Interest on Reserves

The proposal, starting with comments from Senator Ted Cruz, that the Congress force the Fed to stop paying interest on reserves is having more legs than I thought it would. It also seems to be suffused with confusion — at least the AI and top three pages of hits generated by a google search certainly are. Hence this post.

One reason given is that it would save the government money, so including the proposal in the Big Budget Bill would help to hit deficit targets. In addition, interest on reserves is paid to a lot of foreign banks, and sending foreigners money so they can buy American goods is somehow out of fashion.

One answer: If you think that is a good and reasonable idea, here is a better one: stop paying interest on all Treasury debt. Reserves are just another form of government debt, so why stop there? That will generate $1 trillion per year, not in 10 years or so. And lots of foreigners hold Treasury debt too.

Well, duh, you can’t do that. When the government goes to bond markets for about $7 trillion (guesstimate) in order to pay back maturing debt and finance this year’s deficit, if the government says “oh by the way you won’t be getting interest,” nobody will buy the debt.

So, in the first instance, if the government says no interest on reserves, and if other interest rates are unchanged, then banks don’t want to hold reserves. They’d rather hold treasurys directly rather than via the Fed. (Remember, the Fed is basically a giant money market fund. It issues reserves, one-dollar-per-share assets, backed by treasurys.) They will only willingly hold reserves if they get at least the same interest on reserves as on treasurys.

There are then two possibilities: One, the Fed could allow banks to get rid of their reserves and hold only minimal quantities. Or two, the Fed could try to force banks to hold trillions of reserves at no interest.

For option one, the Fed would have to swiftly undo its quantitative easing: it would have to sell treasurys to soak up the reserves. Basically it would exchange its holdings of treasurys in return for reserves.

One problem with this option is that it’s not clear that the Fed has enough treasurys, on a market value basis, to do it. As interest rates rose, the value of the Fed’s portfolio tanked. I don’t have in hand the market value of the Fed’s portfolio (it is not a number that the Fed watches, unlike every other bank in the world) but it might not even be physically possible. It would certainly be embarrassing to realize just how much money (taxpayer money, ultimately) that the Fed lost via QE, i.e. shortening the maturity structure of government debt.

Option two: governments have, in the past, gotten away with paying no or low interest on debt basically by forcing banks, people, and businesses to hold it anyway. This is called “financial repression” and it is basically a tax. Some possibilities with reserves amount to the same thing. Back when the government did not pay interest on reserves, it also forced banks to hold some reserves in proportion to their deposits. But that was only a few tens of billions And now there are no more reserve requirements. But when it tapered QE, the Fed discovered that all the Dodd-Frank regulation forces banks to hold about a trillion in reserves. So it might be able to keep that much outstanding and essentially tax the banks by forcing them to hold that much reserves without interest. The Fed usually likes to subsidize rather than tax banks. And taxes come from somewhere, including lower deposit rates and higher lending rates. This outcome would just further erode the traditional banking system in favor of fintech, stable coins, etc. Maybe that is a good thing, but probably not the good senator’s intentions.

I’ll put in a plug here for my favorite idea: The Fed should get away from trying to set both the price and the quantity of reserves. If the Fed simply set the interest on reserves (and a slightly higher interest rate on a stigma-free discount window) it could target interest rates and never worry if it was providing enough or too much.

Just how does the government make money on this anyway? Wouldn’t the Fed not paying interest on reserves just help the Fed? Answer, no. The Fed makes money on the interest on the Fed’s treasury holdings. The Fed pays interest on reserves. Usually the former exceeds the latter, and the Fed gives most of the profit back to the Treasury. Really all the Fed does here is “ride the yield curve.” Long term debt pays higher interest than short term debt. The Treasury could have made the same money — and taken on the same risk, now coming due — by issuing overnight debt in the first place. Right now, the Fed’s long term bonds pay less (coupons, the Fed ignores market values) than interest on reserves, so the Fed is losing money, and not sending any of the Treasury’s interest payments back. So, by eliminating interest on reserves, and somehow forcing banks to hold a few trillion of those reserves, the government trusts that the Fed will give the Treasury back most of the interest the Fed receives on Treasury debt, thus lowering the Treasury’s interest costs on the debt. We’re back to “just don’t pay interest on the debt,” with a convoluted hope that people will hold zero interest deposits backed by zero interest reserves, people who would not hold zero-interest treasurys.

Another answer is, “inflation.”

If the Fed stops paying interest on reserves, as many have pointed out, banks will then try to dump reserves and buy treasurys instead. This will force up treasury prices, and down treasury interest rates. Indeed, changing the interest on reserves is exactly how the Fed tries to move all market interest rates. So eliminating interest on reserves is equivalent to commanding “set the interest rate target to zero.” Given President Trump’s desire to see lower interest rates, maybe that’s what’s at work.

Monetarist instincts say that forcing zero interest on reserves is just like a huge open market operation of what were essentially bonds (reserves that pay market interest) for money (that does not pay interest). If the quantity of reserves were fixed, then this would be like dropping two trillion of cash on the economy, inflationary.

I’m less of a monetarist these days. And the economic mechanism by which lower interest rates cause inflation is a bit muddy. But for this blog post, let’s just leave it that commanding zero interest on reserves, and maintaining large quantities of reserves (so the government saves any money), is exactly equivalent to commanding that the Fed lower its interest rate target to zero. And that is usually thought to be inflationary.

A closing thought. Much of what I have seen on this has a sort of hazy nostalgia about the good old days with zero interest on reserves, a small amount of reserves, banks lending to each other, and the Fed setting interest rates by rationing reserves. This is about as sensible as most nostalgia. The good old days were awful. Ample reserves that pay market interest are a great innovation. They cost nothing. In fact they save the government money, as reserves typically pay less interest than longer maturity debt. The banks would otherwise hold directly the same treasury debt that the Fed holds, and there would be no rebate. It would be better still if the Treasury offered the same security directly: fixed value, floating rate, electronically transferable, overnight treasury debt. Lots of artful cash management, “settlement Wednesday” market spikes, were just wasted time and effort. We can live the Friedman rule, money pays the same interest as short-term debt. And the zero bound era shows that ample interest paying reserves do not cause inflation. Let’s not throw out this lovely innovation simply because Congress cannot, in a bill larded with silly spending and tax exemptions, find a way to hit reconciliation budget targets other than by an implicit tax on banks.

Update:

Armando Rosselli below asks an important question:

How does the Fed make IOR payments (which are lower than the Treasury coupons holdings)? Printing money?

Answer: Yes! First, of course, the Fed uses the interest it receives on its assets to pay interest on reserves. But when, as now, those are insufficient, the Fed can and does simply create new reserves to pay the interest on old reserves. Unlike private banks who try such a thing, the Fed can do it. After all, the Fed’s liabilities — reserves — only promise to pay more reserves. The Fed can never go bankrupt. The real value of reserves can decline — inflation— but the Fed can’t run out of money. Remember that reserves are just another form of government debt, and the government can print money to pay off any debts. And when the Fed runs out of assets to buy back reserves with, its ability to soak up reserves and thus fight inflation is obviously diminished. So, the Treasury may end up having to recapitalize the Fed if that goes too far. Our future may hold many interesting and unexamined lessons for the intersection of fiscal and monetary policy.

Update 2:

What is the mark to market value of the Fed asset portfolio? Could it sell enough assets to soak up all but $10 billion of reserves? George Hall specializes in this sort of question, having constructed with Tom Sargent time series of the market value of all US debt. George writes:

The NY Fed reports the bonds in its SOMA portfolio, [newyorkfed.org] and I then obtained the prices from the CRSP US Treasury database. I've lost access to the quarterly CRSP updates, so my latest numbers are from December 2024.

In December 2024, the face value of the Fed's Treasury portfolio was $4.2 trillion, and my estimate of the market value was $3.7 trillion (11% below par).

I plot the market and par values in figure 13 on page 31 of this paper with Tom.

Currently, the Fed's balance sheet looks something like:

Assets Liabilities

Treasuries $4.2 trillion Reserves $3.4 trillion

MBS 2.2 Reverse repos 0.5

Other stuff 0.2 Currency 2.3

Treasury deposits 0.3

------------------------------------------------------------------------

Total $6.6 trillion $6.6 trillion

Computing the market value of the Fed's MBS portfolio takes some heroic assumptions. But if both the Fed's Treasuries and MBS are worth only 89% of their face value, I think they could still redeem the reserves and reserve repos.

Still the conceptual question is worth pondering. In the long run it is not obvious that the yield curve will always generate a liquidity premium— an upward slope greater than the expectations hypothesis value which generates no long-term profit. Then, if zero interest currency declines central banks will face negative profits and need treasuries to subsidize them!