Discover more from Posts in the Shell by Demren Sinik
Take the Money and Run
Bank Runs in the Digital Age
Note: this essay was previously published in March 2023
2023 has been a year full of milestones, and not particularly happy ones at that. In a little over four months, we've witnessed an unprecedented global banking crisis, whose causes we will be analyzing for decades and whose consequences are still not fully known. Nobody, except perhaps for quick-fingered Wikipedia editors, were excited to see three new additions to the list of the largest bank failures in U.S. history - First Republic Bank, Silicon Valley Bank, and Signature Bank - which have claimed the ignominious positions of second, third, and fourth place. Unfortunately, the turmoil and fear seem far from over.
It has been a long time since the last major banking crisis during the Great Recession. Consequently, everything about 2023 feels “new,” especially to a younger generation of professionals who came of age in a world subsidized by low interest rates and free capital. According to consensus, this crisis is new because it has occurred in the digital age, an era where people can derail a financial institution from the comfort of their couch, rather than physically running to their local neighborhood branch and demanding cash.
But just because something feels new does not mean it is new. That's the big question for this post. Is the 2023 banking crisis actually new? Is it actually unprecedented? Are the failures of First Republic, SVB, and Signature so unique in the context of our most infamous financial institutions?
Five (In)Famous Case Studies
To answer these questions, we'll start by looking at five major bank runs in U.S. history - each from a different time period and under different macroeconomic and microeconomic constraints. Apart from SVB, each bank in this list set the record for the largest bank failure in U.S. history at the time.
The Bank of the United States (1930)
What was it? The Bank of the United States was the largest commercial bank in the U.S. by deposits and customers in 1930. At the time of its failure, the bank held $268M in deposits (~$5bn today) and served over 440,000 customers, who were primarily working-class immigrants living in New York City. The bank run on the Bank of United States' Bronx branch is often cited as the trigger that initiated the collapse of the banking system during the Great Depression.
Why did it fail? After undergoing rapid expansion in the late 1920's fueled by a handful of mergers, the Bank of the United States faced a handful of legal difficulties. In conjunction, a series of smaller bank runs in the southeast of the U.S. as well as the collapse of the stock market set the stage for an atmosphere of fear. According to historians, when a Bronx businessman was unable to sell his Bank of United States stock back to a bank manager in the Bronx branch, the businessman spread a (false) rumor that the bank could not afford to repurchase the stock. The rumor spread like wildfire, prompting a massive bank run with over 20,000 customers queued to withdraw funds at the Bronx branch alone. The following day, the bank was temporarily closed, and then permanently shuttered when support financing and a merger offer failed to materialize. Ultimately, the bank was illiquid, not insolvent - and depositors eventually received 93 cents on the dollar for their deposits when all was said and done.
What was the timeline? On the morning of December 10th, 1930, a Bronx businessman spread the rumor that the Bank of the United States lacked liquidity. That afternoon, over 20,000 individuals lined up to withdraw from the Bronx branch. The next day, December 11th, the bank directors closed the bank in anticipation of a massive run and gave the New York Superintendent of Banks control over its assets. When a merger was unable to be negotiated, the Bank of the United States was suspended on December 13th, 1930.
Franklin National Bank (1974)
What was it? Franklin National Bank was a bank based in Long Island, New York that failed in 1974. At the time of failure, the bank held ~$3.7bn in assets (worth $21.bn today).
Why did it fail? It is impossible to do justice to the lurid failure of Franklin National Bank in a single paragraph. The long version of the story involves the mafia, the free masons, the Nixon administration, drug cartels, a cyanide pill, and an incredible number of broken laws. In short, however, an extremely shady individual named Michele Sindona purchased a controlling stake in Franklin and, through a series of poor and often illegal decisions was responsible for many losses at the bank, prompting a bank run and ultimately insolvency.
What was the timeline? In early May 1974, Franklin National Bank disclosed to the public significant losses in foreign exchange markets in excess of $60M. Over the next 14 days, deposits declined by ~$325M, or ~11% of total deposits. To cover withdrawals, the bank borrowed over $1bn from the Federal Reserve Bank in May. However, deposit outflows continued for the next several months, reaching a ~50% decline from peak at the start of September 1974. On October 9th, 1974, Franklin National Bank was declared insolvent.
Continental Illinois (1984)
What was it? Continental Illinois is the fifth largest bank to fail in U.S. history. Continental Illinois was a commercial bank headquartered in Chicago that was the seventh largest commercial bank in the U.S at its peak. At the time of its failure, the bank held $40bn in assets (worth $116bn today).
Why did it fail? Following a decade of rapid growth in the 1970's, Continental Illinois made a series of risky investments that proved its downfall. Due to an old regulation that prevented banks from expanding beyond state lines, banks often purchased loans from other banks across states. In this case, Continental Illinois purchased $1bn in risky energy loans from Penn Square Bank, which was based in Oklahoma. Continental Illinois also owned risky assets in several developing countries. Soon after, Penn Square Bank failed, and there was an international debt crisis spurred by Mexico's default. The bank attempted to rightsize its balance sheet, but fear sparked a bank run on the bank's deposits, the majority of which were uninsured. It’s also worth noting that ~40% of Continental Illinois' deposits were foreign leading up to the failure – and historians believe these creditors were more likely to withdraw relative to domestic creditors.
What was the timeline? Continental Illinois' troubles came to light in a poorly received earnings release in April 1982. This release, combined with the failure of Penn Square Bank (July) and Mexico's default (August), resulted in a slew of credit and stock downgrades. In late April 1984 (two years later), the bank publicly announced an increase in nonperforming loans (i.e., a loan that is unlikely to be repaid in full). On May 7th, 1984, rumors arose that Continental Illinois would fail or would be forced to seek a merger. A 9-day bank run commenced on May 10th, 1984 in which 30% of Continental Illinois' existing funding fled, and was replaced by a significant amount of federal assistance from the FDIC as well as support from a coalition of private banks. The government effectively owned and ran Continental Illinois until it finally disposed of its shares in 1991, and the bank was purchased by BofA in 1994.
Washington Mutual (2008)
What was it? Today, Washington Mutual (aka WaMu) is remembered as the largest bank failure in U.S. history - both on a nominal and inflation-adjusted basis. Prior to its collapse in 2008, WaMu was the sixth largest bank and the largest Savings and Loan Association in the U.S. Unlike a traditional commercial bank, Savings and Loan Associations primarily focus on residential mortgages (in addition to other traditional banking services). At time of failure, WaMu held $307bn in assets (equivalent to ~$422bn today).
Why did it fail? To some extent, the failure is self-explanatory - it was 2008 and WaMu specialized in residential mortgages. But to be more specific, WaMu had significant exposures to subprime mortgages because they catered to higher risk consumers and approved variable rate loans despite a lack of provable income and assets. Additionally, WaMu had a lot of business in California (where housing fared worse than average) and had rapidly expanded its physical presence in the years leading to the crash. When housing prices fell, WaMu wrote down billions in defaulted mortgages and struggled to raise cash, which was exacerbated by the collapse of the secondary market for mortgage-backed securities. The final nail in the coffin was a bank run led by depositors, who withdrew $16.7bn over the course of ten days. The FDIC determined that WaMu lacked sufficient funds and seized the bank. JPM eventually stepped in and assumed WaMu's assets.
What was the timeline? Housing prices peaked in early 2006. By December 2007, average home values had declined 6.5% from the prior high. The same month, WaMu wrote down its home lending unit by $1.6bn, cut 6% of its workforce, and closed ~50% of its home-loan offices. In 2008, the Bear Sterns takeover was negotiated in March and IndyMac Bank failed in July. In the first two weeks of September, Fannie Mae and Freddie Mac were taken over by the government, Merill Lynch was sold to BofA amidst a liquidity crisis, and Lehman declared bankruptcy. The latter, which occurred on September 15th, 2008, kickstarted the ten-day long bank run on WaMu, culminating in its failure. Most of those deposit withdrawals were retrieved via internet-based electronic banking and wire transfer.
Silicon Valley Bank (2023)
What was it? SVB was a commercial bank founded in 1983 in Santa Clara, California that specialized in banking technology companies. As of 2021, SVB claimed to bank ~50% of all US venture-backed startups and had relationships with a meaningful number of venture capital firms. At the time of its failure, SVB held ~$209bn in assets, positioning it as the third largest bank failure in U.S. history as of writing.
Why did it fail? Prior to the technology downturn that started in November 2021, SVB had both grown rapidly and catered to a specific clientele. Like any bank, SVB invested client deposits in return-producing securities, like bonds. When the federal reserve rapidly raised interest rates to combat inflation, the current value of those bonds declined. Why? Because if you want to sell a bond that pays a low interest rate relative to market, you will need to sell that bond at a discount to convince someone to buy it. That doesn’t really matter if you are planning to hold the bond until maturity. Unfortunately, SVB’s customer deposits were dwindling because startups were not raising more capital and were continuing to burn through existing cash. To cover a decline in deposits, SVB sold securities from their portfolio for a $2bn loss and would undertake a share sale to raise further cash. Instead of rightsizing their balance sheet, SVB instead spooked the market – its share price collapsed, and existing customers rushed to pull out their deposits. Soon after, the California Department of Financial Protection and Innovation took possession of SVB.
What was the timeline? In mid-2021, a surge in inflation impacted the global economy. On March 16th, 2022, the Federal Open Market Committee enacted the first of nine rate hikes, increasing the federal funds rate from 0.25% to 5.25% as of writing. By December 31st, 2022, SVB had unrealized losses of ~$15bn for securities held to maturity. On March 8th, 2023, SVB announced it had sold $21bn of securities at a ~$2bn loss and was further planning to raise capital from a stock sale. On March 9th, SVB stock declined 60% and customers ran on the bank, initiating ~$42bn in withdrawals. On March 10th, regulators took control of SVB, shutting it down.
Bank runs are as much a function of perception and psychology as risk: Playing “what if” is a fool’s game in history – that being said, it’s not immediately obvious that all failed banks would have failed without a bank run on deposits. Customer confidence may have saved the Bank of the United States, Continental Illinois, and SVB (Franklin Illinois and WaMu, on the other hand, had serious balance sheet problems). Skeptics scoffed at the initial SVB fears, pointing towards recent balance sheet data demonstrating solvency. Nonetheless, liquidity disappeared when enough customers demanded their money back. Legislation aimed at strengthening our financial institutions must not only focus on how banks are run, but also must incentivize customer confidence and trust.
Running on your bank is optimal according to game theory: A bank run is a variation of the most famous problem stated in game theory – the Prisoner’s Dilemma. Imagine a scenario where a bank has two depositors and if either depositor attempts to withdraw, the bank fails. If one depositor withdraws, the other ends up with nothing, and if both depositors withdraw, they only get a portion of their money. In this scenario, if the depositors have trust in the system, it is optimal for them to stay at the bank, but as soon as they think the other will leave, it becomes optimal to withdraw. This is how bank runs start, and this is why they are so hard to stop. It’s just game theory. Depositors are never to blame in a bank run, despite a few recent efforts to paint them so.
Bank runs are fast-moving events: The notion that bank runs occurred over prolonged periods prior to the internet is misleading. The closest parallel to SVB’s collapse, from a timing perspective, is the Bank of the United States, which failed almost a century ago. Money matters and people will rush to protect their cash when they lose confidence in financial institutions, regardless of whether they’re clicking a few buttons on a website or waiting for hours on the street to enter their local branch.
Figure I: Five of the most well-known bank failures in the last century experienced runs that lasted less than two weeks
SVB’s collapse was notable for its magnitude, not its speed: In a single day, $42bn (~24% of existing deposits) was withdrawn from SVB. According to regulators, if SVB had not been shuttered the following day, another $100bn of deposits (a staggering 81% of existing deposits) was scheduled to be withdrawn. This far exceeds deposit withdrawals in prior bank runs.
Figure II: While its length was not without historical precedent, SVB’s bank run witnessed an unparalleled magnitude of deposit withdrawals
It’s too early to assess the impact of digitization on bank run timing: We don’t have enough data points to make a clear-cut assessment of the impact of mobile and internet on bank runs. While 2008 was very different than 2023, it’s worth noting that the WaMu run primarily occurred over the internet during a macro environment experiencing extreme dislocation for over a year. Despite these dynamics, WaMu’s failure took time, and its bank run was noticeably slower than other historical examples. Similarly, we could argue the speed and magnitude of SVB’s failure was not only a function of digitization, but also a function of overconcentration and the size of customer accounts, which makes it hard to isolate variables.
Three common drivers of bank failures are overconcentration, rapid expansion, and bad investment decisions: It is remarkable how most U.S. failed banks demonstrate one (or more) of these characteristics. Of course, just because a bank demonstrates these characteristics, doesn’t mean it will fail. However, it certainly increases risk profile, which increases susceptibility to runs.
Figure III: The largest failed banks in the U.S. share a common set of characteristics
Post-facto government assistance can slow a collapse, but often fails to prevent it: In the case of Franklin International and Continental Illinois, federal assistance allowed the banks to continue operations for longer periods of time. Though this assistance may have enabled more customers to relocate deposits, it did little to prevent the eventual failure of these institutions. Instead, regulators should be focused on aiding banks before bank runs happen. Insured deposit protection is one example of an ex-ante regulation that can build customer confidence. Of course, if these regulations are not comprehensive enough (e.g., the majority of SVB deposits exceeded the insured deposit limits put in place by regulators), they can be rendered ineffective.
Regulators should be worried about the growing role of misinformation and its potential impact on our financial ecosystem: If a single Bronx businessman in 1930 can spread a local rumor and derail a bank, we should be highly concerned about misinformation in the modern era, which can be (and has been) amplified rapidly across social media and the internet. This issue strikes at the heart of so many unsolved, controversial topics today. Should social media platforms take content moderation stances on misinformation that could spark a financial crisis? Will adversarial governments begin to use bank runs as a novel attack vector and geopolitical bargaining chip? Will the next bank run be sparked by an erroneous hallucination from an AI model?
Despite these concerns, smaller-scale banks play an incredibly important role in our economy: Smaller-scale banks include regional banks, community banks, specialized banks, and minority-owned banks. If their risk profile is elevated, why should anyone deposit capital in them? Well, there are numerous benefits to smaller-scale banks, including more personalized customer attention, flexible lending options, community support and investment, and sometimes lower fees. Community banks represent over 95% of all bank charters, over 40% of branches, and roughly 15% of deposits, loans, and assets. For a significant portion of the population, the pros of a smaller-scale bank outweigh the cons. While I have focused this post on large bank failures, it is usually the smaller-scale banks that fail during a crisis, and that is precisely the time to remind ourselves about the importance of these institutions. Which brings us to a final, comforting point…
In major financial crises, the government has your back: To end on a positive note, it’s worth pointing out that the government has historically done a good job at protecting deposits when financial institutions fail. Whether it’s through direct bailouts or regulator-assisted asset sales, regulators have prioritized maintaining confidence in the financial system during times of distress. While this confidence boost has not stopped bank runs from happening in the first place, we can confidently say that U.S. banks have been, and will continue to be, trusted holders of capital for consumers and corporations alike.
Thanks for reading Posts in the Shell! Subscribe for free to receive new posts and support my work.