March 19, 2023
An item in Ezra Klein’s NYT column yesterday really grabbed by attention. Ezra cited a Wall Street Journal column that claimed that the Federal Reserve Board’s stress tests would not have detected Silicon Valley Bank’s (SVB) problems, because its stress tests did not consider interest rate risk.
This struck me as close to crazy. How could a stress test not consider interest rate risk? I recalled the stress tests that the Fed and Treasury performed very publicly in March of 2009, in the middle of the financial crisis. These tests did not consider interest rate risk for the simple reason that, at that point in time, soaring interest rates seemed about as likely as a Martian invasion.
I had not been following the Fed’s stress tests since that time, but I assumed that they did adjust them for circumstances. I recall back in 2002, when I first became concerned about the housing bubble, being on a radio show with the chief economist from Fannie Mae. He assured me that they could not have serious problems with a decline in housing prices, since they regularly stress test their assets. Their tests included a large rise in interest rates. (When the bubble finally burst, Fannie Mae, along with Freddie Mac, collapsed in the summer of 2008 and have been in conservatorship ever since.)
Anyhow, this exchange led me to believe that regulators applied some common sense to their stress test exercises and examined how bank assets would fare in all bad, but plausible, circumstances. In the years 2020-21, when 10-year Treasury rates were at times flirting with 1.0 percent, a sharp rise in interest rates had to be seen as a plausible, even if unlikely, possibility.
Incredibly, the Fed stress tests did not consider this scenario. This means that the Fed’s stress tests would not have detected the vulnerability of SVB to the sort of jump in interest rates that we have seen over the last year. That means that it is possible that, even if Dodd-Frank had not been weakened in 2018 to reduce the regulation to which SVB was subject, the Fed still would not have detected its problems.
I said “possible,” rather than asserting that the Fed would not have caught the bank’s vulnerabilities, because even without a stress test some items should have been apparent to anyone giving the bank careful scrutiny, as would have been required before the 2018 law weakening Dodd-Frank.
First and foremost, the bank had well over 90 percent of its liabilities in uninsured deposits. That has to be a red flag to any bank regulator. These are the deposits that are more likely to run in a crisis, since insured deposits have no reason to flee. Also, most banks have more of their liabilities in the form of bonds or other fixed term debt that cannot run.
The fact that the bank’s customers were highly concentrated in a single industry, the tech sector, also should have been a red flag. This is especially the case because tech has a long history as being a boom-bust industry.
Third, the bank’s assets had nearly tripled in size from the fourth quarter of 2019 to the fourth quarter of 2021. Again, any regulator with some clear eyes should have been asking if SVB was doing anything risky to bring about such extraordinary growth. As an old line goes, they should use their University of Chicago common sense: “If what we’re doing is not risky, why is the good lord being so nice to us?”
Anyhow, I mention these points since it still seems likely to me that if the Fed was applying the strict scrutiny to SVB, that had been required before the passage of the 2018 law weakening Dodd-Frank, it would have caught the bank’s vulnerabilities and required measures to shore up its capital and/or reduce its deposits. However, the stress tests the Fed was using would have been utterly worthless in detecting its problems.
This should be an important reminder that regulation does not necessarily solve market problems. Sometimes liberals seem to work from the assumption that if the market outcomes are getting things wrong, somehow bringing in the government will set things right.
This is often not the case. When I think back to my exchange with Fannie Mae’s chief economist, he insisted that a nationwide plunge in house prices was not even a possibility, since the country had never seen anything like that. While that was partly true (they did fall sharply in the Great Depression), we also had never seen the sort of nationwide run-up in house prices we were experiencing at the time.
But the key point was that he could not even consider the possibility that we were seeing a housing bubble, where house prices were being driven by irrational exuberance, rather than the fundamentals of the market. That was true of almost all the economists I encountered in those years. Even my friends largely did not buy the story, although they might politely nod when I made the case.
Anyhow, if we had more thoroughgoing regulation of the financial system in the years when the housing bubble was growing, there is little reason to think the regulators necessarily would have caught the financial system’s problems. After all, if the house price growth we saw in the bubble years made sense, then the banks’ behavior would not have been especially risky.
To my view, while we need government regulators in many circumstances, the most important part of the story is to structure the market to get the incentives right. That is why I have argued for a system where the Fed gives everyone an account which they can use for getting their paychecks, paying their bills, and other transactions.
This would be enormously more efficient than the current system, eliminating tens of billions in fees paid annually to the banks. The amount saved could be two or three times the price of Biden’s student loan debt forgiveness. It would also eliminate the problem of bankers sitting on huge pots of money where they can make great fortunes by taking big risks.
This is also the story with the pharmaceutical industry. If we paid for the development costs upfront, as we already do with more than $50 billion a year going to the National Institutes of Health, we would not only make drugs cheap (all drugs would be available as generics the day they are approved), we would also eliminate the incentive for drug companies to lie about their safety and effectiveness.
I have my longer tirade on this topic in chapter 5 of Rigged [it’s free], along with a discussion of other sectors. (See also here.) But the key point is that we can’t count on government regulation to right the wrongs of a badly structured market. Regulators can both make mistakes and also be corrupted by the industry they are regulating.
The most important reform is to structure the markets right in the first place, so that we can minimize the need for regulatory oversight. We need regulation in many circumstances, but even the best regulation will not correct the problems of a badly structured market.