Link to the New York Fed’s webpage “FAQs: Reverse Repurchase Agreement Operations”

Charging banks for their reserve balances when the Fed decides someday to go to negative interest rates is probably within the Fed’s legal authority, since the Fed can charge fees related to expenses, and it can be argued that handing funds to the Fed requires the Fed to put those funds into other relative safe assets such as T-bills, which would have negative rates in that situation. But it is worth considering other options that might be even easier legally. 

In that vein, I want to refine what I wrote in “How to Keep a Zero Interest Rate on Reserves from Creating a Zero Lower Bound” by suggesting that after capping the amount of reserves (+ vault cash) banks are allowed to hold to a little above required reserves (plus an extra allowance for small accounts they hold, in line with “How to Handle Worries about the Effect of Negative Interest Rates on Bank Profits with Two-Tiered Interest-on-Reserves Policies”), allowing banks (along with other counterparties) to put funds into the Fed’s Reverse Repo Program, with a negative rate, is the natural way to have the same kind of economic effect as a negative interest rate on reserves while leaving the interest rate on reserves themselves at zero.

I suggested this possibility at Jackson Hole last Saturday as one of several reasons that the Fed should keep its Reverse Repo Program in operation. This was my comment after Jeremy Stein’s brilliantly clear presentation of how the Reverse Repo Program can enhance financial stability in  “The Federal Reserve’s Balance Sheet as a Financial-Stability Tool,” along with his coauthors Robin Greenwood and Sam Hanson.   

As you can see in the screenshot at the top of this post from the New York Fed’s website,  

A reverse repurchase agreement conducted by the Desk, also called a “reverse repo” or “RRP,” is a transaction in which the Desk sells a security to an eligible counterparty with an agreement to repurchase that same security at a specified price at a specific time in the future. The difference between the sale price and the repurchase price, together with the length of time between the sale and purchase, implies a rate of interest paid by the Federal Reserve on the transaction.

While only regulated banks are allowed to put funds into a reserve account, many other financial firms as well as banks are allowed to put funds into the Fed’s Reverse Repo Program. That clearly makes it different from reserves. But I want to argue that other than the more diverse set of participants allowed, the Reverse Repo Program can, in effect, act as if it were the second tier in a two-tier interest on reserves structure, while reserves themselves act as the first tier. As in any two-tiered interest on reserves policy, once the first tier fills up to its cap, it is the lower interest rate on the second tier that serves as the marginal interest rate that matters most for markets.  

There may be institutional and economic subtleties I don’t understand, which I would be glad to be corrected on. But let me write based on my current understanding of the Reverse Repo Program (RRP) until one of you corrects me. The reason funds in the Reverse Repo Program (RRP) can act, economically, as such close substitutes for reserves, is that during the day they become reserves, while at night, for the bank or other financial firm, they become something else that can earn a different interest rate from reserves. That is, with RRP, the Fed is borrowing money overnight using T-bills as collateral, and then repays the money in the morning by crediting the lender with funds that are reserves in some bank’s reserve account at the Fed. 

Above, I said that in the policy I envisioned, reserves would be capped, but that is only partly true: the reserves banks can hold are capped at night, but banks are allowed to have what is potentially a much larger amount of reserves during the day. At night, to meet their cap, banks both on their own account and on behalf of depositors have to sweep reserves above the reserve cap into RRP. An individual bank might find another alternative, but that just shuffles reserves around to another bank. So in the aggregate, the funds above the cap have to go into something with the Fed on the other side of the transaction, and RRP is what is available.

“During the day, reserves; during the night another alter ego that earns a different interest rate” sounds structurally a bit like the blurb for a superhero. The Reserve Repo Program may not be a superhero, but it could come in handy. I hope the Fed keeps it in operation, just in case it is more useful in some future situation than is now appreciated. 

Postscript: One reason the Fed is considering discontinuing the Reverse Repo Program in the future is the fear that too many funds might flee to it in a future crisis, allowing some other markets to collapse. The solution to this is already built into the structure of the Reverse Repo Program: over short horizons, there is a level at which the rate in the Reverse Repo Program adjusts by auction rather than the quantity adjusting with the same rate. Then over longer horizons–that is, as soon as the Fed can meet, which might be within a week during a crisis–the regular rate for the Reverse Repo Program should be adjusted downward sharply to avoid that rate becoming an obstructionist lower bound.