Interest on Excess Reserves: The Hobie Cat Effect

interest on excess reserves, bank reserves, Federal Reserve, operating framework, floor system
Source: Hobie Cat Tiger by Hanninen Markku (CC BY-SA 2.5)
interest on excess reserves, bank reserves, Federal Reserve, operating framework, floor system
Image by Hanninen Markku

Forty years ago, in the spring of 1978, I had no intention of becoming an economist. Instead, I was studying marine biology at Duke University's Marine Lab at Pivers Island, on the beautiful North Carolina Coast. There, when the wind was up, my classmate Alan Kahana and I enjoyed going out on his Hobie 16, with Alan manning the tiller and myself hiked-out on the trapeze. We weren't, truth be told, especially prudent sailors. On the contrary: we were so inclined to push things to the limit that one day we took the Hobie out just as a gale was getting up, and ended up…well, that's a long, sad story. Suffice to say that it doesn't take much to capsize a Hobie, and that on that day we capsized Alan's boat once and for all.

What, you are no doubt wondering, has any of this to do with Interest on Excess Reserves? I'm getting there. You see, although it doesn't take much to capsize a Hobie — a little over-trimming of the sail will suffice — once one capsizes, it's likely to start to turn turtle as its mast fills with water. And as that's happening, it may be all that two reasonably trim lads can do — by pulling for dear life on a righting line attached to the boat's mast, whilst leaning backwards on its uppermost hull — to lever the thing back upright. The more the mast fills, the harder it gets. And the same sort of thing goes for letting a central bank slip into, and then trying to wrest it out of, a floor system of monetary control: the more liquidity the banking system takes in while that system's in place, the more effort it takes to pull out of it.

As faithful Alt-M readers know, I've long insisted that the modest interest rates the Fed began paying on excess reserves in October 2008 were enough to encourage bankers, who long made do with only the slimmest of excess reserve cushions, to hoard all the reserves they could lay their hands on. That modest little bit of Fed sail-trimming was enough to overturn  the Fed's traditional monetary control system. The Fed had long relied on a sort of asymmetrical "corridor" system, with a target fed funds rate set somewhere between zero and the Fed's discount rate, and the effective federal funds rate kept near that target by means of small-scale open market operations. Now it had flipped-over to a "floor" system, with changes in the Fed-administered IOER rate serving as its chief instrument of monetary control.

Some economists, to be sure, refuse to believe that the modest IOER rates the Fed paid in the early stages of the crisis could account for the switch in question, or for banks' subsequent tendency to hoard reserves. Paul Krugman even accused those who thought so of failing a “reality test,” by overlooking how, in the U.S. in the 1930s and in Japan more recently, banks hoarded non-interest-bearing reserves. But Professor Krugman himself might be said to have failed a "logic test," calling for an understanding of the difference between a necessary and a sufficient cause. Then there's the pesky fact that Ben Bernanke and other Fed officials secured permission to pay interest on bank reserves for the express purpose of getting banks to hoard them. Had they done so for no reason? Had Ben Bernanke himself forgotten about the 1930s? A page from his 2014 textbook is illuminating (HT: Alex Schibuola):

But allowing that a modest above-zero return on bank reserves was indeed all it took to establish a floor system, and to get banks to stock-up on excess reserves, it doesn’t follow that restoring the IOER rate to zero would have the opposite effects. The reason has to do with the immense growth in the total supply of reserve balances that has since taken place. That growth matters because, under a floor system, the greater the nominal stock of bank reserves, the lower the IOER rate must be to reduce the quantity of excess reserves demanded to zero. The accumulation of liquidity in a "floored" monetary control system thus acts like the accumulation of water in the mast of a capsized Hobie Cat, making it much harder to revert from a floor to a corridor system than it was to switch to the floor system in the first place. Call it the Hobie Cat effect.

The Hobie Cat effect can be illustrated formally using a diagram showing the supply of and demand for bank reserves or federal funds under a floor system. The supply schedule for federal funds is, as usual, a vertical line, the position of which varies with changes in the size of the Fed’s balance sheet. The reserve demand schedule, on the other hand, slopes downward, but only until it reaches the going IOER rate, here initially assumed to be set at 25 basis points. At that point the demand schedule becomes horizontal, because banks would rather accumulate excess reserves that yield the IOER rate than lend reserves overnight for an even lower return.

For the initial stock of reserves R(1), starting at the equilibrium point “a,” a slight reduction in the IOER rate would suffice to get the banking system back onto the sloped part of its reserve demand schedule, at point “b,” where reserves are again scarce at the margin. But once the stock of reserves has increased to R(2), it takes a much more substantial reduction in the IOER rate — perhaps, as the move in the illustration from “c” to “d” suggests, even into negative territory — to make reserves scarce at the margin again, and to thereby make switching to a corridor system, by a modest further reduction in the IOER rate, possible without any need for central bank asset sales.[1]

Does this mean that the Fed can never hope to escape from its current floor system unless it reduces its IOER rate substantially, and that it might even have to resort to a negative rate? It doesn't. And it's here that the Hobie Cat analogy fails, for while you can't drain the mast of a Hobie Cat that's turned turtle, the Fed can drain the banking system of any or all of the reserves it created after 2008. The obvious, and most prudent, way out of the floor system is, therefore, for the Fed to retrace the steps that got it into that system, by first shrinking its balance sheet far enough to return the stock of reserves to  a point close to the kink in the federal funds demand schedule, and then reducing the IOER rate enough to make reserves scarce at the margin, thereby reviving interbank lending and establishing a corridor system.

Considering how many excess reserves banks are now holding, all of this is still a tall order. But it beats having an operating framework that leaves our monetary system sodden and adrift.


[1] Because the move from “c” to “d,” like that from “a” to “b,” involves no change in the total stock of bank reserves, readers may be tempted to assume that it also involves no reduction in the quantity of excess reserves, and hence no change in banks’ inclination to hoard such reserves. The temptation should be resisted: although banks hold the same total quantity of reserves at “d” as at “c,” the former equilibrium involves a higher quantity of bank lending and deposit creation, hence a higher value of required reserves, with a correspondingly lower value of excess reserves.


  1. The most delightful article on monetary policy I've ever read.

    I will definitely go back over it later today and grind my way through to understanding the simple math of floors and corridors (it never becomes intuitive, no matter how many times I go through it).

    For that second reading, I plan to try to skip over the dream-like images of being young again, being on an Carolina coastal island, and hiking out over the windward hull of a Hobie 16.

    That metaphor is a lovely invention to help one with my mental limitations. I shall evermore think of liquidity as saltwater.

    1. Thanks for the kind remarks! I hope the second reading proves rewarding as well.

      1. @george selgin would you be so kind as to help me with my course on Monetary Policy – a part of a new design for an entire economy which has gained me a promise of two Nobel Ptize nominations for economics and may also attract a peace prize. But that is not for the monetry policy part which needs to be carefully added.

  2. I’d go even further than the above article: like Milton Friedman and Warren Mosler, I don’t see any reason for the state (i.e. government and central bank) to ever pay any interest to anyone for holding a stock of state liabilities, i.e. base money. I.e. the national debt should be abolished.

    One excuse for the state issuing so much liability that it then has to pay interest to holders thereof some interest to dissuade them from trying to spend away that liability, is that come a recession, central banks can cut interest rates. But a flaw in that idea is that it amounts to forcing everyone with a mortgage to pay extra interest just so that central banks can do interest rate adjustments: daft.

    An apparent problem in abolishing or more or less abolishing interest rate adjustments is that implementing stimulus then has to be done via fiscal policy, and it can take time for the collection of economic illiterates known as “politicians” to come to decisions in that connection. In fact the latter is a defect in the system (especially in the US) that can in principle be rectified. In the UK, it is common for the finance minister to announce changes to the sales tax (VAT) and the tax on fuel, the for those changes to be effected within 24 hours.

    Bernanke actually gave his blessing to a system where to effect simulus, the state simply creates new money and spends it, and/ or cuts taxes. See para starting “A possible arrangement…” here:

    1. "like Milton Friedman and Warren Mosler, I don’t see any reason for the
      state (i.e. government and central bank) to ever pay any interest to
      anyone for holding a stock of state liabilities, i.e. base money. I.e.
      the national debt should be abolished." Of course proponents of IOR regularly appeal to Friedman's "Optimum Quantity of Money" argument in defending it. I myself don't oppose IOR as a means for limiting the implicit reserve requirement tax. But the Fed's IOR payments do far more than that.

  3. Wouldn't it be a lot easier if the mast were water-tight, or at least filled with styrofoam?

    1. Yes. Some people do fill the masts w/ foam. But the set of people who ride Hobie Cats in gales and that of people responsible enough to make sure their masts won't sink hardly intersect.

  4. Looking at George's diagram, and his policy proposal to reduce the Fed's balance sheet, let's ask how that will work? Fed holds Treasury bond assets, and owes excess reserves to banks. Selling its assets to reduce the problem will give banks and investors a large increase of liquid Treasury bonds. Perhaps a bond market interest rate movement?
    Assume the Fed then coordinates a drop in IOER to reflect the shifted quantity of ER, closer to the point 'a' on the diagram. Ideally, a non-binding floor IOER could serve to promote larger liquidity, a hedge against future stress? Normal FedFunds would then trade above IOER and not push down to it (not very often).
    Depending on how large the government's balance might be at that time, in the US Treasury account at the Fed, it might be possible simply "to redeem" them, or "sell them back to the Treasury." That ought to cancel both sides of the Fed's swollen balance sheet at one time.

    1. Joe —
      Good to hear from you!
      The Treasury does hold over $200 B on deposit with the Fed, far more than the $5 B that was adequate before 2008. I suspect that this is a "cookie jar" for impulse spending that is already appropriated and within the bloated debt limit. If Congress forced Treasury to use this to buy back bonds from the Fed, it would reduce the debt and revert control over this spending to Congress. But it would make only a small dent in the base overhang.

      1. Of course Congress could mandate the Treasury, but I was thinking let the Fed itself "repo" (or whatever) the bonds back to Treasury, and then just absorb/dissolve the Fed's balance sheet liability to Treasury.
        Shrink balance sheet; "shrink" assets = liabilities shrink.

        1. Yes, the Treasury could use its $200B balance to buy back $200B in bonds, and the Fed could simultaneously sell $200B to the market. This would reduce the base and the debt by $200B, but would require the Treasury and the Fed both to voluntarily disaggrandize themselves. It's more likely it would take a push from Congress.

          1. As Selgin points out, the problem with IOER is a rate above where FedFunds might be, thus a binding floor. Lower the floor, low enough to observe interbank-reserve lending occurring in the market. Lifting and lowering the IOER rates should have the same effect as the former Open Market method.
            We are still cursed, of course, with a "central planning" model, which is itself the best reason to sell Congress on the idea of a free-banking model and demote the Fed from its "church" status. Of course, Congress ought to demonopolize the "dollar" and reconsider the fallacies of monetary nationalism.

          2. Actually, the balance sheet has to shrink one way or the other, for if IOER is reduced dramatically with no contraction of the balance sheet there will be some serious over-loosening of policy.

  5. But what if the Fed simply follows the market? What if money is largely neutral? To test this, we'd have to see how much banks kept their reserves during recessions pre-2008, compared to now. I'd suspect that the more severe the recession, the more reserves banks hold, and, conversely, the bigger the expansion, the fewer the reserves. Once again throwing cold water on the idea of monetarism in general. Indeed, during the "free banking" era of the 19th century, monetary policy was non-existent, and GDP per capita was about the same as now, and if anything monetary policy as viewed from today's vantage was pro-cyclical. Further, back then there was a genuine need for small denomination money (like the penny) but they were non-existent, so much so that people often resorted to script or foreign coins (from a reading of Timberlake's tome on monetary history), yet the economy did fine, again proving money is largely neutral.

  6. Dr. Selgin, I just heard Wells Fargo's head of rate strategy say that the yield curve is broken due to IOR and the Fed's massive balance sheet so it doesn't necessarily predict a recession when flat. Have you written anything on the yield curve related to IOR? Thanks!

    1. I have not, Roger. But I believe that Wells' man has a point. The flat curve is nevertheless bad news for all sorts of reasons.

      1. IOR transformed bank reserves from cash to 1-day T-bills. Looked at that way, the Treasury yield curve has been continuously nverted at the very short end (30-day T-bill rate minus 1-day T-bill rate) since IOR commenced, except for March 2018. It is obvious from looking at commodity prices that money has been too tight since mid-2014. The Fed needs to forget about interest rates and target commodity prices. And, for it to be able to target anything, it has to phase out IOR.

Comments are closed.