Saturday, July 24, 2010

What Happens if the Interest Rate on Reserves Goes to Zero?

The Fed is currently paying interest on reserves at 0.25%. As I have discussed before, the interest rate on reserves (IROR) is currently the relevant policy rate for the Fed. It essentially determines all short-term rates of interest. On Tuesday, as reported here and elsewhere, a rumor spread that the Fed was considering dropping the IROR to zero. On the following day, Bernanke, in his session with the House panel, made clear what the Fed's position is:
We are paying one-fourth of one percent so it's obviously a very, very low rate of interest. The rationale for not going all the way to zero has been that we want the short-term money markets like the federal funds market to continue to function in a reasonable way because if rates go to zero there will be no incentive for buying and selling federal funds, overnight money in the banking system, and if that market shuts down ... it'll be more difficult to manage short-term interest rates, for the Federal Reserve to tighten policy sometime in the future. So there's really a technical reason having to do with market function that motivated the 25 basis points interest on reserves.

That being said, it would have a bit of an effect on monetary policy ...

(and) we'd certainly consider that as one option.
What would happen if the IROR on reserves went to zero? Would the fed funds market shut down? Currently there is a massive quantity of reserves in the system, but nevertheless there is an active federal funds market (though much less activity than in normal times). Why? Even though the system is awash in reserves, Fannie Mae and Freddie Mac don't receive interest on their reserve accounts, and therefore want to lend. A surprising feature of financial markets currently is that the effective fed funds rate (currently at 0.18%) is lower than IROR (0.25%) - there is something inhibiting arbitrage here, as without that the effective funds rate should be 0.25%. Now, if the IROR goes to zero, then the GSEs no longer have an incentive to lend overnight. Maybe the fed funds market essentially shuts down then. Presumably this is what Bernanke is alluding to, though he is not being explicit. But why do we care if the fed funds market shuts down? After all, it would shut down because it's not needed. Is it going to be costly to start it up again once the Fed starts to tighten? Do fed funds traders get out of practice, or what? I don't think this makes sense. Surely there is plenty of scope, if the Fed wants to, to subsequently tighten. It can do this in one of two ways: (i) Conduct open market sales of assets, until the fed funds rate rises above zero; (ii) Raise the IROR.

Now, what would be the effects of a decrease in IROR to zero? As Bernanke says, not much. Reserves would become slightly less attractive to banks, and in the course of trying to shed them, the price level would rise by a small amount, reserves would fall, and the quantity of currency (in nominal terms) would rise.


  1. Reserves are attractive to banks due to settlement needs. Attractiveness of reserves and their quantity does not change when Fed changes interest rates. If anything when interest rate falls to zero reserves become more attractive to banks.

    "Quantity of currency" - do you mean cash?

  2. 1. Interest is paid on reserves held overnight. Overnight there is no settlement. The transactions all happen during the day. Daylight reserves do no bear interest, but banks pay interest on daylight overdrafts (negative reserve balances). Reserves are just another asset on a bank's balance sheet, settlement or no settlement, and IROR has to matter to the bank in its decision concerning how much of the stuff to hold overnight.

    2. Currency:

  3. Banks with excess reserves (excess is defined by the bank itself and not by reserve requirements) lend reserves to other banks. Borrowing banks need reserves to settle transaction before the day-end since they are not allowed to stay in overdraft overnight. Banks will try to lend as long as overnight rate is above zero and exceeds cost of business. The support rate that FED pays is the floor for fed funds. The 0.18% rate is not surprising at all because it is not the rate that banks lend but GSEs which compete with each other while all banks with excess reserves go to FED. One could argue that 0.18% represent the cost of business for GSEs.

    When interest rate falls to zero banks requiring reserves for settlement might have to go to FED to get them because they will be less able to source reserves in the fed funds market. In any case fed funds market rate will be slightly above zero but this will be another "surprising" feature.

    Reserves are a system wide feature. Only FED can decrease or increase them. Banks use it for settlement only (ignoring medieval reserves requirements). So if one bank could try to do something about its reserves, banking system as whole can not.

    Cash is not going to change because cash is driven by settlement needs of NON-banking sector which does not care at all about any support rate that FED pays or does not pay to banks on their excess reserves.

  4. Stephen:

    Ideally, lowering the IROR would be part of a broader monetary policy strategy of easing. With that said, the stock market viewed this development very favorably, which implies the market believes such a move would have a meaningful effect. So even if estimated elasticities indicate otherwise, the effect of dropping IROR could be significant if the market believes it to be so. (e.g. stock market soars, confidence picks up, change in expectations leads to an increase demand for loans, banks are more eager to lend given the change in outlook, the excess reserves get lent out, etc.)

  5. Professor,

    I think that implicit in Chairman Bernanke's testimony it's the fact that a fed funds rate at zero might lead to a rate of return for Money Market Funds bellow its operating costs leading to another "break-the-buck" situation. This is something the central bank wants to avoid.

  6. "Now, what would be the effects of a decrease in IROR to zero? As Bernanke says, not much."

    There's about a trillion dollars of excess reserves that banks would release into the economy. You don't think this would be enough to make much difference?

    What formula or rule of thumb are you using to determine this?

    How much would it take to have a major effect?

  7. Richard, reserves can not be released into economy. Reserves are used for interbank settlement ONLY. In precisely the same way cash is used for retail settlement. Banks do not and can not lend reserves. Banks lend IOUs. This is a basic and simple fact which monetarism can not get. That is why you get all this QE non-sense. It is effect on lending capacity of banks is absolute zero.

  8. "There's about a trillion dollars of excess reserves that banks would release into the economy. You don't think this would be enough to make much difference?"


    No. With IROR at 0.25%, the banks are holding the reserves. Lower IROR to zero, and it doesn't make much difference. I'm not sure what planet Игры рынка is living on. Think about Canada. Currently, reserves go essentially to zero in Canada overnight. During the day, as is the case here, interbank settlement takes place through the Large Value Settlement System on the books of the central bank. Interest is paid on reserves at 0.50% overnight but, again, essentially no reserves are held overnight, with the interbank rate at 0.75%. Now, what we have currently in the US is quite different. There is a large quantity of reserves held overnight, and banks earn interest of 0.25% on the overnight balances. During the day reserves are used in interbank transactions (Fedwire). Of course the system is swamped with reserves, so we can think of the marginal liquidity value of reserves in daylight hours as being close to zero. Now, banks are content to hold the large quantity of reserves overnight at 0.25%. If IROR were higher, they would hold more, if IROR were lower, they would hold less. How much more, how much less? Not much.

  9. I am sorry, it is not me who is living on another planet but you.

    You said "If IROR were higher, they would hold more, if IROR were lower, they would hold less"

    It is FED who controls the amount of reserves. Not banks. Interest rate alone does not change the amount of reserves in the system. Only FED can do but doing something with reserves (e.g. OMO). Amount of reserves is completely independent from fed funds rate or any rate for this matter. And you if check the data on any day or around when Fed changed its target rate you will see that there is no change in the amount of reserves.

    But of course markets clear and interest rates adjust. However it happens on another planet. Not this one.

  10. 1. The Fed determines the quantity of outside money, M, where M = C + R. C is currency (remember that? - it's the quantity you wanted me to define), and R is reserves. The Fed determines M, then the private sector determines C and R. If banks try to shed reserves, the reserves end up as C.

    "But of course markets clear and interest rates adjust. However it happens on another planet. Not this one."

    You'll have to explain that one. You have some non-market-clearing model of financial markets in mind? What is it?

  11. "If IROR were higher, they would hold more, if IROR were lower, they would hold less. How much more, how much less? Not much."

    This is something I have to read more about, but given what you're saying maybe Scott Sumner's suggestion of a negative interest rate on reserves would inject a lot of money into the economy and be a good idea.

    Also, look at this graph of Excess reserves:

    It was at zero for essentially all of the last 50 years and then just recently spikes to about one trillion. Why can't we make it go back all the way down to zero like it always was instead of going down only "not much"?

    Finally, I did add two questions/comments to old posts if you're interested. I think they're good ones:

  12. 1. Yes, in principle you could have negative IROR. It may not be allowed by the Federal Reserve Act, though.
    2. Given the current IROR, the Fed could take excess reserves back to zero with open market sales of assets.

  13. Yes, FED can determine M which is C + R. However this approach died somewhere 20 years ago. Since then FED is targeting interest rates. And the way it works is that banks need R for settlement needs. If a bank is over-liquid it tries to loan out reserves on the fed funds market which is the market for R. If a bank is under-liquid it tries to borrow R on the fed funds market. Interest rate policy means that FED is targeting short term interest rate in this market. So when rate for R deviates from the target rate FED intervenes and adds or withdraws R through OMO.

    Settlement needs is just another side of liquidity coin. Each bank has to define its liquidity strategy. It can go wholesale and hope to be able to borrow R whenever it needs. It can have liquid assets. It can pile up R and so on. So now if we go back to autumn 2008 we will see a complete failure of interest rate policy. Banks liquidity needs increased which caused interest rates go through the roof. The situation has only settled down after Fed started providing pretty much unlimited liquidity to the banking system. At the same time Fed dropped interest rate to zero. And so you end up with excess reserves which are deposited at the FED. Rates are zero and FED is fine having 1 trln of extra liquidity in the system. This is no failure of interest rate policy. However it will have to withdraw this money one way or the other before it can start increasing its target rate.

    Now you see that R is used only for settlement and its amount is driven by the liquidity needs of the banking system. And Fed is the one being driven here and not the driver. By the way from this it also becomes absolutely clear that money multiplier concept is non-sense. It is non-sense on this planet given how the banking system operates.

  14. "Interest rate policy means that FED is targeting short term interest rate in this market."

    Not right now. It's the IROR that is the relevant policy rate. The Fed just sets that administratively.

    "By the way from this it also becomes absolutely clear that money multiplier concept is non-sense."

    Excellent. I agree with you 100%.

  15. Wow. There is base rate which all central banks target. There is support rate which is a floor and which fed pays now but did not pay before. And there is discount rate which is the ceiling. The rate for fed funds in the interbank market is allowed to fluctuate between the two.

    All of them represent interest rate policy. This is the way it works TODAY.

  16. As of yesterday (not TODAY): IROR = 0.25%; effective fed funds rate = 0.19%. Obviously the fed funds rate is not "fluctuating" between what you think is a floor, and the discount rate. Wow. Explain that to me.

  17. sure, this is what our host calls "strange effect". It is like this because GSE also have reserves which they try to lend out but do not have access to deposit facility at FED

  18. More or less. The GSEs have reserve accounts, but do not receive interest. I.e. reserves are zero-interest deposits for the GSEs. Without this wrinkle, the fed funds rate (taking out any risk premia) would be equal to the "floor," which is the IROR, since there is a positive quantity of excess reserves held overnight. I.e. The IROR effectively determines the fed funds rate. If the Fed increased the IROR to 0.5%, the fed funds rate would likely rise to 0.44%.

  19. Correct. That is the the goal of the support rate.

    Related point is the discount rate. It means that central bank will be forced to inject R into banking system if interbank rate hits the ceiling. As a consequence central bank cannot control M because it is driven by commercial banks desire to accumulate R which is equivalent to accumulating liquidity. And this happens during stressful times.

    But of course in normal times central bank does not wait until interest rates hits ceiling or floor and starts acting when interbank rate deviates from the target rate.

    And so money multiplier concept is non-sense. It does not apply to reality on this planet despite the fact that it is described in every mainstream textbook. It will be the case as long as central bank explicitly target interest rate policy as a primary tool of monetary policy. Please note that in fixed exchange rate systems money multiplier still applies because the goal of central bank is convertibility and not interest rate.


  20. We're getting there.

    1. Not every mainstream textbook has a description of the money multiplier. Mine doesn't:

    2. The money multiplier is certainly nonsense. We can never say that, for example, a given increase in outside money results in an increase in M1 of fixed multiple of the base money increase. Right now in the United States, this is true in a fairly obvious manner, but it's true generally, even if the central bank is not targeting a nominal interest rate, or if there is a fixed exchange rate.

  21. It's November of 2010 and the Fed is trying to mess with long term interest rates again in an effort to spur the economy. As many of the developing nations recently pointed out in the G20 meeting in Seoul, there is a serious worry that all this extra money being printed by Bernanke's Fed will just go overseas. Let's hope for the best on this one.