Stress Testing Wouldn’t Have Saved Silicon Valley Bank

After President Trump signed the Economic Growth, Regulatory Relief and Consumer Protection Act in May 2018, midsize banks with assets between $50 billion and $250 billion, such as SVB and Signature Bank, effectively were reclassified as systemically unimportant. Only banks with more than $250 billion in assets are now automatically subject to Dodd-Frank rules for systemically important financial institutions. Yet recent actions by federal regulators certainly seem to be treating midsize banks as systemically important—lavishing deposit guarantees for uninsured depositors at SVB and setting up a new emergency Federal Reserve Bank Term Funding Program to extend liquidity to troubled banks and insulate the financial system from contagion.

But even if midsize banks had been subjected to the same scrutiny as large banks, it isn’t clear that stress testing them would have led to changes that would have prevented failure. Why? Because the tests asked the wrong questions. They failed to encompass the scenarios that ultimately led to SVB’s demise—large and rapid increases in interest rates.

In its February 2022 Stress Test Scenarios, the Fed’s “severely adverse scenario” asked banks to assess their riskiness over a three-year horizon in a hypothetical world in which the three-month Treasury rate stays near zero while the 10-year Treasury yield declines to 0.75% during the first quarter of 2022 and doesn’t change in the subsequent two quarters. Even in December 2021, however, the Federal Open Market Committee’s Summary of Economic Projections was showing the Fed likely targeting interest rates double those of 2022 in 2023, far higher than what it used for bank stress tests.

A reasonable observer would expect FOMC’s policy objectives to have been embedded in the 2023 Stress Test Scenarios. But by February 2023, the Fed still hadn’t changed its regulations to match its monetary policy. While FOMC’s December 2022 projections show its policy rate reaching 5.1% by the end of 2023, the February 2023 severely adverse scenario was almost identical to that used in February 2022: The three-month Treasury rate falls to near zero by the third quarter of 2023, while the 10-year Treasury yield falls to around 0.5% by the second quarter, then gradually rises to 1.5% later in the scenario.

The 2023 severely adverse scenario’s assumptions bore no relationship to reality. In February 2023, the three-month Treasury rate had already risen above 4.5%. Since Feb. 10, 2022, the 10-year has nearly doubled, from about 2% to almost 4%. Even this year, from Feb. 9 to March 10, the 10-year yield has risen by about a quarter of a percentage point.

As has now been well documented, SVB’s deposits were heavily reliant on funds from tech startups and venture-capital firms and most were too big to be covered by deposit insurance. When rates rose, VCs pulled back on funding new innovation, and startups began drawing down their deposits. To meet withdrawals, SVB sold interest-bearing assets into a market where rates were considerably higher than when it purchased them. When SVB reported these losses, markets and depositors got nervous, and a bank run began.

If stress testing is meant to warn regulators about banks in advance of runs, the Fed needs to include scenarios that reflect its own policies. Applying appropriate stress testing to large as well as midsize banks would help ensure that the next SVB isn’t a sitting duck waiting for a run.

Mr. Mason is a professor of finance at Louisiana State University, a fellow at the University of Pennsylvania’s Wharton School and a former senior financial economist at the Office of the Comptroller of the Currency. Mr. Mitchener is a professor of economics at Santa Clara University.