The Surprising Risk That Turbocharged a $142 Billion Bank Run

So he set out to answer what might be the most intriguing question about the bank run: Was it fueled by social media?

Dr. Cookson and a team of collaborators from around the world collected millions of tweets to understand the swift movement of billions of dollars. Their buzzy new working paper presents compelling evidence that social media like Twitter didn’t simply expose the risks of a bank run. It exacerbated them. The technology of Silicon Valley really did contribute to the demise of a bank in Silicon Valley.

The forces of social media have spilled into every part of society, but it seemed improbable that a prominent institution could be toppled in two days by a barrage of tweets, if only because the human mind struggles to picture anything that has never happened before. And never before had a bank with so much money gone bust in such a bizarre way. Now it has.

“SVB faced a novel channel of bank-run risk that is unique to the social-media era," wrote Dr. Cookson, an associate professor of finance at the University of Colorado’s business school, and his colleagues. “We do not expect this risk to go away."

Of course, not every bank’s depositors include startup founders and venture capitalists with sizable online followings, and social media wasn’t the only risk that Silicon Valley Bank mismanaged before its downfall. There was the risk of rising interest rates. There was the risk of its unusually high percentage of uninsured deposits. There was also the risk of not having a chief risk officer.

But the banking panic illustrated another, more prevalent risk that won’t be going away.

It’s the risk of the unknown.

The hardest risks for organizations to hedge are the ones they can’t envision. Even financial institutions in the very business of pricing risk are vulnerable to what they don’t see coming.

There’s a difference between ignoring risks and getting blindsided by them, but SVB fell victim to both. The second-largest bank failure in U.S. history was as much a failure of imagination as it was a failure of executives, supervisors and regulators.

The basic facts of the SVB meltdown remain shocking even if they’re no longer surprising. There wasn’t much reason to suspect that a bank with $209 billion in assets and $175 billion in deposits at the start of the year was in mortal peril as markets closed on March 8. But after SVB said it needed to plug a massive hole in its balance sheet, its stock crashed on March 9 and kept falling until the bank was seized by regulators on March 10. Federal Reserve officials have said that customers yanked $42 billion on the first day of the run and were on track to withdraw another $100 billion the next day.

This was not an obvious risk. Scholars have known for decades that coordination is the key element of bank runs, which require depositors to communicate in real time about whether they intend to withdraw their money. Not long ago, that was difficult. At the very least, it took a while. “Someone would hear something on the radio," said Dr. Cookson, “or they would see people lined up outside the bank." The result was that bank runs couldn’t really be sprints.

But they have been accelerated by the way we communicate now—rapidly, efficiently, often publicly—and the speed of this one made Usain Bolt look slow.

To understand Silicon Valley Bank and this $142 billion game of telephone, Dr. Cookson and his colleagues Christoph Schiller, Corbin Fox, Javier Gil-Bazo and Juan Felipe Imbet took data from a company familiar with the evaporation of money: Twitter.

The whispers about SVB concerns had spread through private WhatsApp chats and Slack channels, but Twitter was the megaphone that amplified them. That was useful for these researchers. It meant they could eavesdrop on the conversations that depositors were having in public.

They pulled more than 5 million tweets about publicly traded banks and focused on the two crucial weeks in March when markets tanked and SVB and Signature Bank failed. The scholars then used a list of keywords to classify tweets by sentiment and identify the posts that came from influential voices in the startup community. (They don’t cite the specific tweets that reached the widest audience, like one that began: “YOU SHOULD BE ABSOLUTELY TERRIFIED RIGHT NOW.")

The next step was to compare the time and tenor of the tweets with the trading activity of hundreds of banks. They didn’t have access to customer withdrawals—researchers are not regulators—so they used every bank’s stock price as a proxy measure of its deposit outflows.

That elegant solution is how they calculated that negative tweets translated to negative returns. The effects were stronger and the stock-market losses greater for banks with higher profiles on social media, like SVB, which catered to the tech industry and called itself “the financial partner of the innovation economy." If there were any bank susceptible to these risks, it was one named Silicon Valley Bank.

The classic portrayal of a bank run appears in “It’s a Wonderful Life," but a bank run today looks nothing like a black-and-white movie from the 1940s. Instead of crowding around Bailey Bros. Building & Loan, you can withdraw your life savings from your phone. You don’t even have to roll out of bed to participate in a run on the bank these days.

If the story of GameStop demonstrated how investors on Reddit could drive a meme stock’s price to the moon, Twitter’s role in the saga of Silicon Valley Bank reiterated that the dynamics of financial contagion have been forever changed by social media.

It won’t be the last time something like this happens. But next time will be different. We just don’t know how.

And the most dangerous risks in any business are the ones we don’t know to fear.