The Federal Reserve just doesn’t make up recessions like it used to.
Four years ago in its annual bank stress test, the Fed predicted that in a severe slowdown Citigroup Inc. would lose nearly $55 billion from bad loans. This year, however, Citi’s estimated lending losses in a recession dropped to just less than $45 billion. That’s not because Citi is lending any less. Fed examiners estimated that a quarter fewer of Citi’s loans would go bad in a recession than they did four years ago. Under their old assumptions, Citi’s lending losses actually balloon to $60 billion. Its trading losses, too, would more than double. Citi passed this year’s Fed stress test with a minimum score on its most important capital ratio of 9.7 percent, the highest grade it has received in years. But under the assumptions the Fed was using four years ago, Citi’s ratio would have been 7.8 percent. That’s still a passing grade, but not an impressive one.
Every bank passed when the Fed reported the results of its annual stress test on Thursday, which is not much of surprise. It’s been a few years since a big bank failed. Nonetheless, the Fed once again said the banking industry’s clean report card showed that the banking system, unlike nine years ago, is prepared for the worst. And while it’s true that the banks have more capital on hand to cover loan and trading losses than they did a few years ago, and are engaged, because of regulation, in less risky lending and trading than a few years ago, here’s something that is also true: The test to determine the survivability of another financial crisis is a lot easier than it used to be. On pretty much every key metric the Fed’s bank stress test is a lot less strenuous than it used to be.
Consider capital ratios. This year the Fed predicted that the average capital ratio of the big banks taking the test would only dip at most to 9.2 percent during a recession. That was the highest that ratio has been since the Fed started administering the test. Also true, the Fed has been applying less pressure. The average bank capital ratio in the Fed’s stress test this year dropped by just 3.3 percentage points. Last year, the Fed stress test model of what a severe recession would look like cut bank capital ratios by 3.9 percentage points. And that was down from a 4.3 percentage point drop the year before.
It’s not just capital ratios. The Fed also assumed lower default rates, 5.8 percent compared with 6.1 percent the year before, and lower trading losses. This year the banks’ traders were predicted to lose $86 billion, about a quarter less than what was estimated in last year’s test. It must be the computers. The banks also appear to deal with adversity better than they used to. The Fed estimated they would bring in $30 billion more in net revenue in a recession than they did in the test a year ago.
It’s not clear why the banks are doing better in the Fed’s stress test. The stated overall economic conditions — 50 percent drop in the stock market, 10 percent unemployment, and a 35 percent drop in real estate prices — are still pretty harsh. Some of the better results may make sense. Perhaps after seven years of taking the Fed’s test, the banks know how to game it better. There are fewer subprime mortgage loans, and many of the bad loans from a few years ago have rolled off the books. So you would expect lower lending losses there. Still, the default rate the Fed used of 2.2 percent for all mortgage loans looks a little generous. That’s down from 9.4 percent four years ago. And not every part of the test was easier. Goldman Sachs Group Inc. and Morgan Stanley, for example, showed higher overall losses from lending than they did a few years ago, but both banks are lending more.
The stress tests are set to get even less stressful under the Trump administration. Earlier this month, the Treasury Department, as part of a plan to loosen regulation and boost lending, proposed switching the stress test to every other year. That might be a good idea, but not because the test has become too much of a burden. Perhaps on that schedule the Fed could cook up a little tougher exam.
This column does not necessarily reflect the opinion of Bloomberg LP and its owners.
To contact the author of this story:
Stephen Gandel in New York at firstname.lastname@example.org
To contact the editor responsible for this story:
Daniel Niemi at email@example.com