply: SVB lost 30%, SBNY lost 33%, and FRB lost 7%. At SBNY, a significant share of these outflows came from passive escrow accounts, which the study assumed were fully insured through pass-through coverage. Notably, many depositors held both insured and uninsured funds, and when they ran, they typically withdrew both.
Conclusion
This review provides unique insights into the largest bank runs in U.S. history. The results show that large and uninsured depositors drove a large share of the outflows, and they withdrew funds faster and more completely than in previous bank runs.
In addition, all three banks had a small group of depositors controlling a large share of total deposits. Whether a business had a lot of uninsured money and whether it ranked among a bank’s biggest depositors were better predictors of whether it would run than factors like how long it had banked there or how many accounts it held. The runs kept happening even after the banks failed, bridge banks were set up and the government guaranteed all deposits, and most of the money moved out through wire transfers. However, deposit insurance helped keep retail depositors from running, and fully insured retail deposits stayed stable or even grew throughout the stress period.
Katie Brown is director, economic research, with ABA’s Office of Economics.
