Why the Fed thinks Goldman is America’s riskiest bank

Goldman Sachs has a new title it may have to live with: America’s riskiest big bank, according to the Federal Reserve.

This week the Fed slapped Goldman with a total capital requirement higher than any other US bank, thanks to a large “stress capital buffer” the regulator wants all banks to maintain to see them through the severest of shocks. The resulting requirement for tier 1 common equity — largely generated from retained profits and share sales — is at least 13.7 per cent of Goldman’s risk-weighted assets. 

Reacting to the verdict, the Wall Street bank said it could continue with plans to invest in new business and would maintain current dividend payments, even though the tier 1 target it must hit by October is a touch above its current position. 

The Fed arrived at that number based on two assessments. 

The first is the regular annual stress test, which models how Goldman and 32 other banks would perform in a garden-variety recession and a severe one. The second is a new “sensitivity analysis,” a rougher exercise that looked at how the banks could fare under the additional strain of the pandemic. Results of that exercise were released only on an aggregate basis.

But for Goldman, the numbers that really leap off the page are the Fed’s assessment of its worst-case loan losses over the nine quarters from the start of this year to March 2022. 

Goldman’s numbers include some stunning outliers, like a loss rate of 25.9 per cent on its mortgage book: almost 10 times higher than the next worst result.

Meanwhile, the 14.9 per cent assumed losses on Goldman’s commercial and industrial lending portfolio are more than twice the 7.2 per cent average of the rest of the participating banks.

Faced with a less jaw-dropping set of numbers and a capital demand it is already meeting, US retail bank Citizens publicly took aim at “inaccuracies” in the Fed’s calculations.

Goldman is so far holding its tongue. Marty Mosby, analyst at Vining Sparks, said the bank faced higher losses for consumer loans largely because its business — which began offering online loans less than four years ago and credit cards in 2019 — had not been tested by a recession, so regulators might be inclined to be conservative. 

Mr Mosby also said the higher mortgage losses reflected the fact the loans “don’t fit” the Fed’s models since they were for wealthy individuals in Goldman’s private bank, rather than the “bread-and-butter lending” the central bank typically reviewed.

Some analysts are less circumspect. “It’s bananas,” says Chris Kotowski, banks analyst at Oppenheimer.

Bananas or not, the damage to Goldman from those bruising assumptions is manageable because the bank’s loan book is relatively small, at $128bn, about one-eighth the size of JPMorgan Chase’s.

That means the Fed’s loan-loss predictions took just $9.8bn from Goldman’s bottom line over the period, far less than the $47bn of losses pencilled in for Citi, BofA or Wells Fargo.

The real damage is the $18.4bn of trading and counterparty losses to Goldman’s trading assets that the Fed envisaged, which pushed Goldman to a net $27.5bn loss over the test period.

Mr Kotowski argues that these assumptions have already proven wrong for Goldman, since the swings of the Fed’s worst-case scenario were similar to what banks actually endured in March. Rather than losing billions, Goldman’s trading operations made money: $2.25bn in first-quarter pre-tax profits, up 75 per cent from a year earlier.

There are legitimate reasons for discrepancies between the Fed’s projections and Goldman’s actual performance. The Fed assumed no policy support — least of all the dramatic interventions from the central bank itself, which steadied the market. The Fed’s assessment was also based on historic snapshots of end-2019 balance sheets, which shifted rapidly during March and April.

Still, Mike Mayo, analyst at Wells Fargo, said the Fed’s models were so at odds with reality that Goldman had a duty to shareholders to contest them.

“You have a fantastic record [on managing risk] and now you’re going to let the Fed . . . make you have the highest capital requirements because of some assumptions that aren’t even clear to them or to us?” Mr Mayo said. “Why are you lying down and taking this?”

Citizens will ask the Fed to “reconsider” its stress capital buffer as part of a consultation process that runs until August, a person familiar with that bank’s situation told the FT. 

So far, Goldman has shown no sign of following suit.

A person at another large bank said it was difficult to confidently argue against an exercise that was a “black box”.

Even Mr Mayo admits that challenging the Fed is risky. After all, regulators oversee a bank’s day-to-day activities and can veto strategic plans such as mergers. 

“You can fight this and make some progress, but then you’re screwed for the next century,” he said.

laura.noonan@ft.com