On January 18, a prominent financial newsletter noted that if Silicon Valley Bank were liquidated that day, “it would be functionally underwater.” Months before the nation’s 16th-largest bank collapsed, incomplete information provided to regulators indicated that the bank was stable, whereas public signals—such as SVB’s overreliance on longer-term securities hammered by rising interest rates—told a very different story. So why didn’t anyone do something?

To help answer this question, I turned to Natasha Sarin, a lawyer and an economist teaching at Yale Law School, who served in senior roles at the Treasury Department under Secretary Janet Yellen.

[Derek Thompson: The end of Silicon Valley Bank—and a Silicon Valley myth]

Sarin thinks that many of us are asking the wrong questions. Instead of focusing mostly on what to do after banks suffer this type of financial distress, federal regulators need to get better at forecasting errors before they become crises. And to do so, they’re going to have to update how they determine whether banks are in good standing.

In our conversation, Sarin described a regulatory system that failed to detect the market’s growing trepidation with SVB and similar banks. In part, regulators were hobbled by 2018 changes to financial regulations that exempted banks with assets below $250 billion from some oversight measures, including the yearly stress testing that larger banks undergo.

In lobbying for those changes, SVB and other regional banks argued that they weren’t systemically important. But clearly, the federal government now disagrees, having guaranteed deposits above the official $250,000 Federal Deposit Insurance Corporation threshold out of concern that failures at SVB and Signature Bank could spiral across the system. In fact, despite federal regulators’ steps to restore confidence, on Monday, the stocks of several regional banks plummeted, reflecting ongoing fear and uncertainty working its way through the market.

If regional banks are not systemically important, the level of public intervention in the SVB crisis is hard to justify. The other possibility is that our laws don’t match reality—making the current regulatory regime untenable.

The following conversation has been edited for length and clarity.

Jerusalem Demsas: The Independent Community Bankers of America issued a statement touting its “local” and “relationship focused” business model and arguing that SVB is not a community bank, given its size. What is the value of regional banks in this intermediate range—too big to be a community bank but smaller than the JPMorgan Chases of the world?

Natasha Sarin: Why is it the case that you see basically all venture-capital firms and all of their portfolio companies banking at SVB? You said it’s not relationship banking. I speculate that a little bit of what you saw with respect to SVB is relationship-based. If you’re a founder of a portfolio company, and your VC suggests this bank, there is a network effect.

Even in its name, Silicon Valley Bank, there is a cultural-institutional thing that I speculate is relationship banking in a way that may be different from how we traditionally define it. It is community, in fact, a fabric. But what you saw is how quickly that can dismantle, and you saw all these founders trying to open a bank account at JPMorgan Chase.

I do think this is going to shift the nature of the industry in ways that we don’t understand yet. It should shift the regulatory framework that we operate under, because clearly these are systemically important financial institutions.

Demsas: The insularity within Silicon Valley is interesting. Although it’s not unique to this community, watching in recent months the layoffs that have spread through the tech world, and now this bank run, do you think something about Silicon Valley’s culture made it ripe for social contagion?

Sarin: When I teach about bank runs, I teach this foundational scene from It’s A Wonderful Life, where you see a bunch of people lined up in front of a bank. They’re trying to get their money out, because the rumor’s been spread that the bank might not be able to pay all the depositors.

[Read: Silicon Valley was unstoppable. Now it’s just a house of cards.]

What I was struck by over the past few days was: This is just a fundamentally different type of bank run. There are no pictures of people lining up out of the branch trying to get their deposits out. This was an electronic bank run, where people instantaneously, with a click of a computer key, were moving out billions of dollars of deposits over the course of a few hours from a financial institution that had been a pillar of Silicon Valley and an incredibly important financial intermediary in this pretty unique ecosystem.

A bunch of it was happening on Slack and through tweets from particular VCs. It’s both totally different and totally the same. Once someone’s nervous, the rest of the community is nervous. The speed at which this all happened is something that regulators were fundamentally not really attuned to.

Financial regulation has a really deep problem in that it relies on a bunch of regulatory information that banks provide the Federal Reserve and FDIC. That information is really useful and valuable, but it’s a pretty static snapshot of financial institutions. It says, “Here is my position a few months ago, based on these regulatory measures of my health, which are pretty easily gamed.” Measures like how much capital do I have based on regulatory risk-weights—what is the value of my securities portfolio based on book measures that don’t adjust for how the market has changed? That was the case with Silicon Valley Bank. [The information available to regulators is] sometimes not reflective of how the market views the value of those institutions. From the perspective of depositors thinking about running, those market values are really, really important.

Demsas: Is there anything to be done about that initial social panic? That’s the point of insuring the initial $250,000 of deposits, right—to prevent regular folks and smaller deposit-holders from freaking out when there’s potential instability? Are there other ways to intervene when this panic begins?

Sarin: A famous Rahm Emanuel saying is that no one should let a good crisis go to waste. And that is the moment we are in with respect to financial regulation.

The reason deposit insurance exists is to stem the incentives for people to run: If I’m under the deposit-insurance threshold, then there’s no reason for me to be concerned.

Do you increase the deposit-insurance threshold, particularly for the types of depository bases that SVB had, where 90 percent of its deposits or some such number were uninsured? If 100 percent of their deposits had been insured, there wouldn’t have been the same incentives to try to withdraw. I’m not sure what the right policy is with respect to the deposit insurance threshold going forward, but we need to think about it and the implicit guarantee we’ve provided to bank deposits writ large.

But I almost think that’s the wrong question, because it’s still about: What do you do ex post facto when there is a crisis? How do you deal with it in the most efficient way? The question that regulators have to ask themselves is ex ante. What can we do better to identify these instances and encourage banks to shore themselves up if they’re about to hit a moment of crisis?

And it’s not like it wasn’t clear that we were on the brink of a potential crisis, because you’ve had people for months saying, “Listen, if [Silicon Valley Bank] had to liquidate today, they wouldn’t have been solvent.”

There were people sounding the alarm. It’s just that regulators don’t have to respond, as part of their structure, to those types of market changes.

Demsas: So once the crisis hit, what options did the Feds have?

Sarin: Ultimately, I don’t actually think there was much of a choice in front of the regulatory community. It’s an irregular situation. The FDIC is incredibly efficient: We come in on a Friday afternoon and shutter the bank, and by Monday morning, the bank is sold to someone else, and the liabilities of the bank are transferred to the liabilities of the new institution. So the depositors still have whatever was in their bank account going into the weekend, and the new financial institution is absorbed.

The word bailout typically refers to “you bailed out with taxpayer money,” “you bailed out the equity holders of a financial institution and people who have shares of Silicon Valley Bank.” That didn’t happen in this case. There is no bailout. Those equity holders were fully wiped out. Silicon Valley Bank doesn’t exist anymore, so it’s not a bailout of anyone. The funds that are being used in this case to protect depositors are funds that banks pay into the FDIC to provide insurance in cases exactly like this one.

I don’t want to understate the severity of the moment. What does this mean about uninsured depositors? Because we have in our mind the concept that after a certain threshold, deposits aren’t actually insured, and I think what [this crisis] means is that we need to think about those deposit-insurance thresholds. We need to think about whether you need to pay in more ex ante to protect against exactly a moment like this one, so we know that we have enough funding to try to support cases where systemically important financial institutions ultimately fail.

[James Surowiecki: What social media is doing to finance]

Another question that regulators have to grapple with is: How do they get better at identifying these types of moments?

Demsas: What sorts of regulations do you think we should be considering, going forward?

Sarin: There’s a ton of really good information in bank regulators’ assessments of financial institutions’ safety and soundness. And there are regularized interactions between institutions and regulators for large financial institutions; there’s a regularized system of stress testing against potential risks. That reveals information about which banks are safe and which banks are unsafe.

You had regional banks such as SVP make the argument that they weren’t actually systemically important financial institutions in a way that necessitated the type of greater scrutiny [faced by] the JPMorgans and the Bank of Americas of the world. Ultimately successfully, regulations were loosened with respect to annualized stress testing, for example, for these [regional] institutions.

The issue with [regulators’] approaches is that they’re incomplete, even for large systemically important banks that are stress-tested annually. They’re incomplete because they’re missing a whole host of inputs that exist in markets but not in regulators’ calculations about banks’ stability and soundness—things like market-based value of a bank’s capital position or market-based measures of the volatility of different institutions, how exposed they are to big fluctuations up or down in the stock market. Those sorts of measures are super easily accessible to regulators and to market participants, but they’re just not incorporated in the picture that regulators paint of financial stability.

Demsas: If they had had that information, what could they have done in advance?

Sarin: Ultimately, what SVB tried to do last week was raise new equity capital. They were like, “We have to sell some of our securities at a loss, and so we’re going to bolster ourselves by infusing into the institution this buffer of stability.” That was just too little, too late.

If they had been encouraged or pushed by regulators to undertake that type of activity sooner, to bring in new capital to the institution in order to buffer it against the losses that it had experienced, that would have been a way to avert, potentially, the crisis that we saw. Or they could have been encouraged to restructure their assets in ways that decreased their exposure to interest-rate changes. But there was no push to do any of that, because even though market measures were showing cause for concern, the regulatory measures were really stable with respect to bank health.

We need to look at more real-time information and at least incorporate it. Sometimes market information is noisy; sometimes it’s incomplete. But the issue is just ignoring it altogether, which is the implicit assumption of our regulatory regime today.