Understanding the Permanent Floor—An Important Inconsistency in Neoclassical Monetary Economics

By Scott Fullwiler

I’ve written numerous times already about how a deficit “financed” by bonds vs. “money” doesn’t matter in terms of inflationary effect.  Notwithstanding my views there (which are not discussed in this post), the point of this post will be to explore the neoclassical paradigm on this matter, since this is at the core of the recent debate between Steve Randy Waldman (see here, here, and here) and Paul Krugman (see here and here) on the so-called “permanent floor.”  (It might be of interest to some that I explained how a “permanent floor” would work back in 2004.)

Let’s consider a time at some point in the future at which the Fed has ceased its current near zero interest policy, IOR, and QE’s, and has completed whatever exit strategy was deemed necessary to drain the reserve balances that the various rounds of QE produced.  In terms of the federal funds market, it would look much like it did before the crisis, as in Figure 1.  The target interbank rate (iinterbank*) is at some positive rate of interest selected by the Fed in accordance with its reaction function, and the Fed has set its penalty rate at some spread above its target rate.  The demand for reserve balances is roughly vertical at the quantity banks desire to hold to settle payments and meet reserve requirements (RB* in the graph), as is now recognized even in neoclassical literature. (Post Keynesian endogenous money folks said this for decades; I explained in my 2002 paper why there could logically be no liquidity effect in the federal funds market when the Fed changes its target rate.  I’m glad neoclassicals are catching up to us here.)  The quantity of reserve balances supplied by the Fed is essentially a vertical supply curve (dotted vertical line) where the Fed simply attempts to accommodate the demand at the target rate, which is well documented in empirical literatures of central banks and also neoclassical economists to be the actual practice of central banks.  (I have argued they have no choice but to do this.)

Figure 1—Federal Funds Market Post QE

Now, what if the Fed were to decide to raise the monetary base via QE again?  Before answering, let’s recognize that neoclassical economists all seem to believe there are two things that stop QE from “working” in the sense that they stop the money multiplier and/or the quantity theory of money in their tracks.  These are the following:

  1. Interest on reserve balances at the target rate, since they believe this is an opportunity cost to bank lending (see here and here as just two examples of literally dozens; I explained why this is incorrect here)
  2. Interest rates at the zero bound, since then “money” (i.e., “dead presidents”) and t-bills are now perfect substitutes.

I don’t know the exact proportion that believe in both of these versus just one of them, but I’ve seen enough that it seems most believe in both.  Note for #1, reserve balances and t-bills are perfect substitutes there, too, while for #2 the rate paid as interest on reserve balances (zero) would be equal to the prevailing interest rate (zero).  So, there’s nearly complete overlap between the two points.  In short, in either case, the quantity theory of money now does not run from “money” to inflation, as velocity falls in kind with the increase in “money.”

What happens if the Fed starts QE again given the context set in Figure 1?  Obviously, the overnight rate falls to zero, since the quantity of reserve balances supplied quickly dwarfs banks’ demand for them (the demand for reserve balances was about $15-$20 billion prior to 2008).  This is shown in Figure 2, where there is an increase in reserve balances from RB* to RBQE and iinterbank has fallen to zero.  Note that this is just supply and demand—an increase in quantity supplied well in excess of quantity demanded at any price simply leads the price to fall to zero.

Figure 2—Federal Funds Market with QE

How can we avoid the price falling to zero?  We can pay interest on reserve balances, which now becomes the new target rate.  This is shown in Figure 3, where the increase in reserve balances to RBQE now results in the interbank rate settling at the rate on reserve balances (iremuneration), which effectively becomes the central bank’s new target rate.  Note again that this is just supply and demand—an increase in quantity supplied well beyond the quantity demanded at any price will require a price floor in order to keep the price above zero.

Figure 3—QE with Interest on Reserve Balances

You probably see where this is going now, if you didn’t several steps ago.  To actually carry out QE, the Fed must either accept an interest rate target at zero or pay interest at the target rate if it wants an interest rate above zero.  But these are precisely the two cases described above for which QE doesn’t work in the neoclassical view.  The only other thing the Fed could do would be to reverse the QE and drain all the excess reserve balances, but this simply reverses the QE as if it had never happened.

What if instead the Fed does QE by purchasing treasuries or other securities with “cash?”  Note that currently this isn’t operationally possible—the Fed settles all of its transactions on Fedwire, which only settle with reserve balances (and US Treasuries only settle on Fedwire), but let’s play along since at least some neoclassicals think this can actually work.  The basic point neoclassicals want to make here is that they believe the Fed can exogenously control the quantity of currency (i.e., “dead presidents”), and thereby generate a “hot potato effect” as the “excess cash balances” created by this form of QE are spent repeatedly by individuals trying to rid themselves of so much cash and hold computers, toothpaste, toasters, new homes, couches, and so forth instead.  (Krugman doesn’t explicitly discuss exogenous control over currency in the debate with Waldman, but he did explicitly argue that the Fed could do this here, which I then countered here.)

Unfortunately for neoclassicals, in the real world we have these things called banks (which have still not made it into neoclassical models of the macroeconomy aside from the crude and inapplicable money multiplier model).  And banks offer these things called checking accounts, savings accounts, money market accounts, time deposits, and so forth.  And individuals that do not want to hold their wealth in the form of cash (i.e., have “excess cash balances”)—QE is, after all, at core a portfolio swap out of Treasuries and into reserve balances, not an increase in income or wealth, or “cash” in this example in which extreme license is being taken with how the Fed actually does things—can costlessly convert their cash into these checking accounts, savings accounts, money market accounts, and time deposits.  This conversion occurs at par value as required by law, and it leaves banks holding the “cash” in their vaults while they’ve credited the accounts of their new depositors.  This ability to convert is thus also unlimited as crediting the accounts of depositors simply requires changing numbers on a balance sheet and accepting more vault cash (i.e., there is no “limit” to the quantity of deposits that can be created via conversion).  (The same process stops any sort of “hot potato effect” that might occur beginning with deposits instead of currency—the ability to costlessly convert undesired balances to savings, time deposits, and so forth is again obvious and unlimited.)

So, what do the banks do with all their new vault cash that is well in excess of what they expect their customers to withdraw?  They sell it back to the Fed in exchange for reserve balances, and the currency is now out of circulation.  But now there is a large excess of reserve balances—what does the Fed do about that?  As above in Figures 2 and 3, it either accepts a zero interest rate or pays interest on reserve balances at its target rate if it wants a non-zero rate.  And now we’re right back to points 1 and 2 that are the very things neoclassicals agree stop QE in its tracks.  Or, the Fed can sterilize by selling securities to drain the reserve balances, which again just reverses the QE.

What does all this mean?

First, the quantity of currency circulating—“dead presidents”—can never be exogenously controlled in a world in which we have banks (!) that costlessly convert currency to deposits, savings, and so forth, effectively taking the currency out of circulation, which banks then convert to reserve balances.  There is no hot potato effect for currency in the real world.

Second, whenever the Fed attempts QE, it must accept a zero interest rate or pay interest at the target rate—the only other choice operationally is to argue that the “laws” of supply and demand have been suspended.  Krugman in fact does this when he argues that the Fed could expand the monetary base without sterilizing and also without paying interest on reserves (his option 3); he then claims this would be inflationary while also arguing it would not be inflationary if instead interest were paid on reserves.  But from Figures 2 and 3, again, either this leaves the federal funds rate at zero (no sterilization) or at the target rate (interest on reserve balances), and neither of these are inflationary in his own view.

What neoclassicals here can’t seem to accept is what Post Keynesians have known for decades and some New York Fed researchers explained in crystal clear language a few years ago, namely that “the cost of reserves, both intraday and overnight, are policy variables.  Consequently, a market for reserves does not play the traditional role of information aggregation and price discovery.”  The Fed simply must set a price—it cannot do otherwise.  Not setting a price of reserves when it attempts to expand the monetary base simply means setting the price of reserves at zero.

Finally, because neoclassicals believe that a zero interest rate target or interest on reserve balances at the target rate stops QE in its tracks, then they are left with no real-world situations in which these can actually create inflation from within their own paradigm.  In other words, unless they are willing to suspend belief in supply and demand, they must suspend belief in QE “working” (at least the quantity effects of QE) in order to be logically and internally consistent.

156 Responses to Understanding the Permanent Floor—An Important Inconsistency in Neoclassical Monetary Economics

  1. Scott,
    Fantastic post. Hopefully this will clear up much of the confusion in the current debate and if we’re lucky, change some minds in the neoclassical camp. As a follow up, do you have any opinion on whether or not the Fed will choose to use IOR as a permanent floor for monetary policy going forward?

    On a somewhat related topic, I have a question regarding fiscal spending with respect to the platinum coin. Under pre-2008 circumstances, when the government deficit spends, either it or the Fed could sell debt to prevent a corresponding increase in reserves. By law, I believe the Treasury was required to sell debt in these instances. Now let’s say the Treasury deposits a $1 trillion platinum coin at the Fed in return for reserves (I presume). Does this increase the monetary base? Or does the base not change until the government spends those reserves? If the latter is true, under this arrangement, does Treasury still have a legal obligation to sell debt? If not, I could see how the platinum coin gives Congress/Treasury some direct power over the monetary base.

    In advance, thank you for sharing your expertise.

    • Scott Fullwiler

      reserve balances wouldn’t increase until the Tsy spent, as the Tsy’s account doesn’t count as reserve balances in circulation. Or, if the Tsy moved the balances to TT&L accounts at banks, which wouldn’t make any sense, but hypothetically it would raise reserves. To my knowledge, the Fed is not yet allowed to issue time deposits or sell it’s own debt. several other CBs do this, though, and there was some talk about giving the Fed this authority when it comes time for it to “exit” the QE rounds and “return to normal.” Obviously, I don’t think an exit is necessary, and neither do neoclassicals if they are true to their own views on the effects of IOR and the zero bound as I described in the post.

    • Scott Fullwiler

      BTW, Krugman totally nailed how the coin operations could work and the effects on Fed operations and inflation at least from his neoclassical perspective. Credit where it’s due . . . .

  2. This article just blew me away. (I’m not an economist or financial person.)

  3. Why do neoclassical believe that cash causes a hot potato effect, but liabilities in checking accounts – which you generally have to pay to use – do not?

    Why do they believe that interest rates as low as they are at the moment induce people ‘not to spend’ when they probably lose more than they receive in interest a month due to transaction mistakes in the bar on a Friday night?

    Is it that old Ivory Tower thing?

  4. The only way for the Fed to deliberately create inflation is to directly put money into a sector where there is already more supply than demand. The Fed HAS managed to do this, by its own admission (bragging, actually) in the secondary asset markets (stocks, bonds, etc.), but, again by its own admission, this has not made it into the real economy in the form of loans (i.e. credit-money). Only government can put money directly into the real economy in this way IF there is no private demand for credit-money.
    The government can inject non-credit-money (e.g. coins, including the trillion dollar coin, if it choose to use it that way, or United State Notes, towards public works projects). Given the depressed economy, i.e. economic under-utilization, it would actually take quite a bit of stimulus to raise wages this way anyhow, though, since wages haven’t risen for the bottom 80% in 40 years, this would be a good thing if they did!
    Money always seeks the greatest return. There is little need to invoke complicated formulas when you can just “follow the money” to see where it is going to get the greatest return (allowing for risk).

    • Scott Fullwiler

      regarding Fed, you’re talking about the Fed changing relative prices. I agree there. my point is that this is not the qty affecting things then, since in the case we agree upon the most efficient thing for the Fed to do is name a price and allow qty to float (so then, obviously, in that case the point is the price, not the qty). it’s obviously doing the opposite.

  5. Shining Raven

    Thank you for the post, it is indeed very timely, I was looking for exactly this after seeing the discussion that you reference.

    You do write something on this (and the post actually came up when I searched for the expression), but can you point me to an explanation of how the “hot potato effect” (that Scott Sumner and Nick Rowe keep going on about) is actually supposed to work in the monetarist view, and a clear explanation what is wrong with that view? This post goes a long way to explain it, so this helps. Is this by any chance related to the “law of reflux”, or is that a completely unrealted issue?

    • Scott Fullwiler

      don’t have best sources on HPE in mind. you might just google it (?). I can see where you’re going with law of reflux, though that has to do with retiring debts, so I wouldn’t put them in the same category.

      • Shining Raven

        Thanks, I did google again and I found a couple of places at Nick Rowe’s where it is discussed.

        Good discussion by JKH and K here, and in particular good comment by K at this point:

        http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/09/goldpunk-strategy-space-and-michael-woodford.html?cid=6a00d83451688169e2017c31b5df3f970b#comment-6a00d83451688169e2017c31b5df3f970b

        There is a second relevant post which even has the title “Money is always and everywhere a hot potato” which discusses the “law of reflux”. Close relation to the comment by K referenced above:

        http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/08/money-is-always-and-everywhere-a-hot-potato.html?cid=6a00d83451688169e2017d3bd482d3970c#comment-6a00d83451688169e2017d3bd482d3970c

        Comment by Sergei along the same lines: As long as the central bank wants to maintain a positive interest rate policy, it has to convert excess (non-interest bearing) reserves back into interest-bearing treasuries (or pay interest on reserves). This means that the “law of reflux” is true by the decision of the central bank to maintain a positive interest rate, since it must offer convertibility of reserves into something interest-bearing, so that an excess “reflows” into an interest-bearing asset.

        I think that was pretty much what I was looking for (and I actually have read this once before, but had to develop a better understanding before it stuck…), also regarding the reflux-idea.

        • Scott Fullwiler

          OK, I see your point on reflux. I think of reflux within the context of the monetary circuit, so a bit differently from what I’m seeing there. But even from that perspective, one could think of there being no hot potato in the sense that banks always accept both their own and the CBs liabilities and always convert both at par, and then the CBs always accept their liabilities and allow free conversion among different types of CB/Govt liabilities.

          • Scott Fullwiler

            Is that where you were going with this?

            • Shining Raven

              Yes, I think so.

              I guess my point is that the banks can extinguish their own liabilities when they get them back, or to say it differently, people can use bank money to pay back what they owe the banks, which extinguishes that money. So there is no “hot potato” that needs to be passed around, if there is too much money created in lending, people pay it back to their banks to extinguish the corresponding debt, and there is no more money to pass on.

      • Good post.

        The problem here is that there are multiple definitions for the hot potato effect. My guess is you’ll have no problem with this one from David Beckworth:

        “triggering a cycle of portfolio rebalancings (i.e. the “hot potato” effect)”

        So HPE = to have too much of something and want to get rid of it. Seems reasonable to me.

  6. Thanks for your contribution to understanding the discussion between Waldman and Krugman. Still a bit heavy stuff to me, but it will be interesting to see who is proved right.

    A typo you’ve made throughout: Waldman writes his name with one n.

  7. Scott,

    Thanks for this post. I am definitely one of the nerds intently following this debate, so I love these kinds of posts, for some odd reason.

    I think I have read all of your papers on monetary operations (at least those publicly available), but I must embarrassingly confess, I still only get half the story, but not the whole. Pre IOR, what enables the Fed to stabilize the FFR at *any rate* between the ceiling and 0%, if demand is virtually interest inelastic and there is, logically, no liquidity effect? To explain where I am coming from, I realize the Fed MUST supply the quantity of reserves banks desire to settle payments and meet reserve requirements. But what sets the price at that quantity supplied, and *why*? I realize the evidence shows the FFR simply moves to whatever the Fed announces (the announcement effect), and that there is no liquidity effect that makes it move (i.e., net changes in reserve amounts across the maintenance period driven by “offensive” OMOs). But *why* will the market simply move to where the Fed says? My question is similar to the criticism aimed at Nick Rowe for his Chuck Norris analogy of the Fed and NGDP targeting – what is the real threat, the transmission mechanism that could be used if need be, that backs up the Fed’s announcement that they want the rate to move from x% to y%? (And for that matter, what got the rate at x% in the first place?!)

    It’s clear to me how IOR works in setting the FFR. The channel system, where there is a gap between the ceiling and the floor, is what I find opaque.

    • wh10,

      Scott will no doubt jump in if I have this wrong, but my understanding is that without IOR, since reserve balances are non-interest-bearing assets, banks have no interest in holding excess reserves, and so market demand itself is enough to establish a “natural” floor. If banks in the aggregate have more reserves than they need, they will swap those reserves for treasuries. The ceiling is defined by the Fed’s penalty rate for direct borrowing in the case of unintended shortages of reserves. The FF rate will never go higher than what the Fed charges for simply crediting the overdrawn reserve account .

      However, if economic conditions create an unanticipated and unpredictable shortages of demand for loans, and hence significant decline in the growth of new deposit balances, then the growth of reserve balances due to the usual actions of liquidity managers will produce excess reserves, and the FF rate will fall toward zero – unless there is an IOR floor.

      • Dan,

        “Scott will no doubt jump in if I have this wrong, but my understanding is that without IOR, since reserve balances are non-interest-bearing assets, banks have no interest in holding excess reserves, and so market demand itself is enough to establish a “natural” floor.”

        Right – but how does that natural floor get established? The Fed sets a zone between 0% and the ceiling rate. How does the FFR settle anywhere between there, if there is no liquidity effect – i.e. the Fed does not adjust reserve balances to move the FFR somewhere between 0% and the ceiling rate (if we consider demand to be virtually inelastic across the maintenance periods, clearly you can’t change number of reserves without forcing the FFR to 0% or the ceiling)? Scott says the Fed announces we will target the FFR to y%, and then they accommodate demand at y%. There is no aspect of changing reserve quantities to move the FFR to y%. But what makes the banks start to trade reserves at y%? Why not any level between 0% and the ceiling? Why don’t banks just say “whatever, Fed wants to do 3%, but I am going to keep trading at 3.5%?” I don’t think the answer is that the Fed will then increase qty of reserves, because Scott is saying a liquidity effect isn’t possible.

        “If banks in the aggregate have more reserves than they need, they will swap those reserves for treasuries.”

        Is this the key? The interbank rank settles around to what can be earned on tsys instead if they swapped with the Fed? But this seems circular. I thought yields on tsys were determined, in large part, by expectations of the future path of FFR. Still, we haven’t determined what definitively sets the FFR in the first place.

        The final two paragraphs of your response are clear and make sense. It’s just that aspect of how and why the FFR settles anywhere between 0% and the ceiling rate that I don’t get, if we rule out the concept that the Fed can adjust the amount of reserves to move the FFR between 0% and the ceiling. Thanks.

        • Wh10, yes as I understand it the floor was sustained on the “old” system by the Fed’s open market policies, and that it is the rates involved in these open market operations, not the quantities, that are important.

          If reserves are non-interest bearing and a bank has more reserves than it need, then it will seek to trade them for something that is interest-bearing. If it can earn more interest in the interbank market than it can earn from securities in the Fed’s repo market, then it will trade the reserves in the interbank market. But once the FF rate drops below what it can earn from repos, then it will stop trading with other banks and instead exchange reserves for securities, both with and without repos. I don’t know how financial folks classify these rates, but it seems to me that there has to be something like an effective “overnight rate” in the market for repo contracts – the amount, given the fact that this debt is constantly being rolled over, that you earn for a very short term repo contract that is about the same as the average duration of an overnight loan. And so, as long as these contracts are sufficiently short term, and the liquidity managers are on top of things, there should no significant liquidity effects. The banks drain their own excess reserves in pursuit of interest from open market trades. And the Fed can adjust the rate on the contracts upward if it sees a need to mop up excess reserves more quickly to support the target rate.

          But it could be that the Fed moved to the channel system with the floor IOR rate because it decided there was too much uncertainty in this kind of procedure, and that banks in the aggregate were frequently overestimating their need for reserves – thus sending the FF rate below the target rate. Or maybe because there just weren’t enough treasuries to trade – so that as longer term debt matures and naturally fills reserve accounts, but growth in deposit balances flags, banks couldn’t unload as many of the excess reserves as they would have liked to. But once you have an IOR floor, then even if banks’ liquidity managers guess wrong, or economic shocks cause sharp changes in the need for reserve balances, funds should never trade below the IOR rate. Instead of being traded back to the Fed in open market operations which have a finite limit based on the amount of government debt issuance, the Fed just offers “securities” themselves by paying interest on the reserves, and the reserves just sit there in reserve accounts earning the interest. The Fed doesn’t have to have tons of treasury debt to sell, because it can effectively provide the same securities in terms of interest-bearing reserves.

          • Looking this over, it seems to me that everywhere I talked about repos, I should have said “reverse repos” – the operations that drain reserves rather than supplying them.

      • Dan K –

        “If banks in the aggregate have more reserves than they need, they will swap those reserves for treasuries.” Are banks actually able/allowed to swap reserves for Treasuries at will? My understanding was that the impetus for this exchange had to come from OMOs conducted by the Fed. If not, the banks could effectively reduce the monetary base at will and would likely have been exchanging reserves for longer-dated Treasuries over the past several years (rather than the reverse). Last comment, I thought banks were limited to exchanging reserves for currency with the Fed.

        Hopefully others can help clear up my possible confusion here.

        • Scott Fullwiler

          buying Tsy’s or borrowing them in the repo mkt doesn’t eliminate the reserve balances. the only way to do that is to buy/borrow the tsy from the Fed or buy them at auction.

        • Yes, see my new reply to wh10 above. I think you are right – the availability of treasury debt via OMO’s is going to be limited by the total amount of debt issued by the treasury. But OMO’s are always going on, right? And Fed holdings of Treasury debt have been going way up:

          http://research.stlouisfed.org/fred2/series/TREAST

          So I assume if the Fed wanted to sell more treasury debt it could.

          By the way, this is why I have been so perplexed at mainstream liberals complaints about Ben Bernanke. Bernanke has been the Obama administration’s best friend. Buy buying up tons of treasury debt, and shifting to an IOR system, the fed has basically been saying to the Treasury, “Don’t worry. We’ll take care of your interest payments. We’ll buy up your debt and hold it to maturity, and then return the interest to you in accordance with law. Then all those interest payments the banks would have received from you by holding your securities will be paid by us in the form of interest on reserves.”

          Effectively, this just shifts the responsibility for interest payments and interest rate management from one branch of the government to another. From the consolidated perspective it doesn’t make much difference. But it gives the administration more breathing room in its political battles with Congressional deficit warriors, since lat year the Fed returned $70 – $80 billion I believe to the treasury.

          The Fed is supposed to be independent. But I don’t think you can read the transcript of Bernanke’s appearance recently at the University of Michigan without concluding he is basically on the administration’s side – at least vs. the radicals in the House.

    • Scott Fullwiler

      Not a ton of research on this, but basically it’s known that the Fed will enter the repo mkt if the rate does trade away from the target, so there’s a “credible threat” there for the rate to be priced there so long as trades are consistent with normal course of business and not some market wide shortage or excess. In fact, prior to 1994, when the Fed did not announce it’s target, it would “signal” the new target via these operations and then stop the operations once the mkt had the rate where it wanted it.

      • In your “General Principles” paper, you wrote:

        “In the case of the Fed, while it might temporarily change the quantity of balances in order to “signal” a new rate to traders or to “nudge” the rate when traders do not move to the new target quickly enough, any changes inconsistent with the given demand for reserve balances—unlike a liquidity effect—are necessarily reversed later in the maintenance period (Krieger 2002, 74). This in fact was the Fed’s operational procedure prior to 1994—”

        Is this what you are referring to? I just find this so confusing, because at first it suggests the Fed is moving the rate via a liquidity effect. But then we recognize they can’t supply more or less than banks need in the maintenance period unless they want the rate at the ceiling or floor. So they reverse the position such that we’re back to the qty of reserves we were originally at (assuming demand doesn’t change). So what maintains the rate they initially ‘nudged’ to?

        • Scott fullwiler

          It’s both/and. Look at how liquidity trap is defined. You can’t permanently adjust balances to change the rate because then you ultimately go to the penalty rate or remuneration rate You can temporarily adjust them for a signal and then take it back. Hope that helps.

          • Scott, thanks so much for your time, but I’m not sure that resolves it for me. I’ll try to use a simple example, with numbers, to explain my confusion.

            Let’s say the banking system needs 100 reserves to settle payments and meet reserve requirements, the Fed has supplied them 100 reserves, the FFR is at 3%, 0% is the floor, and 4% is the ceiling. The Fed wants the FFR at 2.5%, so they increase the quantity of reserves, say to 110, to “nudge” the market from 3% to 2.5%. But as we know, across the maintenance period, banks’ demand for reserves is inelastic at the quantity they need to settle payments and meet reserve requirements, so if the Fed leaves reserves at 110, the FFR is going to fall to 0%. So they drain reserves back to 100. Now what happens to the FFR? Why should it stay at 2.5%, and for that matter, why should it be at any particular level between 0% and 4%? And for that matter, how did the Fed get the FFR to 3% in the first place?

            One answer would seem to be that demand actually isn’t perfectly inelastic, even though it’s close (since banks can’t do anything with reserves besides settle payments and meet RR), such that the Fed *can* manage the quantity of reserves to keep the FFR where they want it. But you seem to say this isn’t true and is not supported by the empirical evidence.

            Alternatively, perhaps the Fed exploits this pre-maintenance period of market elasticity to move the FFR via a liquidity effect, and during the maintenance period, there is an inertia among banks to alter the FFR, even if the Fed changes the qty of reserves back to what the mkt requires. It just doesn’t seem rational or coherent to me, though – that qty could move the FFR during one period but not in the other.

            • (By the way, none of this is to challenge your point about the incoherence of the neoclassical position on this liquidity trap business, which makes sense to me, it’s just to understand this nitty gritty aspect of monetary operations.)

              • No takers :( ? Is it that people think I am hopeless, because this is blindingly obvious? Or that it is complex, and the answer isn’t clear?

                • I have no inside info, but it seems to me that if the Fed does not supply excess reserves somehow, the interbank rate would settle at some point slightly below the Fed’s discount rate. Banks with excess reserves would rather lend them, necessarily at a lower rate else they would have no takers, and banks that need reserves would prefer to borrow interbank than at the discount rate. And there is the thing about the brokers. If most of the loans are brokered, the brokers must be taking a slice as well.

                  • I think this is not as complicated as you imagine.

                    What if the Fed has a target of 3.5%, and a 1/2% tolerance for short-term deviations from it. They could set the discount rate at 4%, and have a standing offer to borrow overnight at 3%. The effect would be the same as offering IOR at 3%. It establishes a band, and they do not allow reserves to trade outside the band.

                    They could set the tolerance at whatever they wanted, maybe as low as 0.01%, so that reserves would trade between 3.49% and 3.51%. They simply supply or drain reserves as needed to maintain their target, within their tolerance.

                    Besides “announcing” the rate they want, they take action in the market to enforce it.

                    Like I said, I have no inside info, but that’s the way I would do it.

                  • That would make sense, but I don’t think they do that. Searching around, I see a daylight overdraft of 50 bps. And I see overnight overdraft rates of at least primary credit rate plus 400 bps. So that’s not really narrowing the band between 0% and the discount window very much at all…

                • wh10

                  I think Scott has answered this (very accurately IMO) across a number of his papers

                  For what its worth, I will remember your question (its a good one) and attempt my own version of an explanation at MR … at some point… could be a while though … priorities and I’ve never written it out before

                  • Hi Guys,

                    Thanks for the responses. JKH, I look forward to your take on the issue. What I get from Scott’s papers is the logic that supplying reserves inconsistent with banks’ demand sends the rate to either the floor or the ceiling. But what I am not getting is what determines the interest rate if you supply reserves consistent with banks’ demand, if we assume, in our simplified model, that banks’ demand for reserves is literally perfectly inelastic at this level, such that there can be no liquidity effect. I don’t see how you can target a rate in the middle ground of the ceiling and floor simply by announcing it, because I don’t see a credible threat. Changing qty of reserves inconsistent with banks’ demand doesn’t appear to be a credible threat to me because doing so would ultimately send the rate to the floor or the ceiling – in other words, it logically shouldn’t work, and therefore is not a credible threat. Moreover, it is known that the CB will reverse the nudge if necessary to make sure only the qty of reserves demanded is supplied, getting us back to where we started and leaving the question of how an interest rate exists at this qty unanswered.

                    If you look at the demand curve, the FFR seems indeterminate at RB*, if we assume the demand curve is perfectly inelastic, and the only tangible, mechanistic tool the Fed has is to change quantity of reserves in the system. In this 0% floor and ceiling rate system we’re assuming, the Fed does not have the ability to supply RB* at different rates – it only supplies the quantity, leaving the rate indeterminate between the floor and ceiling to my eyes. This is different than the IOR system, where the Fed can supply RB* at different rates. Note, however, that Scott isn’t drawing a perfectly inelastic curve – there is some elasticity, which suggests to me that in the real-world, due to various imperfections and uncertainty etc, there is a liquidity effect that is exploited. I recognize it’s hard to pull this out in the literature, but otherwise, I still don’t see how targeting a particular FFR between the ceiling and floor should be possible.

                    (As a side note, let’s assume that the banking system has all the reserves that it needs, and *they’re perfectly allocated to meet every bank’s needs*. Theoretically, in a simple model, shouldn’t there not be any trading after that point, and thus no FFR, assuming this condition continues to hold? So if we imagine the demand curve for reserves, there’s a horizontal line at the ceiling rate approaching RB* with an open circle at RB*, and that’s the end of the story. The FFR is undefined at and after RB* in this simplified model. Now what happens when there are enough reserves but they aren’t yet allocated properly to suit every banks’ needs? According to these graphs and simplified models, the interest rate that the market will settle on seems indeterminate between the ceiling and floor.)

                  • Let me try putting this another way. The way I see it currently, if the interbank market worked completely efficiently, there would definitely be no liquidity effect. But it would also mean that the FFR is undefined at RB* or at least indeterminate between the ceiling and floor. We don’t observe that in the real world though, is what I’m hearing. So reality must be that the market is inefficient, and so either there is a liquidity effect, somehow, and/or the market believes there is a credible threat when there isn’t. Otherwise, based upon what I’ve read in Scott’s papers, and the discussion we’ve all had over the past couple years, I don’t see how the Fed can achieve equilibrium between the ceiling and floor at RB*.

                  • wh10,

                    Looks to me like K has the intuition and the empirical operations right – which shouldn’t be surprising.

                    But I think you already know what he’s described.

                    My intention was to try and describe the elasticity dynamic in more detail at some point.

                    Here’s my quick shot at it now. Hopefully I won’t regret rushing it and that as a result I don’t get it wrong and/or contradict what Scott has described.

                    I think the demand curve is very inelastic – but that in both theory and practice, it can’t be perfectly inelastic and vertical.

                    If the demand curve is perfectly inelastic, that would imply that it would be impossible for the CB to “get traction” by attempting to nudge the rate by open market operations. As you say, the rate would either skyrocket or plummet if that were the case.

                    So what I visualize is a nearly perfectly inelastic demand curve, intersecting what is by construction indeed a perfectly inelastic supply curve. A given supply curve defines the Fed’s reserve setting at a point in time. When the Fed changes the reserve setting through OMO etc., it is shifting its vertical supply curve left or right.

                    The construction of those two curves determines an intersection that is the actual fed funds rate. And you can see that when two nearly vertical curves intersect, the effect of shifting one of them (the vertical supply curve) does indeed get traction on the actual rate versus the target, in a very powerful way. The power of the effect for a given shift in the vertical supply curve is evident in the relatively tiny amount of excess reserves that the Fed needs to work with, compared to the size of the banking and financial systems, in order to achieve a desired traction for moving the actual fed funds rate. The leverage is simply enormous due to the intersection of the nearly inelastic with the perfectly inelastic.

                    So, assume the Fed does indeed achieve convergence of the actual rate to the target rate, through adept control of the location of the supply curve, at a given point in time.

                    Then, assume the Fed drops the target rate.

                    K’s intuitive/empirical explanation then holds, in the following sense of the elasticity perspective:

                    The market will immediately drop the demand curve by the same amount as the rate change. And that will cause the intersection of supply and demand to drop to the new target rate.

                    To see why this is the case, assume the contrary.

                    If the market doesn’t adjust, somebody is going to borrow money at the higher rate – because neither the supply nor the demand curve has yet changed. So the market continues to trade at the old rate. The Fed doesn’t like that. It’s not what it intended. It gets mad. And so it shifts the supply curve right – immediately getting the required traction to move the actual rate down. And after it achieves that, it can reverse the supply curve back to its original position, stabilizing the new intersection point.

                    That hypothetical outcome is the imagined Chuck Norris in action. The guy who borrowed money at the higher rate is then a dead man.

                    But the market isn’t that stupid, and those players in it who are that stupid don’t last very long. The market will kill them. Chuck Norris will just be watching. Instead, the market at large anticipates the threat potential in the form of the looming Chuck Norris. It moves the trading rate immediately to converge to the new target rate, to avoid that. It learns this quickly. And the wisdom of the great majority of the market overwhelms any residual error by a small number of players in this regard. And the move to the new rate level is very, very rapid.

                    BTW, I’ve always hated generalized Chuck Norris. But specific Chuck Norris seems to work here. I guess he’s read STF’s papers.

                  • and note that Scott does say in the post:

                    “The demand for reserve balances is roughly vertical”

                    “roughly”, as in nearly, I imagine

                  • JKH, you’re speaking sense to me.

                    “I think the demand curve is very inelastic – but that in both theory and practice, it can’t be perfectly inelastic and vertical. If the demand curve is perfectly inelastic, that would imply that it would be impossible for the CB to “get traction” by attempting to nudge the rate by open market operations. As you say, the rate would either skyrocket or plummet if that were the case.”

                    Right!!!!!!!!

                    “So what I visualize is a nearly perfectly inelastic demand curve….And you can see that when two nearly vertical curves intersect, the effect of shifting one of them (the vertical supply curve) does indeed get traction on the actual rate versus the target, in a very powerful way… The leverage is simply enormous due to the intersection of the nearly inelastic with the perfectly inelastic.”

                    Which is to say, even though theoretically, in our simple model, we wouldn’t expect one, there is a liquidity effect in the real world. If there is any amount of elasticity, even the slightest, you have a liquidity effect. And if you didn’t, the Fed could never have traction, as you say, or could never be a Chuck Norris. Maybe the near inelasticity is asymmetrical though. If RB RB*, there could be elasticity. So the Fed could keep pushing qty towards that inflection point.

                    “So, assume the Fed does indeed achieve convergence of the actual rate to the target rate, through adept control of the location of the supply curve, at a given point in time.”

                    Liquidity effect!

                    “And after it achieves that, it can reverse the supply curve back to its original position, stabilizing the new intersection point.”

                    The Fed would only reverse the supply curve back if the demand curve dropped in response to the supply curve increasing. I don’t see why the demand curve necessarily would drop in the first instance, but I see why it would happen the next time the Fed announces a rate change. Because, like you say, now everyone knows Chuck Norris indeed exists.

                    But just to reiterate, for Chuck Norris to exist, there has to be a credible threat – such as the liquidity effect. It must be possible for the interbank rate to respond to permanent changes in qty of reserves other than by shooting to the ceiling or floor. And that is what you are arguing here. And that is the conventional, mainstream explanation in undergrad textbooks argues (but they often don’t take it to the next step and explain that as a result, the Fed can just ‘announce’ the rate).

                    But Scott is arguing there CAN’T be a liquidity effect. In this post, he writes “I explained in my 2002 paper why there could logically be no liquidity effect in the federal funds market when the Fed changes its target rate.” In that paper, he’s not just saying that, empirically, we see CBs use ‘the announcement effect’ rather than adjusting the qty of reserves. He is arguing the Fed CANNOT change interest rates by permanently adjusting qty of reserves, even if it wanted to – that a liquidity effect cannot exist. In the General Principles paper, for example, he writes “The mistaken belief that the central bank alters the quantity of balances in circulation in order to change the target rate erroneously implies that banks can “do” something with additional reserve balances when they are supplied, as with the money multiplier model.” This is him not just saying that, in practice, the Fed doesn’t change qty of reserves when they want to change the rate (and instead just announce it). He is saying it’s not possible to change interest rates this way in the real-world, literally. For example, when discussing this concept of “nudging” the market, he says the Fed cannot permanently shift the supply curve, it MUST be shifted back, else you go to floor or ceiling (and my response would be, well then there is no way to target the FFR, as you agreed above): “In the case of the Fed, while it might temporarily change the quantity of balances in order to “signal” a new rate to traders or to “nudge” the rate when traders do not move to the new target quickly enough, any changes inconsistent with the given demand for reserve balances—unlike a liquidity effect—are necessarily reversed later in the maintenance period.” There are many more quotes. You have to go look.

                    That all said, in that paper, he explains it by insisting on the “almost perfectly inelastic demand” for reserves. Now, yes, here, he does draw a slightly slanted demand curve, and in the paper he says ‘virtually inelastic’ or ‘almost perfectly inelastic.’ But there is a difference between that and ‘there could logically be no liquidity effect.’ As soon as you admit any elasticity, it seems to me you must admit a liquidity effect. And if the Fed is to have any traction, a liquidity effect must exist, as you say.

                    The rub is that our simple model of the interbank market would predict no such liquidity effect is possible. And that is partly what Scott is trying to make clear. But I think he is also insisting no such liquidity effect exists in the real-world as well, and that would seem to be over-reaching, because as you note, there must be. The next question, then, is how can this be?

                  • My copy/paste must have been messed up:

                    If RB RB*, there could be elasticity. So the Fed could keep pushing qty towards that inflection point.

                  • Ah, keeps messing up for some reason. I’ll try again:

                    If RB is less than RB*, I would think the rate would have to go to the ceiling – the banks MUST get the reserves from somewhere eventually. However, I could more easily imagine that if RB is greater than R*, there could be elasticity. So the Fed could keep pushing qty towards that inflection point.

                  • wh10,

                    Progress

                    But I now need to make an additional point that was only implicit in what I wrote so far, and need to make explicit – STF is not being inconsistent with what I’ve said, IMO (unless I’m wrong in my own thinking so far, but I don’t think so) – back later, p.m. or a.m.

                  • STF quote I posted below: “The assumption that the Desk can exogenously increase the quantity of Fed balances to move the supply of Fed balances along a downward-sloping demand curve is tied to the fallacious belief that banks can do something with Fed balances beyond meeting reserve requirements and payment needs.”

                    That appears inconsistent with what you said. You say the Desk CAN do that, even though it might not normally because of the Chuck Norris effect. STR is saying it CANNOT, period. But the latter cannot be right if the Fed is to have any power maintaining FFRs between ceiling and floor by just announcing/nudging.

                  • “Right!!!!!!!!”

                    Yes – and I believe Scott would agree, and that he hasn’t said anything to the contrary.

                    “even though theoretically, in our simple model, we wouldn’t expect one, there is a liquidity effect in the real world.”

                    No. The model as illustrated in Scott’s diagram shows a liquidity effect – because the demand curve is still non-vertical. That means a shift in the vertical supply curve changes the rate.

                    Let’s drop the Chuck Norris stuff. I can only stomach that nonsense for one comment in my lifetime.

                    “And that is the conventional, mainstream explanation in undergrad textbooks argues (but they often don’t take it to the next step and explain that as a result, the Fed can just ‘announce’ the rate).”

                    I don’t know if that’s a mainstream element and I really don’t care. That’s the way it works, whether mainstream got this one right or wrong, by accident or by understanding.

                    “But Scott is arguing there CAN’T be a liquidity effect. In this post, he writes “I explained in my 2002 paper why there could logically be no liquidity effect in the federal funds market when the Fed changes its target rate.”

                    This is where you’re wrong in your interpretation.

                    There are two different issues here:

                    a) The liquidity effect as it exists under an unchanged fed funds target rate. There is a liquidity effect. There is no question that the Fed alters the supply of reserves in response to changes in the distribution of reserves in the system and the effect that those shifting distributions have on the effective ease or tightness in the system and therefore on the actual trading rate for fed funds. And that all works according to the graphical logic of a vertical supply curve (that the Fed shifts iteratively in response to changing reserve distribution and reserve demand conditions) and a near vertical demand curve which itself will be shifting due to the effects of shifts in the distribution of reserves among different banks in the system and their behavior in competing for or attempting to shed reserves with respect to their individual positions. There is no way that Scott is going to contradict that or has written anything to the contrary – if he does, I’ll retire from blog commenting.

                    b) A completely different dimension is the presence of a liquidity effect in the case of a change in the target rate. I described that in some detail above. It should be self-explanatory. In that explanation, I separated out the dimension in a) above by assuming that the Fed has at a point in time achieved its target rate with the actual rate. The logic I described then implies that there is no need for the Fed to move its reserve supply curve in conjunction with changing the target rate. Are there “nudges” etc? Of course – some of that will be due to a bit of market imperfection I suppose and some will be due to the fact that changes in respect of a) above are occurring continuously, and can’t be stopped just because b) happens. It’s difficult to untangle the two entirely since both operate concurrently. But they are different forces. The point is that from an analytical and pragmatic operating perspective, the ongoing supply curve changes described in a) are of a completely different breed than any supply curve requirements that might be suggested by b) alone – and there are none in the case of b), once you separate out the effect of a) and any minor market noise otherwise. All those quotes you have from Scott are completely consistent with my description here. E.g. the Fed can’t inject reserves in conjunction with a b) change – UNLESS there is a reserve change required that fundamentally reflects an underling a) type incident – e.g. some unusual reserve distribution pattern that appears by coincidence on the day of a target rate change. Because if the Fed has hit its previous target at the time of the target change, the demand curve drops vertically and any additional reserves put into the system will cause the rate to drop below the NEW target – just follow the geometry I described earlier. That is why even with a b) type “nudge” injection, it must be quickly reversed, other things equal.

                    “And if the Fed is to have any traction, a liquidity effect must exist, as you say.”

                    Yeah – but that’s for a), and not for b).

                    Everything I’ve written should be consistent with what Scott has said.

                    “But I think he is also insisting no such liquidity effect exists in the real-world as well, and that would seem to be over-reaching, because as you note, there must be. The next question, then, is how can this be?”

                    No he’s not saying that.

                    I expect he’ll agree that the a) type effect exists and that a b) type reserve adjustment is generally unnecessary, apart from “nudges” that may be appropriate to weed out market noise.

                    The effects I described in a) and b) are very different – orthogonal in geometric terms – and I think you have confused Scott’s “no liquidity effect” actual interpretation as if he is applying it to both, when I’m pretty certain he isn’t.

                    BTW, I’m using the phrase “liquidity effect” in the sense that you have used it, with my own assumption as to how it might apply to what Scott has been saying. I haven’t had time to review his papers in detail.

                    Again, try separating the a) from the b) (reading closely my description of b) earlier), recognizing that these are separate dimensions in the context of a target rate change and the interpretation of liquidity around that.

                  • “I haven’t had time to review his papers in detail.”

                    meaning:

                    reviewed recently

                    I’ve read them in detail though, at one time or another

                  • JKH, I am pretty sure I am following you, but I think we aren’t connecting because either I am misinterpreting specific passages from STF’s papers / being unfair to him, or you haven’t read those passages as closely as I have / being too fair to him. I understand a) and agree that STF wouldn’t contradict you on it, and implore you not to stop commenting on blogs. I also understand how b) is different. I’ll try to explain below why I think STF, or at least specific passages from STF’s papers, are inconsistent with your view of b).

                    “The model as illustrated in Scott’s diagram shows a liquidity effect – because the demand curve is still non-vertical. That means a shift in the vertical supply curve changes the rate.”

                    I realize he has drawn that. But I contend it is inconsistent with his writing, as I will try to explain below.

                    “The logic I described then implies that there is no need for the Fed to move its reserve supply curve in conjunction with changing the target rate.”

                    This is where I think we aren’t connecting. You are saying there is no need to do this in the real-world given automatic shifts in the demand curve, but that it would be possible (and necessary) to do if the market didn’t move. You’ve stated this very clearly. I get it 100%. But I believe STF is going beyond that, and saying *it’s not possible, period* even if the market didn’t move. For example:

                    “The assumption that the Desk can exogenously increase the quantity of Fed balances to move the supply of Fed balances along a downward-sloping demand curve is tied to the fallacious belief that banks can do something with Fed balances beyond meeting reserve requirements and payment needs.”

                    I take that as meaning that, FUNDAMENTALLY, in the real world, in all types of scenarios, the Desk cannot “exogenously increase the quantity of Fed balances to move the supply of Fed balances along a downward-sloping demand curve.” I do not think that STF is just saying that the Fed has no need to do this, or that they would in fact be erring if they did (assuming there was already a shift in the demand curve), as you write. I think he is saying it’s FUNDAMENTALLY not possible, period – which would be correct if demand is perfectly inelastic, an assumption strongly suggested by the part of the quotation discussing what banks can do with reserves. Moreover, I have never seen him write that the Fed could target a new FFR by shifting qty of reserves if the demand curve didn’t shift – including in that paper (if he did, I would be satisfied). As such, I interpret that passage as being inconsistent with your view.

                    I also realize STF is drawing downward-sloping demand curves here – but I think that is inconsistent with this quotation, or at a minimum, not transparently acknowledged in his papers. While STF may write “almost perfectly inelastic” in the papers, he does not explain the significance of real-world elasticity for establishing the credible threat that removes the need for the Fed to adjust reserves to change the target FFR. Rather, he much more so emphasizes the perfectly inelastic aspect of the ideal simple model, and then uses that to say, IMO, the Fed fundamentally cannot exogenously change qty of reserves to change the target FFR in the real world. There is no intermediate explanation that they could if necessary, but do not need to. This is why I get confused.

                    “All those quotes you have from Scott are completely consistent with my description here.”

                    They are if you assume many things that are not transparently acknowledged, IMO. I haven’t assumed those things, and that is why, in my opinion, the quote above is not consistent with your description.

                    “ E.g. the Fed can’t inject reserves in conjunction with a b) change – UNLESS there is a reserve change required that fundamentally reflects an underling a) type incident”

                    As I contend, that is never transparently acknowledged, if that is what STF believes. Perhaps you intelligently inferred it, or I stupidly missed it. I will completely assume responsibility for the latter though, if the case for the latter is better than the former.

                    “ – e.g. some unusual reserve distribution pattern that appears by coincidence on the day of a target rate change. Because if the Fed has hit its previous target at the time of the target change, the demand curve drops vertically and any additional reserves put into the system will cause the rate to drop below the NEW target – just follow the geometry I described earlier. That is why even with a b) type “nudge” injection, it must be quickly reversed, other things equal.”

                    I completely understand you. I have already said the following, but I will repeat once more. As just one example, the quote I provide above I think is inconsistent with your description. Your description makes clear that if the market didn’t move automatically, the Fed COULD alter the target FFR, somewhere between the floor and ceiling, by changing qty of reserves. I read that STF quote as saying the Fed fundamentally CANNOT do that, ever, both because I think the quote strongly suggests that itself, and also because I have never seen STF acknowledge otherwise.

                    ““And if the Fed is to have any traction, a liquidity effect must exist, as you say.”

                    Yeah – but that’s for a), and not for b).””

                    Let me clarify – a liquidity effect must theoretically be possible for b), under real-world assumptions, assuming the market didn’t move itself. When I say “must exist,” that is what I mean. Not that it happens all the time in practice.

                  • In my above comment, forget this part and skip to the next one (your comment and my reply were on different planes, if you will):

                    ““ E.g. the Fed can’t inject reserves in conjunction with a b) change – UNLESS there is a reserve change required that fundamentally reflects an underling a) type incident”

                    As I contend, that is never transparently acknowledged, if that is what STF believes. Perhaps you intelligently inferred it, or I stupidly missed it. I will completely assume responsibility for the latter though, if the case for the latter is better than the former.”

                  • wh10,

                    thanks for your extended responses

                    you are definitely understanding what I’m saying; sorry for suggesting/implying otherwise

                    “The assumption that the Desk can exogenously increase the quantity of Fed balances to move the supply of Fed balances along a downward-sloping demand curve is tied to the fallacious belief that banks can do something with Fed balances beyond meeting reserve requirements and payment needs.”

                    I hadn’t planned on getting into the detail at this time, but let me review the 2002 paper later today and I’ll return with thoughts

                  • wh10

                    OK – I’ve just read the 2002 paper very closely – from page 1 to page 29.

                    Pages 25 -29 are the key to your question. I will finish the rest of the paper out to the end at page 36 later today.

                    My reading of it is that Scott is saying exactly what I have said – although obviously I can’t be certain that he would say the same thing as vice versa – but I totally agree with what I think his intended meaning is, and that’s the meaning I intended to convey in what I wrote.

                    I will come back later today and attempt to summarize a reconciliation between the two. The precise definition of what I understand to be his meaning of the “liquidity effect” may be an issue here – and I alluded to that earlier. But that is irrelevant in terms of what I think is a near perfect equivalence between the description he wrote and what I wrote.

                    My – what a brilliant paper though.

                  • “If the market doesn’t adjust, somebody is going to borrow money at the higher rate – because neither the supply nor the demand curve has yet changed. So the market continues to trade at the old rate. The Fed doesn’t like that. It’s not what it intended. It gets mad. And so it shifts the supply curve right – immediately getting the required traction to move the actual rate down. And after it achieves that, it can reverse the supply curve back to its original position, stabilizing the new intersection point.”

                    The Fed would only reverse the supply curve back if the market dropped the demand curve (because if the market did that, the rate would change from the new achieved target). But we’re not assuming that in this case. We’re assuming the market doesn’t drop the demand curve. Therefore Fed shifts supply, and you’re at your new point. The NEXT time the Fed announces a new rate, THEN we’d expect the market to drop automatically, and the Fed wouldn’t shift the supply curve.

                • wh10,

                  It’s not *that* complex. The CB’s job is easier if there is a tight corridor like in Canada, but it’s not a big deal. Let’s say the target is 2.5% with IOR at 2.25 and the offer rate at 2.75%. So the rate *can* settle anywhere between the floor and the ceiling. But every night the Bank of Canada somehow makes it settle within a bp or so of the target.

                  Imagine that intraday the market rate goes to 2.6%. I will borrow from you at that rate only if I think that otherwise there is a significant chance of me having to borrow at a higher rate if I don’t borrow now. If I know that the CB disciplines the market hard whenever the marke rate deviates from target, I’m not going to borrow. I’m just going to wait for the rate to drop by itself or the CB to intervene. If they always intervene I know I’ll be able to borrow below 2.6% so I won’t borrow now and therefore I won’t be putting any upward pressure on the market rate. This is where Nick Rowe’s Chuck Norris analogy works really well. If you know Chuck *will* intervene to restore the market back to target you will never put on a position that loses money when the market settles on target. So the market will never be pushed away from target.

                  • But K, you’re describing a liquidity effect, or the demand curve for reserves not being perfectly inelastic. In other words, the CB “disciplining” the market is it changing the qty of reserves to affect the price. This is the old school, conventional story of how the CB targets interest rates. I’m asking you to justify how that could exist, since Scott is saying, logically, no such liquidity effect should exist across the maintenance period, since any qty of reserves above or below RB* should make the interbank rate go to the ceiling or floor (and if that’s the case, no such Chuck Norris effect should exist). It makes sense there shouldn’t be a liquidity effect if the market works as perfectly as a simple classroom model. But it seems it does not work that way in real life – it seems it works as you describe it.

                    Moreover, in Canada the corridor is only 50 bps wide. In the U.S., before the crisis, it was often at least 400 bps. More evidence that something, maybe a liquidity effect, must be at play.

                  • “since Scott is saying, logically, no such liquidity effect should exist across the maintenance period, since any qty of reserves above or below RB* should make the interbank rate go to the ceiling or floor”

                    wh10 – where are you getting this from?

                    and what does any of this have to do with the operation of a maintenance period specifically?

                    And I now notice clearly that Scott actually draws the demand curve non-vertical in the graph above – just as I described in my comment above

                    there is definitely a sensitivity of the rate to the quantity of reserves – in the real life pre-QE regime and in the analytical framework above

                    to assume otherwise, there would never have been any change in the excess reserve setting within periods under the same fed funds target – which is patently false

                    I don’t see where the STF description contradicts this – anywhere

                  • JKH, I swear I’ve read it in all his papers. Read my response to you above for some examples.

                    In the 2002 paper:

                    “To summarize, it is meaningless to talk of the Desk as supplying money or reserves
                    in order to reduce rates through a liquidity effect. Because of the inelastic demand for
                    Fed balances across the maintenance period, an increase or decrease in Fed balances
                    inconsistent with the period-average need would simply lead to wide swings in the funds
                    rate. Changes in the target are carried out through announcement or signaling (or
                    changing the upper and lower limits in other countries). The temporary operations that
                    generate both “signals” and “nudges” are offset if they are inconsistent with the existing
                    period-average need. The assumption that the Desk can exogenously increase the quantity
                    of Fed balances to move the supply of Fed balances along a downward-sloping
                    demand curve is tied to the fallacious belief that banks can do something with Fed balances
                    beyond meeting reserve requirements and payment needs. This point is related to
                    the Fed’s ability to control the monetary base, to which we now turn.”

                    “Through application of the different event sequences in the Fed’s daily tactics, this section
                    defines a liquidity effect as a change to Fed balances sustained throughout a maintenance
                    period to generate a change in the federal funds rate. Shorter-term, temporary
                    operations that “nudge” the rate closer to the targeted rate on a given day or “signal” a
                    change in the FOMC’s target to federal funds market participants are unrelated to a
                    liquidity effect.”

                    Note the word “temporary.” Like I said above, he argues they must be reversed to be consistent with RB*. And it’s not just that he is saying this is not what happens in practice. It’s that it is not possible in the real-world. But if that is the case, we’re back to square 1, as I’ve said repeatedly, and you agreed.

                  • wh10

                    see my 7:56 p.m.

                • Jose Guilherme

                  For the Canadian case (where the operating band is 50 bps wide), let us hear from Marc Lavoie, who wrote the following:

                  “…the realized overnight rate will turn out to be very close to the middle of the operating band…there are essentially two reasons for this. First, the banks know that the Bank of Canada wishes the overnight rate to be around (say) 4% and that it will intervene if the actual rate keeps drifting away from the target rate. Second, competition should bring the overnight rate to near the middle of the operating band because at that point, the opportunity gain of the lenders and that of the borrowers are precisely the same, being equal 25 basis points for both groups of participants in the overnight market”.

                  Couldn’t be clearer.

                  • Jose- please see above.

                    This part about intervening is not answering my question. It’s avoiding the question, or just an incomplete answer. My whole beef here is this issue over the liquidity effect. When he says “intervene,” he must mean the CB changing qty of reserves to affect the interest rate. This is the liquidity effect. Scott is saying the liquidity effect shouldn’t exist across the maintenance period. Scott has acknowledged the CB can ‘nudge,’ as Marc might be alluding to, but he has also said this position has to be reversed to stay consistent with RB*, else the rate goes to the ceiling or floor. But it that case you’re back to square 1! You haven’t explained how a rate can be targeted between floor and ceiling.

                    His point of it drifting to the middle is really interesting, maybe that starts to get at it. But it also suggests an equilibrium that I don’t think should exist in the classroom model where you assume a perfectly inelastic demand curve and therefore no liquidity effect.

                    As such, this doesn’t really answer my question. It might be telling me how the real-world works, but it doesn’t explain why, especially when one would theoretically not expect a liquidity effect.

                • wh10,

                  I think you may be failing to take into account to the fact that because reserve imbalances are caused by internal payments, there is always someone with an excess reserve position for every bank with a deficient one.

                  Imagine that banks have their desired quantity of reserves (zero in the Canadian case). Bank A makes a $1Bn payment to bank B associated with a deposit transfer. But bank A didn’t have those reserves (or wants to get back to their desired reserve balance) and bank B now has $1Bn of extra reserves. If bank B refuses to lend that money to bank A then bank A will have to borrow from the CB at the ceiling rate and bank B will be stuck earning IOR on $1Bn. So they *both* lose 25 bps relative to if they had agreed to a direct loan at the target rate. Non-clearing is not a zero sum game. As Mark Lavoie says, the savings of internal clearing are equalized for lender and borrower *at* the target rate.

                  PS. For some reason, some comments don’t have a “reply” link, so I keep replying to one of your earlier comments. What am I doing wrong?

                  • You aren’t doing anything wrong. It’s just what happens to the comment system after a certain threshold of replies is passed.

  8. This is too easy to get wrong, for someone who does economics and finance for a living, like Bernanke, for instance. That means the only point of QE1 was to raise the price of the assets that bondholders would flee to when their bonds were removed from the market. And the only point of QE3 is to bail out the holders of MBS that are still not marked to market.

    Bernanke must be telling this to Obama, and the corollary that “monetary” policy isn’t and can’t help end the recession, so “fiscal” policy is the only other alternative. I recall one faint reference to this in his many Congressional testimonies. Why isn’t he pounding the table? Or if he is, in private, why doesn’t anyone believe him? Or do they believe him, and they don’t care?

  9. Scott:

    Brilliant, again.

  10. Scott,

    I don’t get this point:

    “Krugman in fact does this when he argues that the Fed could expand the monetary base without sterilizing and also without paying interest on reserves (his option 3); he then claims this would be inflationary while also arguing it would not be inflationary if instead interest were paid on reserves. But from Figures 2 and 3, again, either this leaves the federal funds rate at zero (no sterilization) or at the target rate (interest on reserve balances), and neither of these are inflationary in his own view”.

    In Krugman’s view a zero interest rate on reserves could be inflationary once we are out of a “liquidity trap”.

    Also, Krugman thinks the Fed would pay IOR with money that it “gets from somewhere else”, i.e. from the interest it ‘earns’ on government bonds, so the net effect would be no different to normal government borrowing. You seem to be arguing that the Fed would instead pay IOR with money created ‘ex nihilo’. That’s a major difference which Krugman apparently hasn’t considered. No doubt he would consider that to be inflationary too.

    • Scott Fullwiler

      “In Krugman’s view a zero interest rate on reserves could be inflationary once we are out of a “liquidity trap”.”

      Correct, but he defines the liquidity trap as when bills and currency are perfect substitutes, which is figure 2. so, if you let the fed funds rate go to the zero bound, then you are in a liquidity trap. he doesn’t realize that you can’t increase the base exogenously without ending up in the “liquidity trap” as he’s defined it.

      • Scott Fullwiler

        Correction, as I misread you. My interpretation is he’s saying being at the zero bound IS a liquidity trap. Below I’ve copied what Krugman’s said. My points is that operationally you can only have the zero bound or IOR if you try to increase the monetary base. He somehow believes you can increase the monetary base without having one of those two.

        http://krugman.blogs.nytimes.com/2010/07/14/nobody-understands-the-liquidity-trap-wonkish/
        The Fed can print money, if you increase the supply of something its price will fall, end of story.

        But as I tried to point out a long time ago, this simple story breaks down when short-term interest rates are near zero.

        Here’s one way to think about it: when the Fed conducts an open-market operation, buying short-term debt with newly printed money, this normally affects the short rate because bonds and money are imperfect substitutes: money yields less, but has the advantage of being something you can use directly to make payments, that is, it’s more liquid.

        But when you have bought so much debt and created so much money that rates are near zero, the public is saturated with liquidity; from that point on, they’re holding money simply as a store of value, which makes it no different from bonds — and hence a perfect substitute for bonds. And at that point further open-market operations do nothing — they just swap one zero-interest asset for another, with no effect on anything.

        http://krugman.blogs.nytimes.com/2010/03/17/how-much-of-the-world-is-in-a-liquidity-trap/

        But what’s the definition of a liquidity trap? How much of the world is in one? There’s a lot of confusion on that point; here’s how I see it.

        In my analysis, you’re in a liquidity trap when conventional open-market operations — purchases of short-term government debt by the central bank — have lost traction, because short-term rates are close to zero.

        • Scott, isn’t the old idea of a liquidity trap something more than this? As I understand it the traditional idea is supposed to be that the liquidity trap occurs when there is still a difference between money and short-term debt – they aren’t near-perfect substitutes – but because there is such an intense need for liquidity, the demand for liquidity becomes price-insensitive, and potential lenders are unwilling to exchange liquidity for loan obligations, no matter what the rate of interest on those obligations.

        • Hi Scott

          Nice post!

          To what extent does the liquidity trap rely on the concept of a ‘natural rate’ of interest, as opposed to the ‘nominal rate’ set by the Fed in your opinion?

          If I read Krugman correctly, it is the exit from the ‘floor system’ back to the ‘normal system’ that constitutes the greatest inflation risk once the natural rate has risen above zero / above the nominal rate.

          In other words, when r(nominal) > r(natural) the Fed can sell off securities until r(nominal) = r(natural) after which you get r(nominal) < r(natural), which, according to classical theory, is inflationary. The natural rate describes the condition under which the volume of credit grows at the same rate as GDP (a hidden stable velocity assumption, I guess) and thus inflates neither asset nor consumer prices.

          Krugman conclusion should be, that we either exit now or be stuck with IOR at least until we hit the next lower bound or else tolerate higher inflation. I'm not sure it is, though.

          Does that make sense?

          • Scott Fullwiler

            Yes, I see your point. I’ve struggled interpreting PK, too, a bit on this point. I’ve looked for somewhere that he says something about the liquidity trap (which he defines as currency and tbills perfect substitutes, and so overnight rate=0; this isn’t at all Keynes’s liquidity trap, btw, as Jan Kregel pointed out in critique of PK’s original paper on this) relative to his view of a “natural rate.” I didn’t see anything explicit (have you seen anything?), but I think that’s where you’re going, too (?).

            If one wants to argue that there’s no liquidity trap if the natural rate rises above 0% even if currency and tbills remain substitutes, then my counter would be that even from within the neoclassical paradigm the Fed can simply raise the floor to whatever one thinks the natural rate is and this again from their own view cuts off the inflationary effect of an excess qty of the monetary base. And in that case, again, currency still isn’t a hot potato because excess balances can be costlessly converted at financial intermediaries, taking the currency out of circulation as I described.

  11. “The basic point neoclassicals want to make here is that they believe the Fed can exogenously control the quantity of currency (i.e., “dead presidents”), and thereby generate a “hot potato effect” as the “excess cash balances” created by this form of QE are spent repeatedly by individuals trying to rid themselves of so much cash and hold computers, toothpaste, toasters, new homes, couches, and so forth instead.”

    Market monetarism guilty as charged

    Krugman seems slightly ambiguous at times

    Doesn’t matter

    There’s just no excuse for any of them not clearly understanding this

    Neoclassical economics is still running a short position in accounting … apart from accidentally stumbling on small parts of what’s required from time to time

    • Scott Fullwiler

      Nailed it! As usual. Good to see you here. Sorry I was kind of a poop several months ago–I’m over it now. Thanks for your support of my work–it means a lot.

  12. Scott,

    Somewhat unrelated I guess, but how much leverage do primary dealers have in tsy auctions? If the Fed decides to hold funds rate at or near zero indefinitely, is there any likelihood that the dealers would refuse to make reasonable bids? I guess they would always rather have tsy’s that yield more than the .25% that excess reserves pay?

  13. Scott,

    Great post, spot on! I’ve been advancing this argument on Sumner’s blog over the years and quite intensely over the last few weeks, but to little avail. What *has* come up, though, is that Sumner, for example, isn’t really talking about our current monetary system when he says that the Fed should just control the size of the base (which I have been saying is operationally impossible). Sumner doesn’t really understand how the current operations work, but, it turns out, he doesn’t really care. If you push him a bit, it turns out that he is talking about a system in which there is no IOR *or* discount window and the Fed basically ignores the banking system, and as a result, the banks keep vastly (hundreds of billions or trillions) higher excess reserves.

    It’s quite absurd for him just to say that “that’s how it works” when it obviously doesn’t. But, there *is* an operating framework in which demand for reserves is probably quite flat. It’s a system dominated by no central guarantor of the clearing system and wide spread distrust between banks and much higher equilibrium interbank rates. It probably is not a very stable system because of the well known tendency for leverage bubbles followed by crashes accompanied by unlimited demand for liquidity. But, in theory at least, there exists a universe in which the economy is controlled via the size of the base (it still happens via indirect control over the real rate (the IS curve), but that’s another matter). But he really should start focusing on his desired institutional reforms and stop pretending that the system already works like he says it does. It clearly doesn’t.

    On the coin that inspired this whole debate: It’s interesting to note that if they minted a $1Tn coin, the Fed would not be able to reduce its balance sheet to the $800Bn or so, which was the base (required reserves plus equilibrium demand for currency and non-interest bearing excess reserves) at the end of 2007. This means that they would not be able to raise the short rate to 4% *without* paying IOER. So *if* they began to see inflationary pressures they would face the choice between keeping rates at zero (and letting the administration control inflation via fiscal policy) and paying IOER and driving themselves into negative equity. The latter would not be a major problem if it wasn’t for the fact that congress might not be predisposed to bailing them out of an inflationary problem of Obama’s making. Since I don’t think the bankers who run the Fed would ever risk the Fed becoming “insolvent” I suspect they would leave IOR at zero, and let the White House run monetary policy via spending (probably targeting a zero equilibrium short rate). Is that not, as it happens, the preferred MMT policy outcome?

    • My question about the upshot of the new regime is this K: With large excess reserves, what is the point of the FF rate target? How does it function to regulate inflation? For a couple of decades, this has been the key operational target. But the idea is supposed to be that raising the FF rate is anti-inflationary because it raises the commercial rates banks have to charge as they are expanding their lending in order to maintain a profitable spread between the costs of additional reserves and expected returns on the loans.

      But if the banks are already swimming in excess reserves, then raising the cost of getting more reserves doesn’t matter all that much right? If a bank doesn’t need to acquire more reserves to expand its lending, then the price it would have to pay for acquiring them is irrelevant.

      In debates about this before, the question has come up as to whether the Fed already has all the authorization it needs to impose a negative rate of IOR to drain reserves directly by in effect taxing them away. There seemed to be some legal disagreement about this.

      • Shining Raven

        I am neither K nor an expert, but perhaps the bank simply prices things at the margin, independently of the actual (excess) reserve position? I.e, since the Fed funds rate is what it would need to pay to acquire additional liquidity, this is the price that it assumes for internal calculations, and this is the price that goes into the profitability calculation for loans? And then the interest rates for loans are set accordingly, at a fixed spread above the FFR?
        And if no additional liquidity needs to be acquired, so much the better, higher profits?

        Or interest rates are still simply set as they ever were before, at a spread to FFR, out of institutional inertia….

        • Dan:”If a bank doesn’t need to acquire more reserves to expand its lending, then the price it would have to pay for acquiring them is irrelevant.”

          Shining Raven: “And if no additional liquidity needs to be acquired, so much the better, higher profits?”

          To be more clear than in my previous answer…

          When a bank lends, it creates new deposits which are quickly scattered throughout the banking system. When a deposit moves from bank A to bank B, an equal payment of reserves must go with it. If bank A has excess reserves on hand they simply transfer those. The “cost” of doing so is that they no longer earn IOR on those funds. If bank A *doesn’t* have excess reserves (banking system circa 2007), they have to borrow them in the interbank market. (In a world with no excess reserves, the borrowed reserves will effectively come from bank B, because they are the ones who would otherwise end up with an exactly matching reserve surplus.) The cost of doing so is the FF rate. So in one case the marginal cost is IOR, and in the other it’s FF. It doesn’t make a big difference: either way the Fed controls commercial banks’ base marginal cost of funding their loans. I don’t think it has any very direct impact on profits.

          Shining Raven: “somebody who has researched monetary policy all his professional life can believe that “money” is essentially pictures of dead presidents.”

          Yeah, I think I’ve given up commenting at The Money Illusion. It’s all circles and rabbit holes and turtles all the way down. And in the end you are back exactly where you started.

          [Scott: Are all commenters moderated here, or is it just me? It’s a bit hard to follow the discussion, but maybe there has been a lot of abuse?]

          • Yeah, I think I’ve given up commenting at The Money Illusion. It’s all circles and rabbit holes and turtles all the way down. And in the end you are back exactly where you started.

            That’s how I feel, K. I used to ask them about the transmission mechanisms and how they are supposed to function. But when one alleged transmission mechanism was shown to be faulty, they would move on to some other one, and then double back to one of the earlier ones. It just seemed like a matter of pure faith to them that there must be some transmission mechanism that worked. I got tired of playing whack-a-mole, and don’t even read the blog any longer.

            Nick’s blog is better, because he is a much more open-minded guy with an inquiring mindset. But he also falls back on that simultaneous causation business and lampooning “concrete steppes” when the initial stabs at transmission mechanisms don’t seem plausible.

          • K, regarding the opportunity cost of lending in an IOR regime, with or without excess reserves, it seems to me that while that makes sense from the point of view of the individual bank, it doesn’t help much from the broader perspective of aggregate monetary policy. If the Fed changes policy that somehow successfully induces a higher or lower rate rate of lending, that will change the volume and rate at which interbank reserve transactions occur, it doesn’t change the total amount of reserve balances or the aggregate costs to the whole system. Banks might incur an additional cost per loan associated with the quantity of reserves the new demand deposit balance ends up moving out of their bank. But they are compensated by higher earnings on the reserve balances that are flowing into their bank by their competitors’ corresponding increase in loans. So their earnings per period from reserve balances haven’t changed.

      • Scott Fullwiler

        Dan . . see “paying interest on reserve balances” that I link to in the paper. I deal directly with that.

    • Shining Raven

      K, thanks for keeping at it and for making all of this so clear. I always find your comments extremely helpful.

      I am not sure that Scott Sumner perhaps simply does not understand that things do not work the way he imagines.
      He always wants to assume a world without banks, or without deposits, or whatnot, and says that he is only interested in monetary policy, not in loans, when anybody brings up credit. For the life of me, I cannot understand how somebody who has researched monetary policy all his professional life can believe that “money” is essentially pictures of dead presidents.

      I have posted this before at Scott’s blog, but it really seems apposite:

      “It appears that with RPD [reserve position doctrine], academic economists developed theories detached from reality, without resenting or even admitting this detachment. Economic variables of very different nature were mixed up and precision in the use of the different concepts (e.g. operational versus intermediate targets, short-term vs. long-term interest rates, reserve market quantities vs. monetary aggregates, reserve market shocks vs. shocks in the money demand, etc.) was often too low to allow obtaining applicable results. The dynamics of academic research and the underlying incentive mechanisms seem to have failed to ensure pressure on academics to ensure that models of central bank operations were sufficiently in line with the reality of these operations.”

      From a paper by Ulrich Bindseil from the ECB: http://www.ecb.int/pub/pdf/scpwps/ecbwp372.pdf

  14. Thanks for this Scott. I remember Dr. Wray scratching the surface of this – floor vs ceiling explanation – during the course of the MMT Primer, although he did not touch on IOR there. With respect to Open Market operations, however, I decided to seek out some more information from the Fed site itself and found this:
    http://www.federalreserve.gov/pubs/bulletin/1997/199711lead.pdf
    Very interested in annotation (2) from a functional perspective, particularly this bit:
    “Depository institutions can arrange transactions
    directly between themselves, or for large transactions they can use a
    federal funds broker. Typically, the term ‘‘federal funds rate’’ refers to
    the rate at which the most creditworthy institutions borrow and lend
    balances in the brokered market.”

    Two questions for you:
    1. Can one then assume that most Open Market transactions are indeed brokered?
    2. Can the Fed ever mandate that all transactions be brokered as another means of keeping trading within target?
    Much obliged for any light you can shed here.

  15. Why do neoclassicals ascribe so much importance to whether reserves and treasuries are perfect substitutes? It’s like a phase transition to them, one basis point difference and boom! banks behave completely differently. I don’t get this. Does it make sense within their paradigm?

  16. The elimination of Reg Q ceilings induced dis-intermediation within just the non-banks during the credit crisis of 1966. The Fed’s policy response (countervailing intervention), using the Fed Funds “Bracket Racket” then subsequently caused stagflation.

    Likewise, the payment of interest on excess reserve balances (the establishment of a floor system using IOeR) will inevitably cause a depression.

    I would trust the politicians with MMT (& eliminate the compounding interest expense on the Federal debt) before playing Russian roulette (eliminating reserve requirements altogether).

    • Shining Raven

      I don’t understand this. Paying interest on reserves is only a technical means to maintain a positive FFR. It is no more deflationary than maintaining the same FFR by draining excess reserves in OMOs.

      So I *think* that it makes sense to say that to high an FFR target can lead to a depression, but then your beef is with a wrong (too high) FFR target. Interest on reserves as such has nothing to do with it. The same outcome should result as from the same FFR in the world without FFR and without excess reserves.

    • Shining Raven

      Last sentence in my previous post should read “….without *IOR* and excess reserves.”

  17. Dan,

    If there are lots of excess reserves then FF will be IOR, but as you said there won’t be much interbank lending (like right now). Banks just make payments by transferring reserves they already have on hand. When banks make a loan, they assume that the funds will be dispersed to other banks in the system and they will therefore have to transfer an equal amount of reserves. The profit is therefore relative to what they would have made just holding those reserves. So (on a floating rate loan) they charge IOR plus credit spread plus cost plus profit. So loan rates are relative to IOR=FF.

    So the relevant risk free rate relative to which

  18. Dan,

    The real issue in the liquidity trap is that the real rate can’t be lowered because the nominal rate is floored at zero. This happens because of the existence of currency, whether there is IOR or not.

    • Scott Fullwiler

      Have you seen PK explicitly state it this way? I’ve been looking but haven’t found it. Might have just looked too quickly.

      • Scott,

        No, that was my view.

        He used to say that a liquidity trap is whenever the policy rate is at zero. That’s obviously wrong since the policy rate could be too low. Krugman ’98 suggests that he must know this very well though. 

        But he recently said this:

        “On the other hand, if the government borrows from the Fed and the Fed does not raise IOER, the government will be printing money – literally – to cover the gap, expanding the quantity of currency. And this will be inflationary unless you’re in a liquidity trap.”

        That is strictly speaking correct. But it sounds like he is thinking the wrong thing (price level increases with money quantity unless money and bonds are equivalent) or he wouldn’t explain it like that. Instead he’d say that if you increase the base such that excess reserves exceed a few billion dollars the FF rate will immediately drop to zero (if IOR=0). A zero FF rate is inflationary unless the nominal natural rate is below zero. I’m almost certain that’s what Woodford would say. 

        BTW, I think it’s important to distinguish between the neoclassicals and the New Keynesians in this discussion. Woodford, for example, most definitely understands how the banking system and the liquidity trap works. Krugman is a bit of a cross breed, his intuition solidly founded in macro 101, but way too smart, for the most part, to say things that are unequivocally wrong. The terms of the debate could be much clearer, though, and that definitely includes Waldman and Kaminska. Your posts are definitely an oasis of clarity. The debate here is also good and super civilized (maybe that’s the moderation).

        • Actually that Krugman statement is not “strictly speaking correct”. It would have been if he’d said “base” instead of “currency”. But he didn’t. So maybe he really is confused.

      • Bill Mitchell has many posts in which he complains about Krugman’s and DeLong’s reliance on liquidity trap theorizing, and argues that the MMT-related reasons for the limits on the central banks’ impact on aggregate demand do not depend on a liquidity trap for their existence. Here’s one.

        http://bilbo.economicoutlook.net/blog/?p=15168

    • K, I don’t think “liquidity trap” and “zero bound” mean the same thing – although maybe Krugman uses them interchangeably. Fischer argued that anything that could be stored cost free would never be loaned at negative interest, and I think that is the theoretical foundation for the very existence of a zero bound. A liquidity trap is supposed to be one specific phenomenon (or set of phenomena) that explains why interest rates might fall to the zero bound.

    • Eric Tymoigne

      I would say that the liquidity trap is a situation where nominal long-term rates can no longer be pushed downward because the state of expectation about future nominal long-term rates is such that nobody is willing to buy bonds below the current rate: expectation of future capital losses so that total return is negative. Additionally, the “widows” type of buyers (buy and hold, don’t care about capital gain/loss) are unwilling to buy at a lower rate.

  19. Pingback: interfluidity » A confederacy of dorks

  20. Dan,

    “anything that could be stored cost free would never be loaned at negative interest”

    Right: nobody will lend at negative nominal rates if they can store nominal value at zero nominal cost *by holding currency*. Negative nominal rates are only prevented because the monetary authority supplies currency in whatever quantity is demanded. If there is no currency, there is simply no
    nominal-cost-free store of nominal value.

    Lets say the CB sets IOR at -5% and supplies a large quantity of base. Banks will just convert their reserves to currency that they will store in vaults. A corridor system won’t help either: if they lower the offer rate (discount rate) to, say, -4% banks will borrow currency in unlimited quantities and make big arbitrage profits. If you want negative rates, you can do so easily by limiting the quantity of currency or cancelling it altogether.

    But I’m quite sure we discussed this elsewhere in great detail!

    “A liquidity trap is supposed to be one specific phenomenon (or set of phenomena) that explains why interest rates might fall to the zero bound.”

    I don’t agree. The liquidity trap is a story about how, when the nominal natural rate drops below zero, disinflation will occur setting off a spiral of rising real rates and accelerating disinflation. It’s fundamentally about how the nominal ZLB can cause a very important price (the real rate) to be very wrong for a long time resulting in a big output gap. I don’t see a good story about how it makes rates *go* to zero.

    ” If the Fed changes policy that somehow successfully induces a higher or lower rate rate of lending, that will change the volume and rate at which interbank reserve transactions occur, it doesn’t change the total amount of reserve balances or the aggregate costs to the whole system.”

    It’s not about affecting “costs to the system”. It is quite literally about affecting the *real* rate of interest that lenders *inside* the system charge to borrowers also *inside* the system. Like you, monetarists also focus on seignorage costs, i.e. opportunity costs of holding cash, equal to the nominal rate on the outstanding quantity of currency (plus interest-free excess reserves). I.e. outside costs to holding medium of exchange. But imagine a world of zero currency. The CB *still* controls the short rate and therefore the interest rate used in private loans. If you raise that rate you will still induce a reduction in consumption and investment. The real rate is hugely important in the economy of the typical household (e.g. mortgage interest), and dwarfs opportunity costs of holding currency.

    • K,

      Keynes pooh-poohed the natural rate concept in the General Theory, so I don’t think that can be the idea behind his concept of the liquidity trap. My understanding is that he was arguing that for any give set of economic circumstances the speculative motive for liquidity will have a certain strength, and that there will be an interest rate in those circumstances below which nobody will be willing to part with liquidity for debt. Ordinarily, we would expect that as the rate of interest falls, the demand for borrowings would increase. But the liquidity trap hypothesis is that below a certain interest rate the demand ceases to change – so the demand curve for borrowings goes vertical below that price line.

      The zero bound is a phenomenon that occurs whenever there is cost-free storage of money, and has no logical connection with liquidity traps. Even in an economy without a liquidity trap there is never an incentive to pay someone to borrow your money if you can hold it at no cost. You might be inclined to imagine that supply and demand curves always have their usual curvature and are defined for all values on the a axis, and so intersect somewhere below the x-axis if they don’t intersect above it, but it seems to me a more natural picture is that with cost free storage the curves are truncated and simply come to an end at the x-axis. So is there is no positive equilibrium point there is no equilibrium at all.

      The reason the presence of a liquidity trap could explain rates falling to the zero bound is that we would ordinarily expect that since suppliers and customers have to discover their equilibrium in the bargaining process, we would expect the supplier to decrease the price and the customer to increase bids until they settle at an equilibrium. But if there is a liquidity trap, you will get to a point where the suppliers keeps reducing with no movement from the customer – until they are at or near the zero bound and the supplier’s price stops moving.

      On the negative interest rate stuff from before, I have a new argument that the effect of a negative central bank corridor in a currency-free economy would necessarily be a continuous drain of member bank net financial assets and a reduction of the member banking sector to insolvency.

      • During the depths of the economic crisis – late 2008-early 2009, short-term Treasuries actually turned negative. That is, they actually had below zero yields. This was for a short time, and was dismissed as a technical glitch. Less easy to dismiss was/is the near zero yields ever since then. Why would anyone, especially professional investors, bother to own Treasuries that yield near zero interest, certainly well below inflation? Of course, there is the absolute guarantee of repayment of principal, but an escrow account, or even a safe bank CD would do the same thing. No, something else was/is going on. Despite repeated warnings of a Treasury bubble http://blogs.wsj.com/marketbeat/2013/01/18/gluskins-rosenberg-bear-market-for-treasurys-dont-bet-on-it/ the yield even on the 10-year bond, is below 2%!
        Well, if the Fed is the buyer of last resort, and you are a trader able to leverage huge blocks of bonds using derivatives, you can make a great deal of money by using OPM (Other People’s Money). OPM is the best form of money, especially in an unregulated market where there are no limits. With OPM, it’s possible to “invest” 2, 5, or even 10 times more money than you actually have. Lehman Brothers, prior to its collapse, was estimated to have a 40:1 ratio of leverage to actual assets. Today’s derivatives market, admittedly mostly in currencies, not Treasury bonds, is calculated to be $1.2 quadrillion http://www.dailyfinance.com/2010/06/09/risk-quadrillion-derivatives-market-gdp/. I don’t know if anyone knows how much of the derivatives market is in Treasuries, but if you had a guaranteed market for something, and you were a TBTF bank, how high could YOU leverage an investment in that asset?

      • Actually, I don’t feel good about my last paragraph right now. An argument that seemed sound by the midnight oil doesn’t look so good to me at 9:15 am. But no problems yet with the first two paragraphs!

      • (sigh) Where I said, “demand for borrowings” above, I should have said “demand for interest bearing assets.”

      • “Even in an economy without a liquidity trap there is never an incentive to pay someone to borrow your money if you can hold it at no cost.”

        And yet banks offer, and people use, checking accounts that pay no interest and charge monthly fees. Perhaps the demand curve for debt is not quite vertical, even at negative interest rates.

  21. Scott,

    “If one wants to argue that there’s no liquidity trap if the natural rate rises above 0% even if currency and tbills remain substitutes, then my counter would be that even from within the neoclassical paradigm the Fed can simply raise the floor to whatever one thinks the natural rate is and this again from their own view cuts off the inflationary effect of an excess qty of the monetary base.”

    I’m not following this. Are you talking about the *nominal* natural rate? Are you saying that the effects of the liquidity trap on the real economy are intermediated via the quantity of the monetary base? That’s not at all Krugman ’98, or the standard NK liquidity trap. In those models the effects on the real economy are entirely via the real rate.

    I’m an outsider at this blog, so maybe I’m asking things that have been answered a million times before. If so, I apologize.

  22. Pingback: The Coin is Dead! Long Live the Coin! | Monetary Realism

  23. ” It is no more deflationary than maintaining the same FFR by draining excess reserves in OMOs”

    Fundamentally wrong. It’s been exceedingly deflationary (& remains so). The Fed’s path was already contractionary in Dec. 2007 (gDp would have contracted regardless). But as gDp fell, the Fed simultaneously tightened even further (via the introduction of IOeR).

    The CBs compete with the non-banks (NBs) for their share of the loan-pie, but not for their funding. But the NBs cannot compete with the CBs (money never leaves the CB system, except for hand-held currency & even then OMOs have offset their withdrawal- since 1933). CBs are credit creators. NBs are credit transmitters.

    The remuneration rate permitted the CBs to pay higher rates on funding, than the NBs could competitively meet. Thus IOeR induced dis-intermediation (where the size of the NBs shrink, but the size of the CB system remains the same). Historically, an outflow of funds from the NBs would just reduce the overall supply of loan funds (but not the supply of money). Actually, the money stock also fell as shadow banks do have the capacity to create new money & credit (processing thru the pool of re-pledging) They operate like the unregulated, prudential reserve, E-D banking system has since the 1960’s.

    IOeR alters the construction of a normal yield curve, it inverts the short-end segment of the yield curve –known as the money market. The money market is differentiated by its position on the yield curve (i.e., short-term borrowing & lending with original maturities from one year or less). @ .25% the remuneration rate on excess reserves is still higher than the Daily Treasury Yield Curve almost 2 years out.

    The liquidity run during the Great-Recession was compounded again by a flight to safe-assets by both the CBs & the NBs…

    The non-banks are the most important lending sector in our economy — or pre-Great Recession, 82% of the lending market (Z.1 release, sectors, e.g., MMMFs, GSEs, etc.). Every effort should encourage the flow of savings thru the NBs (the customers of the CBs). IOeR does the opposite. It stops & reverses the flow of savings into real-investment. It is important to understand that the CBs don’t loan out existing deposits (saved or otherwise).

  24. “your beef is with a wrong (too high) FFR target”

    Given the different concentration in long-term earning assets that has historically existed between the CBs & NBs, whenever the central bank raised its FFR (policy rate) faster than the NBs could adjust (maturity mis-matches when borrowing short & lending long), the Fed induced dis-intermediation within the non-banks – true.

    So not understanding the difference between money & liquid assets the Fed just turned 38,000 financial intermediaries into 38,000 CBs thinking this would correct their problem (DIDMCA of March 31st 1980). Hence the S&L crisis was inevitable. Leland Pritchard (Ph.d, economics, Chicago 1933), predicted in June 1980 that the GSEs would have to replace the thrift’s dominance in the housing market because the Fed just forced the MSBs & S&Ls to act like banks in order to survive.

    IOeR is confusion compounded. Lending by the NBs matches savings with investment. CB held savings represent a leakage in National Income Accounting. CBs create new money when they lend or invest. CB lending expands the volume & directly effects the velocity of money. NBs lend existing money that has been saved (& unspent savings, or the non-use of savings, exerts a depressing effect). The CBs could continue to lend if the public ceased to save altogether. Thus it will take increasing infusions of Reserve & CB credit to generate the same inflation-adjusted dollar amounts of gDp. I.e., more & more financing will have to be accomplished by the creation of new money (stagflation).

  25. Jose Guilherme

    I think one or several of the many STF papers deals with the problem inherent in setting the floor at 0%: as I recall it (old readings…) the Fed has extra difficulty in keeping the rate near the desired target, and so the rate will tend to fluctuate between the upper and lower bounds and becomes much more volatile.

    A corridor system, Canadian style, with upper and lower bounds within 25bps of the target makes the management of the funds rate much easier.

    Btw, I think the time has come for STF to synthethize his many papers on monetary policy, Randy Wray style, in a new, not too technical Primer. The lack of this type of publication in the MMT literature (dispersed as it is among countless blog posts, seminars and technical papers) goes a long way towards explaining its difficulty in making its voice clearly heard in the present debates.

    Even the much announced MMT macro textbook by Wray/Mitchell has been languishing for years, its date of publication constantly postponed. No wonder many people still fail to take MMT with the seriousness it deserves.

  26. Ben Johannson

    I’ve wondered if a lot of it is simply that the Fed declares what the target interest rate will be and banks comply or else find the Fed undercutting them if it has to step in and conduct OMO. In other words managing a specific rate, like the 2.5% in your example, is simply god shouting down from the mountain and his followers obeying because its easier than pissing him off.

  27. Wh10,

    This continues our discussion above:

    Starting here:

    http://neweconomicperspectives.org/2013/01/understanding-the-permanent-floor-an-important-inconsistency-in-neoclassical-monetary-economics.html#comment-100077

    Stopping here:
    http://neweconomicperspectives.org/2013/01/understanding-the-permanent-floor-an-important-inconsistency-in-neoclassical-monetary-economics.html#comment-100888

    As noted, I’ve now read STF’s 2002 paper quite closely.

    One factor relevant to our discussion so far may be my misuse of the term “liquidity effect” as it relates to Scott’s 2002 paper. However, I think that’s a minor technicality within the substance of understanding the elasticity argument. But let me correct it now:

    STF defines “liquidity effect” on page 26:

    “This section defines a liquidity effect as a change to Fed balances sustained throughout a maintenance period to generate a change in the federal funds rate. Shorter-term, temporary operations that “nudge” the rate closer to the targeted rate on a given day or “signal” a change in the FOMC’s target to federal funds market participants are unrelated to a liquidity effect.”

    I can relate this directly to what I wrote.

    I divided the dimensions of the analysis into two distinct dimensions, which I labelled as a) and b) in the discussion above.

    Scott’s use of the term “liquidity effect” relates only to the scenario I discussed in b).

    His description of “shorter-term, temporary operations that “nudge”” corresponds only the scenario I discussed in a).

    I erroneously applied the term “liquidity effect” to my description in a), thinking of the effect of a vertical supply curve shift on a near vertical demand curve (that effect being a change in the funds rate) as the liquidity effect of a change in the quantity of reserves supplied on the fed funds rate. That seemed to me like a common sense use of a phrase, absent specified meaning otherwise. But STF does specify liquidity effect in his paper otherwise as pertaining to what I described in the b) scenario. So let me drop “liquidity effect” in application to a) and use “nudge effect” instead.

    So apologies there – but notwithstanding that, my descriptions of a) and b) scenarios map to Scott’s own descriptions of the difference between the two, in substance, IMO.

    Here are my a) and b) descriptions reproduced verbatim from the discussion above, but where I’ve replaced “liquidity effect” in a) with “(nudge effect)”:

    ………….

    a) The (nudge effect) as it exists under an unchanged fed funds target rate. There is a (nudge effect). There is no question that the Fed alters the supply of reserves in response to changes in the distribution of reserves in the system and the effect that those shifting distributions have on the effective ease or tightness in the system and therefore on the actual trading rate for fed funds. And that all works according to the graphical logic of a vertical supply curve (that the Fed shifts iteratively in response to changing reserve distribution and reserve demand conditions) and a near vertical demand curve which itself will be shifting due to the effects of shifts in the distribution of reserves among different banks in the system and their behavior in competing for or attempting to shed reserves with respect to their individual positions. There is no way that Scott is going to contradict that or has written anything to the contrary – if he does, I’ll retire from blog commenting.

    b) A completely different dimension is the presence of a liquidity effect in the case of a change in the target rate. I described that in some detail above. It should be self-explanatory. In that explanation, I separated out the dimension in a) above by assuming that the Fed has at a point in time achieved its target rate with the actual rate. The logic I described then implies that there is no need for the Fed to move its reserve supply curve in conjunction with changing the target rate. Are there “nudges” etc? Of course – some of that will be due to a bit of market imperfection I suppose and some will be due to the fact that changes in respect of a) above are occurring continuously, and can’t be stopped just because b) happens. It’s difficult to untangle the two entirely since both operate concurrently. But they are different forces. The point is that from an analytical and pragmatic operating perspective, the ongoing supply curve changes described in a) are of a completely different breed than any supply curve requirements that might be suggested by b) alone – and there are none in the case of b), once you separate out the effect of a) and any minor market noise otherwise. All those quotes you have from Scott are completely consistent with my description here. E.g. the Fed can’t inject reserves in conjunction with a b) change – UNLESS there is a reserve change required that fundamentally reflects an underling a) type incident – e.g. some unusual reserve distribution pattern that appears by coincidence on the day of a target rate change. Because if the Fed has hit its previous target at the time of the target change, the demand curve drops vertically and any additional reserves put into the system will cause the rate to drop below the NEW target – just follow the geometry I described earlier. That is why even with a b) type “nudge” injection, it must be quickly reversed, other things equal.

    …………

    Some quotes from Scott’s paper, as I see them relating to my a) and b) scenarios:

    “Open market operations are not intended to continuously add/subtract to the quantity of reserves, since doing so would send the federal funds rate substantially below/above the FOMC’s targeted rate.” Page 9

    – Refers to b) in that easing for example does not require a permanent addition to reserves – or even a temporary one, apart from coincidental intersecting influences of type a); that’s the textbook error – that the Fed increases the base when it eases

    “Temporary open market operations are aimed at keeping the federal funds rate at its target on average through temporary additions or subtractions to the quantity of Fed balances.” Page 10

    – Refers to a)

    As I said, the maintenance period is not an issue for what we are discussing – which is the difference between a) and b) type adjustments. This difference would hold whether the maintenance period is 1 day or 2 weeks. Scott’s point in discussing the maintenance period is to highlight the tendency toward increased inelasticity of demand as bank reserve managers come into the end of a reserve maintenance period – which BTW is something I can corroborate from direct personal experience in a bank reserve management function. But that is not pertinent to the core differentiation issue we are addressing here.

    “When the Desk does intervene to increase or decrease Fed balances, even if the intervention occurs on the day of an FOMC rate change, the intervention could have been made in accordance with the already existing period-average need and would thereby be unrelated to a liquidity effect since the operation would have occurred regardless.” Page 26

    – Completely in synch with my a)/b) differentiation, in that reserve adds or subtracts in the case of b) are only due to intersecting influences from a)

    • Scott fullwiler

      Thank you, JKH! I’ve been traveling this weekend so couldn’t take part.
      What youve written inthis thread is exactly the same as I’ve intended in my own writings. My use of the term liquidity effect only refers to your (b). I have been very careful in everything I’ve written to differentiate between (a) and (b) as you’ve defined them. My point has been that (b) is related to a belief that banks can “do” something with more balances. As you say, (a) is not controversial, but (b) is a huge barrier for neoclassicals to overcome. Hence my focus on (b). Hopefully this does clear things up for wh10.

      • Scott, I do think I am on the same page as JKH. Do you agree with the following, because I think JKH would –

        JKH has said that if banks were stupid, and their demand curve (which slopes down slightly in the real world) didn’t move in response to the announcement of a new target, the Fed could in fact change qty of reserves to achieve a new target FFR, and leave it there (not just a temporary nudge). We don’t see this in the real world because banks KNOW it can do this, thus they move their downward sloping demand curve in response to the announcement/signal automatically. BUT the fact that qty CAN change the rate *in the real world* IS the reason why banks do this automatically. Otherwise, there’d be no credible threat. Temporary nudges of the a) variety are not what underpins the credible threat. What underpins the credible threat is that the Fed can permanently move a vertical supply curve along a slightly downward sloping demand curve if banks were stupid.

        • “Temporary nudges of the a) variety are not what underpins the credible threat. What underpins the credible threat is that the Fed can permanently move a vertical supply curve along a slightly downward sloping demand curve if banks were stupid.”

          Not exactly wh10 – permanence is not the effective threat. If banks were stupid, the Fed in fact could do a nudge move – but it would have to be temporary – because other things equal, the trading rate would definitely move, but it would move too far unless reversed. The stupidity factor would be unlearned (i.e. bank reserve mangers would get smarter after being killed) – at which point there are now too many reserves in the system, other things equal – because the real world move without stupidity in fact requires no permanent injection. Therefore the stupidity compensating nudge has to be reversed once the stupid have actually learned that they were wrong.

          Here’s perhaps a useful point, I hope:

          When the Fed does a nudge move in mode a), it is permanent in the sense that it is an adjustment required to move the actual rate to the target rate – with no necessary expectation of reversing that move – but it is temporary in the sense that the demand curve is constantly shifting due to the dynamic effects of reserve distribution among competing banks. The system requires some excess to operate just because of friction – and that friction can be conceptualized in terms of the continuously changing reserve distribution as bank reserve managers are constantly deciding what their next tactical move will be in order to to manage their positions. This set of factors is ever changing, with the consequence that the system demand curve is ever changing and that the Fed is ever reacting to those changes by shifting its supply curve through intervention.

          But the Fed doesn’t need to do such nudge moves in mode b) simply due to a change in the target rate, because bank reserve managers as a whole are too smart to get caught by the kind of error I noted earlier.

          And the only reason to do a nudge move in mode b) is because the reserve distribution influences characterized in mode a) are intersecting with the b) factor at the moment the Fed moves its target – i.e. a) influences are superimposing on b) while b) is in progress. But that’s an intersection of two different dynamic effects.

          These two effects are very separate analytically and in the real world – proven because the a) mode can definitely be quiescent at times (i.e. smooth reserve operations at the initial target) when the target change is announced – and then you get a pure b) effect without much Fed intervention.

          re “same page”:

          Words are words and pages are pages, and pages change … the quest for mutual understanding is never quite done or perfect is it?

          sort of like managing the a) effect

          :)

          • Scott Fullwiler

            Very well said. thank you again.

          • JKH, I have but one remaining quibble, which is below. The rest of your post I get, and I am honored that you have cared about achieving mutual understanding with me through this massive nerd-fest.

            “If banks were stupid, the Fed in fact could do a nudge move – but it would have to be temporary – because other things equal, the trading rate would definitely move, but it would move too far unless reversed.”

            Perhaps my use of the word ‘permanent’ was a poor choice of words. Let me try again by doing a very simple model. Imagine a downward sloping demand curve like Scott has drawn in this post, and imagine that this characterization of demand stays constant for the near future (is not naturally dynamic). Imagine a vertical supply curve intersecting it at 4%. The Fed wants to set the rate at 3% and announces it. Banks are stupid and don’t get it, such that the demand curve is not going to shift. So, in response, the Fed shifts the supply curve to the right ever so slightly. Now supply and demand are in equilibrium at 3%. I don’t see why this nudge needs to be reversed. The demand curve did not shift down in this instance, there is no reason for it to (YET), and we assumed this characterization of demand stays constant for the near future. Now, the next time the Fed announces a rate change, banks will get it, and they will shift the demand curve down themselves, so as to avoid getting burned this time. In that instance, the Fed will not do any nudge, or if they do, they’ll overshoot, and have to reverse it. In fact, for the infinite future, this is how it should work, since banks will forever understand this is how the system will work.

            What I am trying to say is that banks’ ***understanding*** of that initial burn, even if just hypothetical, is what imbues the Fed with omnipotent power, is what makes the announcement/signal a credible threat for the banking system to move its demand curve automatically. And that burn is theoretically possible only because the demand curve slopes downward.

            Next phase of understanding: If the demand curve was perfectly inelastic, I don’t think that type of ‘burn’ would be possible in this channel system with a floor of 0%, assuming we want target a rate between the floor and ceiling. As such, it would seem that the interbank rate would be indeterminate between the ceiling and floor, unless there was something else the CB could do (such as say Lavoie’s point about the interbank rate naturally settling in the middle of the ceiling and floor, and the Fed being able to move the ceiling). But for another time, I beg of you.

        • Scott Fullwiler

          It depends on the country/system. In Canada, the demand for overnight reserves is vertical at zero balances. In the US, there’s a slight slope to it largely because in the US the payments system is so decentralized and the Fed’s method of getting balances into the banking system is so indirect (via repos with dealers). So, relying on “stupidity” is context specific and doesn’t apply everywhere. Also, even in the US, to deal with “stupidity” would require very, very small changes to reserve balances anyway, because the demand curve is steep enough that it takes a much larger change in the target rate than the Fed normally makes to significantly alter required reserve balances and desired excess balances. As Sandra Krieger says, the Fed’s estimates of how much it should add/subtract to accommodate banks are almost completely insensitive to more typical changes in the interest rate target–and she’s not relying on a “stupidity” effect, there.

          Note that my more recent use of the term liquidity effect comes from a paper in which I’m describing the “general” principles of central banking–I don’t find relying on “stupidity” to be a significantly general way to consider open market operations. I find a system like Canada’s to be a much more appropriate way to begin thinking about central bank operations, because that ingrains much better payments systems, endogenous money, the necessity of an interest rate target, and so forth right off the bat. Then we can see that there are peculiarities in the US system that make the potential for “stupidity” relevant, if not always economically significant.

          Another point–the 2002 paper (interesting to see people going back to that! that’s the first paper I ever published, btw–so just a year removed from being a runny-nosed graduate student) was trying to counter the primary research to that point on the liquidity effect, which mostly relied on monthly data, or at least not daily and certainly not intraday data. For the “lack of stupidity” effect, you’d need intraday data, for instance. At a monthly frequency for instance, you are essentially getting the reverse causation–the interest rate changes, shifting the demand, and the Fed accommodates–not a liquidity effect. So, there’s a historical context to the paper, and the “open mouth operations” or announcement effects view was pretty new at the time. As usual, even then I was trying to provide a more general perspective (hence the point in the paper about beginning analysis of CB ops and liquidity effect by assuming no RR) that would show the money multiplier and exogenous control of the monetary base views was inconsistent with how things can and do work.

          Also, I did discuss the credible threat a bit in “setting interest rates in the modern money era.”

          Anyway, after all that, YES (!!), I do agree with your characterization by and large regarding how things work in the US even if I might not say it the exact same way myself.

          • “Anyway, after all that, YES (!!), I do agree with your characterization by and large regarding how things work in the US even if I might not say it the exact same way myself.”

            I might cry tears of joy. This has been bothering me for 2 years.

    • JKH, thanks for sticking this through with me. You have said that I understand what you said, so I trust that we are on the same page (and I think we are).

      However, you have not acknowledged my argument that STF never transparently explains what *theoretically* underpins the credible threat of the announcement effect – of why we observe b) to happen as it does in the real world. And what theoretically underpins it is if the demand curve didn’t move (let’s say the market is stupid, wording taken from your example at 3:33), the Fed could move the vertical supply curve along a near but NOT perfectly inelastic demand curve, and leave it there (not just a temporary nudge later reversed) proving it has the power to change the target interbank rate by altering qty, if necessary. As you say, from there on out, it should never need to do that again, since the market will learn to drop the demand curve in response to the announcement. I have never seen that explanation from STF. Have you, and have I just missed it?

      I do not think a) is sufficient in explaining or implying the above, since a) is always discussed in the context of defending an existing interest rate. Moreover, when STF discusses b), such as in the quote I provided in the prior thread, which you did not address, it is strongly implied that the demand curve really is perfectly inelastic in the real world, making it fundamentally impossible, *in theory,* for the Fed to change the target FFR by changing qty of reserves (if the demand curve didn’t drop). All of that, combined with no explanation as to what theoretically underpins the threat, leads me to my position.

      So there is a difference between saying –

      1) for b), there is no need to do change qty of reserves to change target FFR in the real world, and we don’t observe such a thing, but it would be possible and necessary if all of the sudden the market went stupid

      2) there is no liquidity effect in the real-world for b)

      1) is much more transparent and comprehensive.

      “- Refers to b) in that easing for example does not require a permanent addition to reserves – or even a temporary one, apart from coincidental intersecting influences of type a); that’s the textbook error – that the Fed increases the base when it eases”

      It’s an error in the sense that we do not see that in the real world. But it’s critical to understand that the Fed’s ultimate ability to change qty of reserves ***is what underpins*** its ability to change the target interest rate without changing qty of reserves, ***in the real world*** – otherwise there’d be no credible threat behind the announcement effect! So I think you would agree that the textbook explanation is actually critical to understanding WHY the real world works as it does, even though it misses how it actually works in practice, and the author probably doesn’t realize it himself.

      This is quite ironic, because the textbook misses the point that *theoretically*, in the *ideal simple model,* demand should be perfectly inelastic, such that the Fed ***couldn’t*** change qty of reserves to change the target interest rate between the ceiling and floor! It’s quite a twister to reconcile this with the real world. In that model world, I don’t know what you do. I suppose you narrow the channel, to minimize variation, or you decide to adopt a different system (IOR).

      • Or you believe in what Marc Lavoie says, and set the floor and ceiling such that your target is halfway between them.

        • If either Scott or Marc said it about the facts of monetary operations, its pretty likely I agree with it – but I’ll come back on this

        • Jose Guilherme

          Let us then hear again from Marc Lavoie, writing about the Canadian system:

          “…we can say that there is a horizontal money supply of base money, (the curve) SS, at the target overnight rate. Whenever the demand for base money increases so does the supply of base money…the standard assumption (is that) the demand for base money is downward sloping (the interest rate being a measure of the opportunity cost of holding coins or bank notes – the higher the interest rate the less base money one wishes to hold). If there is a shift in the demand for base money…the supply of base money will accommodate at the target interest rate…”.

          And – in an enpirical illustration that takes into account the fact that, there not being reserve requirements in Canada, under normal circumstances bank deposits at the Bank of Canada are zero or very close to zero – Marc Lavoie goes on to write:

          “…tha Bank of Canada sets the supply of settlement balances at the level demanded by the banks. The supply is demand-led. In the case of…the subprime crises, there was a temporary large demand for settlement balances because banks were highly reluctant to lend to each other. Banks preferred to hold secured deposits at the Bank of Canada rather than lend these funds to other banks at the slightly higher overnight interest rate and risk losing their funds. This induced the Bank of Canada to set the supply of settlement balances much above zero, to keep the overnight interest rate on target”.

          Again, couldn’t be clearer IMO.

          • I need to learn how the Canadian system works. Do you have a link to this paper?

            In any case, we were specifically talking about the U.S. system here (not the general case, as STF puts it). And STF just wrote “I’ve had the same questions when I’ve thought about pure theory, so I totally understand where WH10′s been going with this and why,” so I feel my inquisition has been vindicated :).

            • Jose Guilherme

              There is no link available. I simply made a transcription from the Macroeconomics textbook by Baumol and Blinder – adapted (with the monetary chapters re-written from scratch) for the Canadian market by Lavoie and his colleague Seccareccia.

              It’s likely the best Macro textbook available, btw. The two chapters on money should be made mandatory reading for the incredibly simple yet realistic description they manage to make of the way the Canadian system works. Hard to match for clarity of exposition and depth of analysis, IMO.

              • Jose Guilherme

                Forgot to add: the web site of the Bank of Canada has lots of useful information and description of how the country’s monetary system works. Definitely worth a visit.

      • My immediate reaction on reading Scott’s 2002 paper was that our respective explanations were almost perfectly isomorphic – but I’ll have to come back later on this.

      • Scott Fullwiler

        See my comment above. For me, it’s about which approach is sufficiently general. I find “lack of stupidity” a workable explanation for the US case, but I don’t find it sufficiently general to be the “theoretical” case for understanding the essence of CB operations. Even the “nudge” (for me at least) is occurring because the Fed has set up its operations in such a way that a nudge might be necessary, rather than setting 0perations that transparently and efficiently accommodate demand at the target rate as is a more general case for understanding what CBs do. If you start from the theoretical position that there is a liquidity effect, then you miss this, and it’s not too far off from believing in exogenous control of the monetary base, etc. For me a proper “theoretical base” should not enable those sorts of mistakes in understanding something so central.

        • I understand what you are saying and the value of the general perspective.

          “rather than setting 0perations that transparently and efficiently accommodate demand at the target rate ”

          My problem is that if demand for reserves is perfectly inelastic at RB*, and the floor is at 0%, and the ceiling is x%, I imagine a vertical supply curve intersecting a vertical demand curve at all points between 0% and x%, and that seems problematic to me from the perspective of the Fed being able to set a particular interest rate between the floor and ceiling in a system like this.

          • wh10,

            Two answers:

            1) That’s why the US system *is* unstable. FF is way more volatile than the Canadian interbank rate because traditionally the “corridor” was from IOR=0% to the discount rate.

            2) If reserve demand is perfectly inelastic, then the CB just has to borrow $1 via repo and the rate will settle at the penalty rate (ceiling), or just lend $1 and the rate will settle at the floor? So if everyone knows that the CB might borrow $1 if someone offers below target, or lend $1 if someone bids above target thus severely punishing those who did that, then nobody will offer below or bid above target. The more inelastic the demand, the more omnipotent the CB. I’m not seeing the mystery.

            • Hi K,

              “If reserve demand is perfectly inelastic, then the CB just has to borrow $1 via repo and the rate will settle at the penalty rate (ceiling), or just lend $1 and the rate will settle at the floor? ”

              Right.

              ” So if everyone knows that the CB might borrow $1 if someone offers below target, or lend $1 if someone bids above target thus severely punishing those who did that, then nobody will offer below or bid above target. The more inelastic the demand, the more omnipotent the CB. I’m not seeing the mystery.”

              Is this a valid interpretation of the intent of the system? It’s intriguing but doesn’t feel right to me. Who is to say the banking system won’t prefer to have the interest rate at the floor or ceiling, and goad the CB into making that move, at which point the CB is SOL? Just doesn’t seem right to me. The CB should have a real, underlying mechanism to get the rate where it wants it through its own operational means.

            • Right on the intervention effect – except I wouldn’t define that as a vertical demand curve.

              The fact that the CB lends at some point above target means that the true demand curve at that time must slope down to the right – because a right shift in the supply curve moves the rate down – because the rate has to move down at some point in order to hit the target, ever.

              The fact that the CB borrows at some point below target means that the true demand curve at that time must slope up to the left – because a left shift in the supply curve moves the rate up – because the rate has to move up at some point in order to hit the target, ever.

              • Scott Fullwiler

                Yes, that’s true, for sure. And that’s probably why WH10 started the inquiry in the first place–for a perfectly vertical curve, it’s not in a pure theoretical sense clear why a fall in rbs would send the rate up, for instance, rather than down. I’ve had the same questions when I’ve thought about pure theory, so I totally understand where WH10’s been going with this and why–I think the distinction b/n (a) and (b) is the important one, overall.

                • Scott Fullwiler

                  Yes, that’s true, for sure. And that’s probably why WH10 started the inquiry in the first place. I’ve had the same questions when I’ve thought about pure theory, so I totally understand where WH10’s been going with this and why–I think the distinction b/n (a) and (b) is the important one, overall.

        • Hi Scott,

          Yes – my description applies more to the US.

          My personal experience with this goes back to when the two systems – Canada and the US – were more comparable. I’m admittedly not completely up to speed on the full detail of the Canadian system today.

          My use of the “stupidity” idea was extemporaneous. I meant it as a contradictory counterfactual, where there is a material liquidity effect in a target rate change. Any individual (stupid) trader who gets it wrong gets fired, as the market in large part does get it and prices accordingly. There’s probably a more elegant way for me to have expressed that idea – and then there is the difference with the Canadian system etc.

          • Scott Fullwiler

            It was all perfectly fine, so no worries, and obviously applicable. I was just trying to draw a distinction b/n that and a more general approach, but I might be splitting hairs at this point.

  28. Wh10,

    This may well be redundant given our overall discussion to this point, but in response to your:

    http://neweconomicperspectives.org/2013/01/understanding-the-permanent-floor-an-important-inconsistency-in-neoclassical-monetary-economics.html#comment-101317

    i.e. “I don’t see why this nudge needs to be reversed.”

    First, the opening demand curve is drawn on a graph in which the target funds rate is 4 per cent. It’s implicit in this that there is no a prior reason to assume that this demand curve is independent of the funds target rate. You can’t specify that it is, but you can’t specify that it isn’t. So you have to be open to the possibility that the demand curve is a function of the target rate. And we can show that it is as follows:

    Assume that the Fed drops the target funds rate to 3 and nothing happens. Funds continue to trade at 4.

    The Fed doesn’t like that and shifts the supply curve – in response to the fact that the market is still trading on the opening demand curve.

    The funds trading rate then moves down to 3.

    Now the situation is that the funds rate is trading at 3 instead of 4, but with more reserves in the system.

    At that point, it’s important to note that by assumption, absolutely nothing else has changed – i.e. we are segregating away any reserve distribution effect of the type described in mode a) discussed above or any other noise in the system, apart from the increase in the quantity of reserves.

    Given that ceteris paribus assumption, there’s no reason to believe that the system needs more reserves than before in order to clear and settle interbank payment liabilities.

    The system is set exactly as before, including reserve distribution and assumed payment patterns, by assumption. Only the supply of reserves and the funds trading rate has changed, which everybody is now comfortable with – at least momentarily.

    But if there’s no reason for more reserves needed than before to clear the system – that means that somebody is going to end up with excess reserves relative to what is they require to “square” their position.

    And I’d say that this is conceptually the point at which at least that particular reserve manager suddenly realizes he/she is on the wrong demand curve. And they will shift their demand curve down – as they should have done but failed to do so at the timing of the announcement. And that will now cause the funds rate to overshoot – because the supply curve is still hanging out to the right – so that the funds rate trading level drops down to 2 per cent in the case where the opening, intermediate, and reversing demand curve all have the same linear shape for example. The reserve manager is forcing the demand curve down by trying to get rid of his surplus reserves. And with the excess reserve increase as a system effect, you can spread the reaction of that particular reserve manager out to many reserve managers who now share the unnecessary increase in the quantity of reserves.

    So, the prospect of that realization by the market that it is on the wrong demand curve – combined with the central bank’s own learned anticipation of the inevitability of such outcomes under those conditions – means that the central bank would reverse its supply curve back to the opening position – before that hypothetical 2 per cent trading level is allowed to occur.

    • But JKH, we’ve assumed that there is elasticity around RB*, even if just barely. And we’ve also assumed that RB* is IS insensitive to a new rate – not an unrealistic assumption, in the short-term at least, according to my reading of STF’s papers (but regardless, I’m assuming it for simplicity, to understand what underlies banks incentives in this market). So, it seems to me, you’re changing the rules of the game when you start to insist that the bank reserve manager will start to get rid of their excess reserves. I don’t think that’s right – it’s only right if we assume a perfectly inelastic curve. But we’re not. We’re assuming some elasticity, we’re assuming they’re willing to hold some extra reserves.

      If we do it your way, it seems to me we’re assuming a world of perfect inelasticity. And if we assume that, we’re in a world where the Fed has no traction, and thus it’s hard for me to conceptualize why banks would shift their demand curves to match the Fed’s proclamations.

      • And in that perfectly inelastic (literally) world, it seems to me the CB has 3 choices – it can accommodate demand at the ceiling rate, the floor (0% in our assumed system), or have the rate be volatile in between. Seems to me they can minimize volatility by narrowing the channel. I guess like Fullwiler says, this is still de facto interest rate targeting, they’re still “accommodating demand at the target rate.” But if the demand curve is perfectly inelastic in a system like this, I don’t see you have credibility stabilizing the rate between the floor and ceiling, in this simple model (maybe throw in some imperfections, game theory, and you have Lavoie’s outcome of getting the rate to be stable in the middle). I think K is confirming this. I need to better understand Canada’s system, though – seems there is an important lesson there.

        Or maybe I’m just wrong about this. But I’m interpreting STF here as agreeing that there are some conceptual difficulties if we think about the pure theory of perfect inelasticity.

    • don’t overlook the fact that I’ve described a compound counterfactual

      i.e. a counterfactual within a counterfactual

      so be careful when attempting to pin down internal contradictions

      :)

      • Can’t tell if you’re saying my critique is definitively wrong haha. I need to step away from this for a bit, unfortunately. I’ll definitely revisit your hypothetical very soon though (day or two). thanks so much again.

  29. But JKH, we’ve assumed that there is elasticity around RB*, even if just barely. And we’ve also assumed that RB* is IS insensitive to a new rate – not an unrealistic assumption, in the short-term at least, according to my reading of STF’s papers (but regardless, I’m assuming it for simplicity, to understand what underlies banks incentives in this market). So, it seems to me, you’re changing the rules of the game when you start to insist that the bank reserve manager will start to get rid of their excess reserves. I don’t think that’s right – it’s only right if we assume a perfectly inelastic curve. But we’re not. We’re assuming some elasticity, we’re assuming they’re willing to hold some extra reserves.

    If we do it your way, it seems to me we’re assuming a world of perfect inelasticity. And if we assume that, we’re in a world where the Fed has no traction, and thus it’s hard for me to conceptualize why banks would shift their demand curves to match the Fed’s proclamations.

  30. Elementary question: What is meant exactly by “elasticity” and “inelasticity.” For example, wh10 above:

    “we’ve assumed that there is elasticity around RB*”

    Also, here is a place that Nick Rowe uses these terms:

    “In Canada, over a longer horizon, the supply of reserves is perfectly elastic at 2% *inflation*. It is perfectly *in*elastic with respect to the rate of interest (except at very short horizons).”

    here:

    http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/03/banking-mysticism-and-the-hot-potato.html

    I’m going to guess that Scott will disagree with that statement from Nick, but how is Nick using these terms, and what is meant by “(in)elesticity” “with respect to” inflation or the rate of interest?

    • I don’t think that’s an elementary question. I became interested in this particular discussion because I think the concept of elasticity is not so intuitive, and requires a fair bit of hard thinking, depending on context.

      I can’t speak for Scott, or Nick, but here’s one possibility that occurs to me:

      The type of supply and demand curves we’ve been discussing are focused on short term Fed operations. The idea of a vertical supply curve for reserves suggests that we’re talking about a single Fed money desk decision to target reserves at a particular level (at least implicitly) at a particular point in time – e.g. open market trading on a given day. In making that decision, the Fed estimates all sorts of factors involving the system flow of funds in aggregate, and what it thinks the (implicit) demand curve is for funds. They decide on a reserve add or subtract on that basis. So they shift their supply curve, but once that’s done, it’s vertical and fixed – until the next decision to shift it.

      We’ve also been discussing what the demand curve looks like – vertical or near vertical – for a given target situation, as well as what everything looks like when the Fed institutes a target rate change.

      Regarding your question, this is what I think (and it may not be related to what Nick was saying):

      If you assume that the Fed has a given target funds rate for a given time period, then you can look at supply and demand for reserves over the full span of that period. I.e. instead of looking at a point in time for a given supply curve decision in a given target rate situation, look at the time period for all such decisions between the time points of target rate changes.

      From that perspective, you can interpret the path of the Fed’s decisions as being a sequence of “nudges” (as we’ve been speaking about) over time, either adding or subtracting reserves, all of which are intended to maintain the funds rate trading at target. That pattern will consist of back and forth shifts of a short term vertical supply curve, as the Fed responds to changes in reserve distribution among banks, and corresponding back and forth changes in the reserve demand curve. The system will behave as if reserves are tight or easy, depending on how reserves are distributed, which in turn depends on constantly changing flow of funds patterns for any given reserve setting.

      That back and forth shifting of supply by the Fed becomes is a horizontal pattern – meaning that the Fed will adjust (shift) supply in whatever way is required in order to main funds trading at a given target over time. And that horizontal zone constitutes a horizontal supply curve in that context – which could be interpreted as infinitely elastic – for a given supply range over a given time period – as opposed to infinitely inelastic at a point in time as earlier discussed.

      The key is that the time horizon is different for the two different interpretations of the supply curve.

      (I have no idea if Nick was talking about anything like that.)

  31. Canada’s central bank takes the PD’s out of the equation? – excellent. But it’s so much tripe to think reserves & the policy rate can be disassociated. There is not a liquidity trap. The Keyneisan trap is to believe that the money supply can be managed through interest rates. Those that actually trade know this well. Greenspan & Beranke were directly responsible for the depth of the Great-Recession.

  32. “I’ve written numerous times already about how a deficit “financed” by bonds vs. “money” doesn’t matter in terms of inflationary effect.”

    I agree with that in the present. What about in the future?

  33. “You probably see where this is going now, if you didn’t several steps ago. To actually carry out QE, the Fed must either accept an interest rate target at zero or pay interest at the target rate if it wants an interest rate above zero. But these are precisely the two cases described above for which QE doesn’t work in the neoclassical view. The only other thing the Fed could do would be to reverse the QE and drain all the excess reserve balances, but this simply reverses the QE as if it had never happened.”

    What about raising the reserve requirement?

  34. “In short, in either case, the quantity theory of money now does not run from “money” to inflation, as velocity falls in kind with the increase in “money.””

    Define money in the quantity theory.

  35. Pingback: Understanding the Permanent Floor—An Important Inconsistency in Neoclassical Monetary Economics « Economics Info

  36. Pingback: Floors and ceilings | Fifth Estate

  37. Pingback: Drop It: You Can Call for Helicopter Money but Drop the Call for "Coordination" | New Economic Perspectives