Banking goes back to the 1920s


The Federal Reserve has created a lot of new tools in the past 15 years. But to respond to the retrograde challenge of a depositor panic, its member banks have gone back to the basics.

While a new Fed facility got some use in its first days of operation after SVB’s failure, the Fed’s discount window — its oldest and normally its least popular option for emergency bank borrowing — saw far more activity.

Banks have tapped $12bn of liquidity from the Fed’s brand-new Bank Term Funding Program, or BTFP, according to the latest H.4.1. release. They had drawn nearly $153bn from the discount window by Wednesday. That’s a new record for discount-window volume, reports mainFTer and former Alphavillain Colby Smith.

In the early days of the Fed, the discount window was an actual window, with a teller and everything (see the picture above). And if a bank found itself in a liquidity bind, it would send someone over to their local Reserve Bank’s window to take out a short-term collateralised loan.

There is some nice historical symmetry here. As some readers will remember from the classic paper by Yale’s Gary Gorton and Andrew Metrick on the Fed’s first 100 years, the whole point of the discount window was to prevent bank-run panics. They quote Federal Reserve Act sponsor Senator Robert Owen, who said the law would give “assurance to the business men of the country that they never need fear a currency famine. It assures them absolutely against the danger of financial panic . . .”

That assurance was ultimately tough to provide. Because the discount window’s loans were made at below-market rates, the Fed spent much of the following decades trying everything (other than raising costs) to discourage bankers from regularly using it. By 2003, when the Fed raised the cost of accessing the discount window, there was a well recognised stigma associated with its use.

That’s one reason it’s notable that the newer (and presumably less stigmatised) BTFP did much lower volumes than the discount window:

All vibes aside, the BTFP offers loans for up to a year, compared to the 90-day maximum at the discount window.

More importantly, banks can take out loans against the par value of certain types of bonds pledged as collateral. That means Fed will lend against safe bonds as if they hadn’t experienced significant losses from rising interest rates. In other words, banks can avoid the unrealised losses that helped kick off the panic over Silicon Valley Bank.

Banks can only get that deal on a few types of bonds, however. The term sheet for the BTFP shows that it only accepts securities used in the Fed’s open market operations, meaning mostly Treasuries and agency securities, along with a few others you can find here.

This may have posed a challenge for some stressed banks, including possibly First Republic.

At the end of 2022, First Republic had a significant amount of its portfolio of securities invested in long-dated municipal bonds, which are not listed as eligible collateral for the BTFP program. The discount window does accept munis, along with the mortgage loans to wealthy clients that make up an even larger part of First Republic’s balance sheet.

Unfortunately for us, the Fed has also made it impossible to tell whether high discount-window volumes are a sign of trouble, even if we compare them to BTFP volumes.

Here’s why: the Fed said Sunday that banks can borrow par value of some bonds at the discount window as well. But only the ultra-safe bonds that are eligible for the BTFP.

So let’s say you’re a bank. The two programmes are basically identical if you have good collateral (and if you don’t need to borrow for more than three months). And if you don’t have good collateral, your borrowing will be grouped with whichever banks decided they wanted to take a retro approach to borrowing against their bond holdings at par.

It’s tough to beat a classic.