In just a few years, news specials and academic papers will mark 100 years since the start of the Great Depression. Archival photographs will be dusted off to display the restive crowds gathering outside bank doors in desperate attempts to collect their life savings.
A kind of dress rehearsal for the coming commemorations took place in recent days as the financially besieged Silicon Valley Bank (SVB) collapsed. Following tradition, people did appear at the bank’s doors. But the inventiveness of Silicon Valley in the decades since that earlier economic shock have offered alternate pathways for depositors to show up en masse for a bank run. Every available electronic channel—Slack, Twitter, online banking—was brought to bear, all without the necessity of showing up at 3003 Tasman Drive in Santa Clara, Calif., the bank’s main office. Private Slack channels filled with frenzied messages from those pulling out their money. Depositors could sit at home and obsessively refresh their browsers for hours on end to try to complete an online transfer of their cash to another institution. In a recent statement, House Financial Services Committee chairman Patrick McHenry characterized what was happening as “the first Twitter-fueled bank run.”
One thing that hasn’t changed during the past 100 years is the abject terror that can be induced at the prospect of losing all of one’s savings or the cash to keep a business afloat. The discipline of behavioral economics and the related fields of behavioral finance and neuroeconomics specialize in exploring the biases and irrationality that can lead to the “madness of crowds” in financial markets. To better understand the psychology of tech start-up heads and venture capitalists rushing for the proverbial exits, Scientific American turned to Colin Camerer, a professor of behavioral economics at the California Institute of Technology and a MacArthur Fellowship winner.
[An edited transcript of the interview follows.]
We wanted to ask a behavioral economist about the types of flawed thinking that could lead to a bank failure in Silicon Valley.
I do have a theory of what's going on. It relates to something called “skewness.” Positive skewness is upside potential. That means there's a small chance of something really great happening such as buying a winning lottery ticket or a company becoming a billion-dollar “tech unicorn” in the start-up world. And negative skewness is the opposite: it’s a small chance of something terrible happening such as a bank run.
So venture capitalists [VCs] and the start-up world are very good at two very interesting things. One is that VCs don’t mind losing all their money. They don’t want to, but they understand that if you’re investing in a portfolio of these potential unicorns, positive skewness, or high upside, also means that nine bets out of 10, or some similar large percentage, are going to end up at zero. And there’s a 10 percent chance of a huge payoff.
The idea is: you’re trying to manage the portfolio where you have as many unicorn winners as you can. And so a lot of the venture capital analysis and pricing is: How do you tell the good long shots from the bad long shots? And as I mentioned, [VCs are] quite used to the idea of losing money. So they have almost an immunity. You can invest $20 million in a company, and three years later it’s worth zero. That just doesn’t faze them because they understand that’s the price you pay. You’re just like somebody buying lottery tickets.
And what about you second point, the negative skew?
What I think VCs are not very good at is worrying about the downside potential. What I mean by that is: If you read interviews with start-up people, there’s a tremendous amount of puffery. They will say, “This is the greatest product, and we’re going to be the next Facebook” or “the next Google” or the next whatever. “We’ll be the Uber for schoolchildren” or something like that. And there’s usually a sense of unbridled optimism among this crowd.
Banking and finance, however, are the exact opposite. It’s related to something called risk management. So companies that deal with large amounts of money, particularly financial companies, have risk managers. They usually report to a C-suite executive, and they’re very important. Their job is to worry. Their job is to ask, “What are some scenarios in which we could lose a lot of money?” And they try to protect against downside risk. And I think most of the tech companies that have their money in places like SVB are just not thinking about risk management. [Editor’s Note: SVB itself reportedly had no risk officer for most of last year.]
A lot of the SVB customers who were founders of companies had personal money, mortgages, and so on in the bank. SVB was known to be supportive of founders. Putting all your assets and company money into one bank is not usually the smartest thing to do. The number one rule of household finance is diversification: don’t put all your eggs in one basket.
From a risk management point of view, the bank customers were just not thinking about the strength of their bank, SVB. It’s not really their job to do so—they rely on regulators and SVB’s managers to worry about the bank’s financial health.
So I think risk management culture, as it’s usually practiced, is antithetical to the Silicon Valley culture. Silicon Valley has the least risk-averse people in the world, but they take risk because they like positive skewness. It is not in the nature of the bank’s start-up customers to be thinking about bank run risk.
I am not sure how much this blindness to negative skewness extended to SVB executives, who mostly had regular banking backgrounds. They did a lot of unusual lending such as accepting collateral in shares of start-ups, which were probably not always very liquid. If the loans were not paid back, they had to sell these shares somehow. They were known for “understanding” VC and start-up lending in a way that big banks did not. A major mistake seemed to be putting a lot of SVB assets in “safe” long-term mortgage-backed securities, which can drop a lot in value when interest rates rise.
From the standpoint of group psychology, does this culture create a certain group mindset that breeds a false sense of trust? Paul Krugman, the economist and columnist wrote in a tweet, “In a deep sense—though not a legal sense—what SVB actually did was a kind of affinity fraud a la [Bernard] Madoff. It managed to convince the VC/startup/crypto etc world that it was one of them, part of their community, and hence trustworthy.”
Affinity fraud refers to a preexisting group connection between people that can allow scams to happen. Usually, the affinity is within a religious group or based on some shared background. In the tech industry, there is also a kind of affinity: who’s done deals with who, whether you went to Stanford, and so on. There is a sort of attitude in the tech industry that you’re part of a tribe, a group of people that you can naturally trust. And often what that means in practice is: “I met this guy; he went to the same college as I did” or “I met this guy, and he put some money into our start-up.” And so “anybody who went to the same college as me could not possibly run a bank into the ground.”
But things don’t always work that way. Human beings are complicated. SVB seems to have been seen in tech as “our bank,” charismatic in the way Bernie Madoff or Elizabeth Holmes from Theranos were, although not as badly intentioned.
Do you have any other thoughts on the psychology that actually spurs bank runs?
There’s a phenomenon that banking experts talk about called contagion. If I, for example, think other people are going to pull all their money out, I’m going to try to pull mine out first because there’s not enough money in the bank. Douglas Diamond and Philip Dybvig won the Nobal Prize in economics last year for a mathematical model of how this can happen and how to prevent it.
The recipe for what causes contagion and how to prevent it is not at all well defined, however. We are nowhere near to having a formula that predicts when a contagion is likely.
There are various fields that have studied behavioral contagion. If you model a bank run, for example, and three large customers in the network being modeled take their money out, does that cause the others to continue a run on the bank or not? Or is there some tipping point, such as “Three customers are fine, but five is too many”? The answer for these kinds of problems is usually “It depends.” And there is almost surely some indeterminacy—even with similar economic conditions and the same degree of publicity in the press, you might get a run on one bank but not on another.
I think future studies to find out more about contagion should be a mixture of group psychology and ideas from other fields such as the study of collective behavior—flocking stampedes in animals is one example. The group and tribalism and information flow are also part of it. Apparently, a lot of tech start-ups and VC firms were communicating with each other and becoming worried about the bank, which triggered the first couple of large customers withdrawing funds.
Has brain research explored any ideas about these behaviors?
There’s quite a few studies of the neural signature of conformity. In a typical study, people hear a few seconds of a song. And they’re told, say, three other people loved it but one did not. When people agree with the majority and say, “I loved it, too,” there’s activity in the brain associated with reward. Conforming to what other people think seems to be a general reward just like money or food. I thought there might be reward for nonconformity instead, but the general finding is a neural reward for conformity.
As far as SVB, brain reward generated by conforming to what others are doing (“They’re taking their money out, and so should I”) is probably only a piece of the story that explains SVB. But it could still provide a small amount of the fuel that accelerates a bank run.
Some of the solutions to problems like contagion seem like they inevitably require political and regulatory measures.
In  the Dodd-Frank [Wall Street Reform and Consumer Protection Act] did institute a lot more control and regulation, including larger bank reserves and “stress testing,” in which you look at all the numbers and try to guess what the bank balance sheet would look like after a change in interest rates or economic conditions. You want to financially earthquake-proof the bank. But as you may know, a 2018 supplement to Dodd-Frank was passed that said, We’re going to raise the size of the biggest banks that need this type of scrutiny most from $50 billion to $250 billion in assets. SVB was $209 billion, so without the 2018 law, it might have picked up possible distress from better stress testing. Then the bank regulators and SVB itself might have had an early warning signal and could take action by requiring more reserves or raising capital, which protect depositors better.
What steps can be taken to remind people to keep their guard up?
I think, ironically, one of the things that actually works best is to have something bad happen very publicly. And that raises everyone’s consciousness. It’s hard to change the culture to one of prevention. And the political economy doesn’t help, as the 2018 law shows. These are people who are proud of being risk takers. Putting all your money in a bank that may have a bank run, that’s a kind of risk, too. But the tech industry, in my opinion, is blind to this kind of rare risk and is not used to worrying about it. Now banks will get a talking-to from their risk managers and from investors, maybe even at shareholder meetings. So I think there’ll be this raised consciousness, but there will still be this view that risk managers are cautious worrywarts. They’re like forest rangers who are going to a campsite and saying, “Your campfire can only be two feet wide. This campfire is two and a half feet wide. You have to put it out.” No camper wants to hear that.
I think you’ll get a little bit of a corrective, but it’ll mostly be in the form that says, “Let’s not be the next SVB or the next Signature” [another bank that failed in recent days]. For the moment there probably won’t be another big bank run because this one happened. And that could be because banks voluntarily do more stress testing, even if they’re not required to by regulation, or because venture capitalists have a portfolio of 20 companies but do not encourage those companies to all use the same bank.
Editor’s Note: The Department of Justice and the Securities and Exchange Commission have launched investigations into the SVB collapse, according to the Associated Press, and they are in their early stages of examining the actions of the bank’s senior executives. Silicon Valley Bank had not replied to a request for comment by the time of this article’s publication.