Friday, April 27, 2018

There's water everywhere, but John Taylor wants us all to be thirsty

"Water water everywhere, and not a drop to drink" - Rime of the Ancient Mariner (Gustave Doré woodcut)

In a recent paper, John Taylor rhapsodizes about bringing back the good ol' federal funds market:
I think the case can be made for such a framework. Peter Fisher ran the trading desk at the New York Fed for many years, and knows well how these markets work. His assessment is that such a framework would work, saying “we could get back and manage it with quantities; it’s not impossible. We could just re-engineer the system and go back to the way we were.” I spent time in the markets for federal funds watching how they operated in those days, and I wrote up an institutional description of how good experienced people traded in these markets, and I developed a model showing how the market worked.
The fed funds market is currently moribund, but just a few years ago it was buzzing with activity. Banks that didn't have enough reserves at the end of the day to meet requirements could go to the fed funds market and buy them from banks who had excess reserves, the price they negotiated referred to as the fed funds rate.

I disagree with John Taylor. Resuscitating the fed funds market is not a good idea. The fed funds market is no longer used because the Federal Reserve has stuffed the market with so many reserves that banks no longer need to buy them from other banks to meet their requirements. But this cornucopia is a good thing. Any effort to bring back the fed funds market would ruin it.

Let's set up an analogy. Imagine a country called Waterland that gets tons of rain and has plenty of lakes and rivers. Since everyone has immediate access to water, there is no market for the stuff. The price of water is zero. Say that the government establishes control over the waterways and rainfall. It decides to limit the amount of water that is available to the citizens of Waterland. In response to this artificially-imposed scarcity, a market develops in which citizens buy and sell water among each other. 

Markets are great. They allow those with too little of something to trade with those who are good at conserving what they need, both sides improving their lot in life. But this particular market should never have existed in the first place. Water is plentiful in Waterland, and so it should be a free good, not a market-traded one. The entire apparatus that has been built around the exchanging of water—informed dealers, speculators, exchanges, warehouses, networks for transporting water to and from market, auditors and lawyers involved in verifying water transactions—represents a waste. By consuming resources in constructing and operating the market, other more important projects never see the light of day. If the absurd water scarcity were to be removed, the market for water would disappear, freeing up resources for more socially beneficial uses. 

Reserves, like water in the previous example, should by all rights be free. The only effort the Fed incurs in introducing a new unit of reserves into circulation is a keystroke or two. This means that the Fed can provide a bunch of new reserves, say by conducting open market operations, without incurring any costs whatsoever. As the Fed continues to mouse-click new reserves into existence, the demand that each individual bank has for reserves will eventually be satiated. Once that point is reached not a single bank will need to bid for the reserves of another bank, and so there will be no activity in the market for reserves. The fed funds market is effectively dead, as is currently the case.

Taylor wants to bring back the fed funds market. But this would mean putting an artificial constraint on the amount of reserves that the Fed supplies, much like Waterland's frivolous constraint on water. Banks, their satiation for reserves now being replaced by an artificial hunger, would suddenly be willing to pay a fee to other banks in order to get their hands on some reserves.

A whole fed funds trading apparatus would re-emerge. Traders would have to be hired and trained to to fill newly-formed fed funds desks. Bank resources would be diverted away from other valuable projects towards plotting the best way to time outgoing payments, the idea being to reduce the need to hold reserves in order to lend them out in the fed funds market. The Fed itself would have to rehire Peter Fisher to run its open market desk. All of this would be an expensive investment of time and money, diverting resources from other more socially beneficial activities.

In calling for a return to the days of an active fed funds market, it is as if Taylor were advocating for an artificial constraint to Waterland's supply of water, solely because he admired the market for water that emerged. Never mind that the whole water trading apparatus, though wonderfully efficient, represents a massive missallocation of resources. Given that I'm pretty sure Taylor would not want to kickstart a water market in a hypothetical Waterland, I don't understand why he is so keen to reboot the fed funds market.


  1. The premise of this post, that liquidity, like water, is or can be a "free good," is false. FWIW, in Floored! ( I directly address this issue as follows:

    "although central banks can create nominal reserves “out of thin air,” as it were, increasing the stock of such reserves effortlessly, they cannot effortlessly increase an economy’s real supply of savings. For that reason, when banks elect to accumulate excess reserves in response to the Fed’s creation of additional nominal reserves, because they’ve been encouraged to do so by a relatively attractive IOER rate or for any other reason, real savings are directed toward the Fed, and thence toward those from whom the Fed itself has borrowed, leaving that many fewer such savings for other prospective borrowers. In a phrase: there ain’t no such thing as a free liquidity lunch."

    Note here the crucial distinction between the quantity of "reserves," which the Fed alone (largely) determines, and that of "excess reserves," which it can determine only by making reserves more attractive relative to other assets. Liquidity is a fn. of total excess reserves, not total reserves. Before October 2008, bank loans of all sorts were equal to almost 100% of bank deposits, whilst excess reserves made up roughly 0%. The figures are now about 80% and 20%. Like I said, there's no free lunch here.

    All this is apart from the damage done by destroying the unsecured interbank funds market, which normally provides important incentives for interbank risk monitoring (see Craig Furfine's work). There are far better ways to economize on resources than by having banks sit on trillions of dollars of supposedly free (but actually quite costly) excess reserve balances.

    1. "...real savings are directed toward the Fed, and thence toward those from whom the Fed itself has borrowed, leaving that many fewer such savings for other prospective borrowers."

      I think I agree that that making reserves plentiful rather than scarce cannibalizes certain private sector borrowing and lending activities. More specifically, it destroys the fed funds market, which was a silly market anyways!

      "There are far better ways to economize on resources than by having banks sit on trillions of dollars of supposedly free (but actually quite costly) excess reserve balances. "

      C'mon George, this is hyperbole. No reasonable person thinks that a policy of plentiful reserves requires trillions of dollars. Surely you don't think that's what I mean?

      "All this is apart from the damage done by destroying the unsecured interbank funds market, which normally provides important incentives for interbank risk monitoring."

      Or alternatively, the fed funds market is an artificial government-created market foisted on banks that imposes on them a set of monitoring costs that, in a free market, they wouldn't choose to enact themselves. By retiring the fed funds market, the ongoing damage it imposes on banks is removed and their resources can be better deployed. Who knows, banks might even choose to redirect liberated resources into their own cheaper and more efficient systems for monitoring risk.

      George, I find it odd how in our ongoing debates on this topic, I'm the one defending the free market solution, plentiful reserves, and you're against it, arguing in favor of a government constructed market!

    2. JP, as you are among my favorite living writers on money, I promise that I'm not at all inclined to disagree with you on any topic. Still, I still disagree with you on this one, sincerely if perhaps wrongly.

      My disagreement doesn't stem from considerations of which policies or arrangements are or are not consistent with a "free market solution." Despite my work enthusiasm for past free banking systems,a and commitment to understanding how market forces work in money and banking, I remain, at bottom, a conventional utilitarian sort of economist, and as such favor those arrangements that seem to me more stable and more efficient over others, whether free market or not, that are less so. My beef concerning the Fed's present, liquidity-enhancing floor system of monetary control is based mainly on my opinion that it wastes resources (and would still do so even if the stock of excess reserves were considerably smaller than at present) and, secondly, on my belief that the presence of an active unsecured interbank lending market contributes to monetary stability through its tendency to encourage interbank monitoring.

      Having said that, I think you are wrong to characterize the old interbank market, and the lower bank reserve ratios that went hand-in-hand with it, as a product of government intervention. So long as banks have existed, they have endeavor one way or another to economize on reserves, especially by taking advantage of opportunities for netting of interbank dues but also by other means, including taking advantage of opportunities for overnight interbank borrowing to make up for last-minute reserve shortfalls. The Fed Funds market was no exception: it was started by a handful of NYC banks in the early 1920s as a cheaper alternative to the Fed's discount window to make up for temporary reserve shortfalls (see I suppose you could blame the Fed for not keeping the discount rate below market, and thereby credit the Fed for this development. But I think it more reasonable to regard any below market discount rate itself as a distortion, and a very problematic one at that. It's also true that a substantial share of activity in the Fed Funds market was largely driven by statutory reserve requirements. But even in the absence of such requirements (which I believe quite unnecessary) banks would still have found it desirable to resort regularly to overnight interbank loans as a means for avoiding precautionary reserve shortfalls.

      (To be continued!)

    3. (Continues previous):

      Banks have traditionally sought to hold as few reserves as possible, relative to their total liabilities, so as to take the fullest possible advantage of more profitable lending opportunities. Adam Smith eloquently explains the advantages this practice had for economic development in his day. The context of Smith's argument was, of course, a specie standard; and specie is an especially costly reserve medium. But although claims held against a central bank are less costly than specie, they are costly nonetheless. As I said earlier, this wouldn't be so if central bank portfolios merely replicated the portfolio of an efficient commercial banking system. But that is never the case in practice, where -- for both good and bad reasons -- central banks generally limit their portfolio holdings to a very narrow set of assets, consisting mainly of domestic government securities, foreign exchange, and (yes) gold. Only rarely to they make loans of any sort, except to banks. Now would we want them to do so.

      And so I repeat that the supposedly "free" liquidity the the Fed supplies under its floor system, by making reserves bear interest at a relatively attractive rate, and by then increasing the nominal quantity of reserves, is not really free at all. The cost is a decline in the share of financial intermediation that's directed toward financing the assets on the Fed's portfolio, and an equal decline in the share directed toward assets, apart from cash reserves, that normally make up the bulk of commercial bank portfolios--namely, bank loans of all sorts. This decline is far from consisting in a decline in interbank loans only, which ordinarily (that is, before October 2008) made up but a very small portion of commercial bank assets. Mostly it is a decline in lending to businesses, farmers, and consumers.

      Finally, although my reason for preferring a corridor system in which banks hold very few reserves isn't simply that that system would preserves the fed funds market, I do believe that market to be valuable, not merely as a means for reducing banks' overall need for cash reserves, but as a vehicle for interbank monitoring. Concerning the advantages of such monitoring, and how the Fed Funds market encouraged it before the crisis, I refer you to Craig Furfine's research,

    4. For the first "decline" in my penultimate paragraph kindly read "increase."

    5. George, thanks for your kind words. I'm honoured. But I've had some great teachers, including yourself! Thanks for the link too, I'll check it out. I confess that I'm not too familiar with the historical development of the fed funds market, I'm just trying to muddle through by thinking about the theory first. So perhaps I should do my homework.

      Part of our disagreement seems to come down to which private lending markets get displaced by a policy of plentiful reserves. I think that the fed funds market, the overnight interbank market, is the first to be displaced. As long as the central bank stops at that point, then no other lending activities are displaced. In Canada's case, a targeted policy of plentiful reserves would require just $2 billion in settlement balances, so maybe the Fed would require $50 billion. (The Fed's current policy of trillions in reserves would not qualify as 'targeted'). Whereas you think that a targeted policy would displace not only the fed funds market but lending in general, for instance loans to farmers, businesses, and consumers.

      If you did agree with me that a targeted policy was capable of knocking out just the fed funds market, then our disagreement comes down to whether the fed funds market is a good thing or not.

    6. "Part of our disagreement seems to come down to which private lending markets get displaced by a policy of plentiful reserves.I think that the fed funds market, the overnight interbank market, is the first to be displaced."

      I agree with that statement. But it's necessary to distinguish between "first to go" and "the only market to go providing the amount of new settlement balances is limited," for these aren't the same. When banks collectively find themselves with extra reserves, the first adjustment they make does indeed consist of reduced participation in the Fed funds market, because that is the part of their portfolios that's most readily adjusted. But over time they cut back on other forms of lending as well, and will do so regardless of the total quantity of new settlement balances made available.

      The best way to think of this, I believe, is in terms of the marginal equalization of returns on alternative assets. At the margin, adjusting for risk, maturity, etc., the return on all sorts of bank lending must equal that on (excess) reserve balances. That generally means that, as the IOER rate goes up, lending of all sorts adjusts downward. Because the IOER rate itself is fixed, while rates in other markets are declining at the margin, any further increase in reserves involves a increase in the real quantity of reserve balances demanded, and no increase in other asset holdings, so that the share of excess reserves again increases relative to that of all other assets.

      Often I'm tempted, in thinking about the consequences of a floor system, to set aside altogether changes in the nominal stock of bank reserves, and to ask only what the logical consequences are of an increase in the return on bank reserves relative to that on other assets, where the increase suffices to make reserves earn at least as much as some other assets earned before the change. Thinking in these terms brings the real consequences of the policy change into bold relief. Then one can go on to ask what difference it makes if the change in question is accompanied by a like change in the stock of nominal reserves, the main effect of which is (I believe) to offset what would otherwise be a deflationary outcome of the policy in question.

    7. Concerning the Canadian case, JP, as I'm sure you know, the B of C experimented very briefly (2009-10) with a plentiful-liquidity floor system, only to return afterwards to a corridor. Do you know why the Bank chose to go back? In other instances such decisions were based on concerns to avoid reserve hoarding and also on authorities' desire to have an active interbank lending market.

    8. "Do you know why the Bank chose to go back?"

      Here is David Longworth on the episode. He doesn't go into the reasons for switching back.

    9. "The best way to think of this, I believe, is in terms of the marginal equalization of returns on alternative assets. At the margin, adjusting for risk, maturity, etc., the return on all sorts of bank lending must equal that on (excess) reserve balances. "

      I agree, I think that's a good way to go about it.

      That being said, while a policy of moving from rationed reserves to plentiful reserves does involve a shift in IOER, it doesn't involve a shift in the return on reserves. I'm not sure if you agree or not.

      Here's an example. If we start with rationed reserves, the effective return on reserves is the fed funds rate, or what banks can get by lending overnight. Say that is 5%. Alternatively, the return on reserves can be thought of as the return banks earn by holding reserves themselves overnight, in which case they collect IOER (4.75%) as well as a non-pecuniary return on overnight holdings (0.25%), the sum of which is equal to the fed funds rate of 5%. The non-pecuniary return arises due to the rarity and usefulness of reserves.

      In moving to plentiful reserves, the IOER is ratcheted up 5% while at the same time the quantity of reserves is increased until their non-pecuniary return has been removed. Those banks that hold reserves overnight now collect IOER of 5% but the non-pecuniary return on overnight holdings has collapsed to 0%. So in moving from one regime to the other the 5% return provided by reserves hasn't changed.

    10. We agree that, even absent IOER, banks will equalize the marginal return on all assets, including non-pecuniary returns. So, if the FFR is 5%, the marginal unit of non-i-bearing reserves yields a non-pecuniary return of 5%. As the IOER rate is raised, banks adjust by holding more reserves, until, as you say, the sum of the pecuniary and non-pecuniary return on a marginal unit of reserves is equal to the going FFR. But the going FFR isn't given: as banks cut back on lending to increase their reserve ratios loan rates, including the FFR rate, will increase somewhat. The new equilibrium involves both a fallen non-pecuniary return on the marginal reserve unit and higher FFR and other lending rates. The higher the IOER rate goes, the less the banks lend, notwithstanding that there can be a non-pecuniary return to reserve holding. Of course, at a high-enough IOER rate, that non-pecuniary return falls to zero. Achieving that point is, in my opinion, the correct interpretation of the Friedman Rule. Any further increase in the IOER rate creates additional "liquidity" that supplies, by assumption, no further (non-pecuniary) benefits. But notice that this means that an IOER rate above other short-term market rates is not a source of valuable liquidity. In practice floor systems involve huge liquidity "overdosing."

      I am not, as you know, against IOER. But I do believe that a corridor system with a positive but modest IOER rate is far more likely to approach the efficient ideal than any floor system. On this see (and my discussion of same in Floored!)

    11. FYI,on the B of C's decision last month to further lowerits end-of-day settlement balance target:

      Of course even $250 million is ten times the former level of just $25 million.

    12. Ok, but you never really answered my initial question and associated example at April 30, 2018 at 11:18 AM.

      Specifically, assume that IOER is at 4.5% and the fed funds rate is at 5%, so that the non-pecuniary return on reserves is 0.5%. Now the Fed simultaneously increases IOER to 5% while increasing the quantity of reserves so that the non-pecuniary return is reduced to 0%. The return on reserves hasn't changed, right? It's still at 5%.

      The reason I'm focusing on this example is because it speaks to our original dispute over which private lending markets get displaced by a policy of plentiful reserves. We agree that returns on all bank assets are equalized, including non-pecuniary returns. But as I show in my example, implementing a system with plentiful reserves doesn't involve a change in the return on reserves, the rate staying at 5% throughout, so it needn't alter bank lending behavior. All that changes is that banks stop buying and selling fed funds.

    13. "Specifically, assume that IOER is at 4.5% and the fed funds rate is at 5%, so that the non-pecuniary return on reserves is 0.5%. Now the Fed simultaneously increases IOER to 5% while increasing the quantity of reserves so that the non-pecuniary return is reduced to 0%. The return on reserves hasn't changed, right? It's still at 5%."

      I don't disagree with this statement, assuming that reserves are held voluntarily, that is, that statutory reserve requirements are not binding at either the old or the new IOER rate. But I think you miss my point that the new equilibrium ipso-facto involves a greater demand for excess reserves, such as makes the marginal reserve unit have a pecuniary yield of zero, and that this increased demand for reserves comes at the expense of reduced bank lending on all markets.

      The reason this may not be evident in your example is that you treat the fed funds rate as given, whereas it actually depends on the IOER rate owing to the decline in fed funds lending that will accompany any increase in the IOER. That goes for other (marginal) lending rates as well. In practice, as the supply of interbank and other bank loans declines, the FFR will rise somewhat above its initial value of 5%, as will other marginal lending rates, while the banks will settle on excess reserve holdings that leave the marginal non-pecuniary return on reserves at some positive value. In concrete terms, you might end up with an equilibrium in which marginal returns on all loans are equalized at 5.2%, and the non-pecuniary return on reserves is .2%.

      The main point remains that the adjustments include both an increase in reserves held and a decline in bank holdings of other assets of all sorts.

    14. To be more clear: I agree with your statement as a description of a partial equilibrium outcome, but insist that the general equilibrium is one that involves a general reduction in bank lending of all sorts. This statement is, by the way, consistent with the general understanding that raising the IOER rate is a means of monetary tightening.

    15. George, I sort of get where you're coming from but I'm still struggling to keep up.

      Maybe I can better understand by asking another question. Pre-2008, the return on reserves was purely a non-pecuniary return, since there was no IOER. Say the fed funds rate was at 4.5% (ie the non-pecuniary return was 4.5%), and the Fed increased it to 5% (ie the non-pecuniary return is now 5%), so that the return on reserves suddenly exceeded the return on other assets and loans. Given the points you've made above, wouldn't this sweetening of the return on reserves have created a greater demand for reserves, and therefore come at the expense of reduced bank lending?

      I think I'm trying to tease out is why you think that IOER changes the game. A tightening of interest rates is achieved by a sudden increase on the return on reserves; why would it make a difference to commercial bank balance sheets if this increase is achieved by an adjustment to the pecuniary return (IOER) or the non-pecuniary return (pre-2008)?

    16. Because when the pecuniary return on reserves keeps up with the return on other assets, banks don't have to cut back on their real reserve holdings for the sake of raising the marginal non-pecuniary return on reserves to the increased level of pecuniary returns on other assets. The equilibrium portfolio allocations will therefore be different, with more reserves and fewer non-reserve assets in the IOER+FFR case as compared to the one in which IOER<FFR.

      But we must be very careful here as we are "reasoning from a price change" without considering why it has come about.

    17. For "IOER+FFR" please read "IOER=FFR."

    18. Thanks George, I'll have to think it through. I must confess that after all this I still wouldn't be able to pass an intellectual Turing test.

  2. I should add that there would be no cost IF central banks replicated commercial banks' uses of savings. But of course they don't.

    Here is a reductio to consider: suppose we want to really jack-up liquidity, and therefore raise the IOER rate enough to encourage banks to hold reserves equal to 100% of their deposits. We would then have an exceedingly liquid banking system, albeit one with no bank lending at all. Water is all well and good. But man does not live by water alone!

    1. Wait. If commercial banks could write bonds with a floating IOER+0% interest rate, and the Fed would buy these for exactly the bond's principal value, then the banking system could get as much liquidity as it needs, without affecting anything in the real economy. Both the Fed and commercial banks would enlarge their balance sheets with assets and liabilities that have exactly the same value and interest rate. Nothing happens, except banks are now swimming in excess reserves. (Because the excess reserves are regulated as having no credit risk, the banks don't even run afoul of the BIS capital requirements.)

      Yes, the shaman really can summon water in unlimited quantity.

  3. Unlike water, reserves don't already exist independently of the Federal Reserve's decision to create those reserves. It is not as if the Federal Reserve is restricting private banks from tapping into some primordial pool of reserves that is just independently floating out there in the economy all on its own.

    If reserves were specie, then it would be a different story. Specie exists on the market independently of whatever the Federal Reserve does. In that case, it WOULD be an infringement on the market for the Federal Reserve to stipulate that the banking system can only acquire such-and-such a level of aggregate gold reserves, especially if that level were below the amount of gold that gold-bearing depositors would voluntarily wish to deposit into the private banking system in return for an interest rate on gold deposits voluntarily set by the banks themselves.

    However, there is no "free market" level of reserves of Federal Reserve notes (or their electronic equivalents redeemable for paper FRNs on demand). A policy of supplying 1 trillion FRNs in return for 2% interest is just as much a policy decision as a policy of supplying 1 trillion FRNs in exchange for 0% interest, or supplying unlimited FRNs for 0% interest, or supplying unlimited FRNs at 2% interest, or supplying 1 trillion FRNs in exchange for a certain basket of assets, or supplying k% more FRNs each year for 0% interest.

    There is no "hands-off" policy, no way that the Federal Reserve "can't act" and "leave it up to the free market" aside from liquidating the entire Fed balance sheet in exchange for all outstanding FRNs and their electronic equivalents, retiring all FRNs, and changing regulatory requirements to allow banks to back their loans with an asset whose quantity is endogenously-determined by the world economy itself, such as a commodity like gold, rather than exogenously-determined by a monetary authority. (In the case of having banks back their loans with a foreign currency, the Fed has merely traded its policy constraints for the policy constraints and decisions of the monetary authority governing that foreign currency).

  4. Plus, to make the analogy between water and FRNs truly parallel, we'd have to imagine that, in addition to allowing banks to tap into as much water as they want, the Fed is also deciding to pay banks 1.75% interest for holding their water reserves. JP, you must surely recognize that this, at the very least, is a political decision, and not simply the free market going about its merry way.

    This will become increasingly obvious to the public once banks start getting paid positive real returns (>2%) for simply sitting on idle reserves.

    And it will ESPECIALLY become obvious once banks starting getting paid an IOER rate that is higher than the average weighted rate that the Federal Reserve earns from the assets on its balance sheet (for example, 4% IOER vs. ~3% return on Fed balance sheet), at which point the Federal Reserve will start becoming a net drain on the fiscal balance sheet (negative net remittances to the Treasury). If this continued for long enough, it would outweigh the accumulated positive remittances that the Federal Reserve has sent to the Treasury since the QEs, making the Federal Reserve a net drain on the national debt.

    It will also become obvious that reserves paying IOER are simply another name for perpetual, variable-rate Treasury bonds, and that excess reserves ought to be counted towards the national debt if they are going to cut into net remittances and thus the fiscal budget for the foreseeable future.

    "But surely IOER will never need to go up to 4%!" I don't know, if short-term interest rates need to get that high to curb inflation at some point (remember that short-term rates rose to 5.25% at the height of the last business cycle!), then the Federal Reserve will either need to reverse-QE to make reserves scarce enough in an absolute sense, or hike the IOER to 4% to make reserves scarce enough in a relative sense (relative to the quantity of reserves that banks would like to hold given that they will receive 4% interest on them).

    1. Matthew, you don't like my analogy very much. Your objection is based on what qualifies as "free" and "hands off", and that nothing that the Fed does qualifies.

      I think we are thinking about "hands off" in a different way. In a competitive market, firms produce up to the point where marginal revenue equals marginal cost. Monopolists that are able to escape the forces of competition can under-produce, imposing a dead weight loss on society.

      My argument is that under the sort of system that Taylor advocates, the Fed is limiting the quantity of money it produces so that it comes short of producing at marginal cost. Plentiful reserves is what a competitive firm would do, and even though the Fed isn't a competitive firm, shouldn't it do it's best to replicate that policy? That would remove some of the dead weight loss imposed on society.

  5. Regarding the "artificial" nature of the Fed Funds market: an analogous, and entirely private, market developed and was heavily relied upon for many years by the member banks of the pr-Fed Boston clearinghouse. For details see Cannon, pp. 233ff (

  6. One last thing and then I promise to shut-up, for now. Regarding this exchange:

    Me: "There are far better ways to economize on resources than by having banks sit on trillions of dollars of supposedly free (but actually quite costly) excess reserve balances. "

    JP: C'mon George, this is hyperbole. No reasonable person thinks that a policy of plentiful reserves requires trillions of dollars. Surely you don't think that's what I mean?

    I don't think it is what you think you mean, JP; but it is in fact what your analogy implies: if liquidity, like water, is or can be a "free" (that is, cost-less) good, then why would we not prefer to have as much of it as we possibly can? Why, indeed, settle for $2 trillion when we might have 4, or, 6, or 8, or any higher multiple, of perfectly liquid real reserve balances, all, as you say, by means of "a keystroke or two" (and, lets not forget, a little IOER)? What is the argument, consistent with your analogy, against having the Fed expand the quantity of excess reserves ad libitum, if not ad infinitum? The hyperbole, it seems to me, resides not in my reply to your argument, but in that argument itself.

  7. If water is scarce and there is a market for it, adjusting its quantity is a potent tool of aquatic policy. If it is abundant to the point of being a free good, there is one less lever to pull at the disposal of the head of aquatic policy.

    To make matters even more interesting, one of the remaining levers (the interest rate) appears to have very little movement left in one direction (it's against the ZLB). Those who read your blog know that it's a SNLB (Slightly Negative Lower Boundary) in the first place, and at least three ways in which it could be circumvented -- but Mr. Taylor and Fisher probably don't read this blog.

    The main point of the linked paper is, however, that if there were a market for this purpose, then its price, set by a multitude of self-interested actors, would respond faster and more accurately to economic shocks than a single team at the Fed. Even though it takes somewhat more resources due to duplication, by outsourcing macroeconomic analysis to commercial banks, you get an alarm clock to wake Rip van Winkle.

    One particular quote from the paper:
    "Without that connection, you raise the chances of the Fed being a multipurpose institution."
    It's already a clearinghouse, a LoLR, the issuer of physical circulating cash, and a monetary policy authority. Four in one, all of which could be separated from all others.

    1. "The main point of the linked paper is, however, that if there were a market for this purpose, then its price, set by a multitude of self-interested actors, would respond faster and more accurately to economic shocks than a single team at the Fed."

      Is that what you got from Taylor's paper? I had troubles picking out why he thinks it is so important to have a fed funds rate, apart from his repetitive contrasting of the words "market-determined" and "administered" and the underlying assumption (which he leaves unexplained) that the first must be better than the second.

      You're right that he also digs into the Fed for becoming a multipurpose institution, since it can presumably run a policy of industrial lending without hurting the policy rate, which is held up by interest on reserves. But this disconnect has always existed. A pre-2008 Fed could simultaneously set its policy rate and make massive industrial loans by sterilizing that lending, say by selling t-bills out of its portfolio, or issuing sterilization bonds, or asking the government to issue bonds and leave the proceeds deposited in the government's account at the Fed.

  8. I don't know if it is a good way to resolve the various concerns expressed here, but I like John Cochrane's proposal to consolidate all of the US Federal debt in the form of central bank reserves. It maximizes the government supply of settlement media, or "water" (as JP puts it). The interest rate paid on these reserves would be a policy variable useful for controlling the price level.

    But I also share George Selgin's concern that if banks are able to deposit into this fund and get a good rate of return, it could easily crowd out commercial lending and we would probably end up with the central bank needing to make large-scale asset purchases to keep the system going. Or even worse, it could create a need to loosen the purse strings by the system's spender of last resort. Thus I would have the size of this debt governed completely by the federal government's legitimate spending needs, and have shares available for purchase by commercial banks to add to their portfolios (as they may do now with government bonds). But I would not have this fund open for additional deposits by banks as a substitute for lending to businesses and consumers.

    I don't know if this is making sense as my understanding of the current system is weak and incomplete, but I think it's worthwhile to attempt a synthesis of the various proposals advocated by others.

    1. Is there a link for Cochrane's proposal, Anwer?

    2. Here is the published version of the paper John Cochrane has discussed a few times on his blog:

      He makes a point similar to yours: there is no point in starving the system of liquidity by issuing bonds in an inconvenient form that must be repackaged by financial intermediaries like money funds. We can just turn it all into settlement media, or "water" as you put it.

      Some people also apply the monopoly argument to issuers of safe assets (USA, Germany) and ask them to increase their activity as spenders of last resort. I don't think this is a reasonable demand, and improvements in systemic structure and organization are probably a better way to go, for all involved.

      A better structure in my thinking is to let money funds get access to these reserves, integrate these entities into the payments system, and then let consumers keep their deposits there. We'll be able to stop worrying about the system that is currently fragile and requires intensive supervision by a LOLR

    3. Thanks for that. In this post I'm just describing a way to implement the Friedman rule, which--as you point out--is something that Cochrane is trying to do too in his paper: "Financial arrangements designed to save on the lost interest of holding money are a social waste."

      Cochrane's solutions are often quite radical. The one I imply in this post is small and incremental. For instance, the Bank of Canada currently leaves around $250 million in settlement balances in the system overnight. All the Bank would have to do is boost this to around $2 billion and it would achieve the Friedman rule. It already implemented this policy from 2009-10, so we know it works.

    4. For sure we don't need to go all the way and convert the entire debt to satiate the demand for money, and what you suggest would be adequate. I'd add that if the major economies ever come to an agreement to stabilize exchange rates, we probably want two categories of government debt: these reserve balances, which are highest priority and part of the pool of exchange media that is subject to the rate stabilization agreement, and a second category of sovereign debt that carries default risk. This is to avoid the structural problem of the Bretton Woods system where US default on gold obligations ended the system for everyone else.

      Radical or not, the increase in size of reserve balances might come about as a byproduct of a systemic change like the one discussed by George Selgin in a recent proposal: "He proposes abolishing the current primary dealer system and expanding the Fed’s number of counterparties"

      I think this is similar to what I advocated in some earlier comments above

  9. We certainly want banks to compete for funding, and we hardly have that now with the stickiness of deposits as well as the dormancy of interbank lending. I think it would be great if loan originators competed for wholesale funding from money funds, themselves funded with deposits. This is closely connected with another of John Cochrane's proposals: namely banking based on 100% equity (which describes MMMFs perfectly).

    We need our system to complete a transition that was forced by the financial crisis and massive expansion of the Fed's balance sheet. We once had a system in which banks were constrained by required reserves. As part of Cochrane's proposal to reorganize US federal debt, he explains that we can use the new tool of interest on reserves to manage the price level, and we no longer need to constrain reserves artificially to attain policy goals. To bridge the gap between JP and George Selgin's concerns, I think the way is to convert all federal debt into reserves, but not to accept new deposits from private banks at the Fed.

  10. 1: Desired reserves: liquid bank assets, earmarked (i.e. kept on deposit at CB or as cash) for paying for withdrawals in excess of new deposits based on the bank's own risk assessment.

    2: Required reserves: minimum level of liquid assets required by the authority to be held by a bank, usually defined as a proportion of demand liabilities.

    3: Excess reserves: desired reserves in excess of required reserves.

    When will banks desire more reserves than required? 3a: If, for example, there are no required reserves such as in the Canadian case. 3b: Or, if holding reserves is lucrative, e.g. if the CB pays interest on reserves.

    Both 1 and 2 require either direct borrowing from the CB or an interbank market to function, thus imposing a systemic cost. JP's argument seems to be that that cost has no positive effect on the banking banking system and is purely artificial.

    George Selgin's arguments seem to be as follows:

    a) that (in light of deposit insurance and implied baylouts?) 1 does not function in the desired, free market way. Banks will economise on reserves to the extent that it becomes systemically problematic, which is why one needs an exogenous authority to define required reserves as in 2.

    b) that there are two ways in which 1 & 2 can function. By banks borrowing directly from the CB through the discount window or by first borrowing from each other. George Selgin's argument seems to be that the latter is better. I admit, I can't quite follow that argument.

    c) that 3b crowds out lending activity for business opportunities with risk adjusted returns between 0% and IO(E)R.

    There seems to be some consensus between the two that, in theory, there is a sweet spot between 2 and 3 where required reserves = desired reserves. Following George Selgin's point a), that theory will not hold in practice, though.

    The argument that I'm aware of in support of the view that required reserves as in 2. are unnecessary, is that stability of a bank is not a function of its reserve holdings but rather of its capital. Liquidity can / will be provided as long as there is solvency. The logical conclusion of that argument being that required reserves in excess of desired reserves are indeed an unneccessary cost imposed on the system. That says nothing about a market for desired reserves, though.

    Nor is it an argument for IOER, i.e. killing the whole market for reserves. The case for, or at least not against IOER, is that interest bearing reserves are a more direct alternative to riskless, government bonds. As for George Selgin's question about what the limits are to the amount of excess reserves in the system, I think my answer is thus: the amount (and type) of interest bearing assets the CB purchases from member banks in exchange for interest bearing deposits. Under IOER, banks are not being payed to borrow unlimited reserves from the CB, after all. They are voluntarily giving up one type of asset for another.

    So do interest bearing reserves crowd out investment? Not beyond the extent that the interest baering assets purchased by the CB didn't already do so.

    Should desired reserves come at a cost? Hmm...

    In conclusion, depending on whether JP's aim is to kill reserve scarcity altogether or only the market for reserves in excess of desired reserves (my understanding was the prior), I think I find myself somwhere in bewteen JP and George Selgin territory (not sure, whether that's a wet or dry place...). Not because I believe that having to manage liquidity is a virtue per se. More because, apart from secondary effects of CB balance sheet expansions, which I admittedly have limited understanding of, I don't find the case for IOER particularly compelling. It's a 'why?' as opposed to a 'why not?'.

    1. "When will banks desire more reserves than required?"

      I believe capital regulations are an important part of the answer: reserve balances are treated favorably by risk-weighting systems. When banks reach their capital constraint, parking deposits at the central bank might be the best option - especially if these reserves pay interest. Then it is up to the central bank to continue the expansion of credit by buying assets from banks and freeing up bank regulatory capital. Central banks can do this because they are free from capital regulation. But if money funds were better integrated into the payments system - with access to reserve balances - we could rely on them to fulfill our demand for deposits and shift them out of the regulated system. I'm pretty sure George Selgin would like that, based on utilitarian considerations and not just because of his long history of criticizing bank regulations that have bad side-effects!

    2. "When will banks desire more reserves than required? 3a: If, for example, there are no required reserves such as in the Canadian case. 3b: Or, if holding reserves is lucrative, e.g. if the CB pays interest on reserves."

      Oliver, U.S. banks desired more reserves than required even before the Fed implemented IOER in 2008 and made them lucrative. See here:

    3. Oliver, U.S. banks desired more reserves than required even before the Fed implemented IOER in 2008 and made them lucrative.
      Yes, right. Although I did qualify my comment with 'for example' and 'lucrative' is a relative term. But I'm not sure I added anything new with my comment one way or the other. & Thanks for the link.

      Not sure you'll agree, but on the whole I think one can put your question of whether 'banks should pay for settlement services' in the tent together with: are we dealing with one banking / currency system that happens to be divided into separate institutions but can, for all practical purposes, be treated as one bank (there it is again). Or should we be thinking (pretending?) in terms of competing, corporate entities that happen to be using the same unit of account. The Friedman rule seems to fit in with the prior as far as I can tell.

    4. That last paragraph may be going too far. But I do think the watchman function that George Selgin mentions cannot be relegated to the institutions themselves. In an architecture with no natural scarcity it must, logically come from independent authority.

    5. What would be the natural scarcity in a system of free banking where deposits could be redeemed in gold? Credit would still be limited to those who were worthy of it. Bank equity would still sell at equilibrium prices.

    6. Hmm, higher withdrawal rates would deplete gold reserves up the point where some or all institutions had to close their doors. In a system with LOLR and deposit insurance, that scenario doesn't exist. So you can't create a little subordinate market which pretends that it does. But yes, the upward limit to credit is credit worthy customers in any setup and bank equity is another buffer.

  11. Seeing how this discussion has gone, I'm more appreciative now of the original post and the comparison to water. Water is indispensable to us all, yet it is provided so efficiently that its price is effectively zero for most of us. Likewise, the non-pecuniary benefits of central bank reserves could be so freely available that their price would be zero. I don't see what's wrong with that. We want loan originators to grant credit when these loans are expected to make a profit. It would be great if we could simplify their operating system so that they could focus just on that.

    1. Yes, I think that's right. George's counterargument is that reducing the non-pecuniary benefits to zero somehow imposes an extra set of costs, say by displacing other lending and/or removing the "watchmen" effect imposed on banks by the overnight market for reserves. But (as of yet) he hasn't been able to convince me of this.

    2. I don't know the current system of reserves very well but I think the key is to prevent banks from being able to increase the size of consolidated government debt. Any new federal debt could only come into existence through spending on goods, services, and interest - not by passive acceptance of deposits as part of normal system functions. If demand for settlement media increases, their price would be bid up until equilibrium is reached. The central bank would presumably vary IOER (like a thermostat) to meet its nominal target. When the central bank closes its deposit facility, banks would not be able to lend to it anymore and would be forced to lend to the private sector - if at all.

      Surely there are other (better) ways to get the watchmen effect. If banks are bidding for settlement media, then what they have to offer in return is being evaluated every night. This valuation might be the price of shares issued by MMMFs overnight. Again I'm following John Cochrane here and thinking about his potential replacement of the current corridor system, though he has valuation of MMMF shares done with every transaction - via HFT - which is not how money funds do it now, probably for good reasons.

    3. J.P., as I haven't been able to nudge you, what say you about the arguments made in the article I cite concerning the Friedmam rule and IOER ( and to Craig Furfine's work on monitoring and the interbank market?

      I have a post coming out soon addressing Dudley's defense of the floor system citing this and other relevant literature. It also mentions how most central banks have chosen not to maintain or implement floor systems for the reasons I mention. The Fed, ECB, and Band of England's inclination to stick with their present systems may hae more to do to their reluctance to shrink their balance sheets (and the practical problems involved in doing so) than with any true advantages of a floor system over a (reasonably narrow) corridor system with a modest IOER rate.

  12. This is certainly a strange discussion!

    It seems to have no firm framework of theoretical structure!

    Now I say that money is like a gift certificate issued by a merchant (which promises merchandise from the merchant's store). I say that 'reserves' are those certificates despite now being held by a bank.

    Gift certificates (if properly managed) are always issued accompanied by accounting acknowledging their issue. 5000 certificates per bond record is common accounting.

    The practice of central banks borrowing-their-own-gift-certificates throws a curve to most people. Why would banks borrow gift certificates when (just moments ago) they also create them (at initial printing)? Simple--it reduces the supply of actual gift certificates floating around.

    That's my story--I'm sticking to it.

    What's the story behind the reserves you are discussing?

    1. Roger, let me catch you up. If your zero-interest gift certificates are used as reserves, banks may want to borrow/lend them between each other to satisfy their demand for overnight liquidity or statutory reserve requirements, one party paying the other a greater-than-zero overnight interest rate on gift certificates. All I am arguing is that for efficiency's sake, enough certificates should be created by the merchant so that this market is reduced to its smallest possible size. Put differently, there should be no overnight interest rate apart from the interest already offered by gift certificates (which in this case is 0%, but it could by 3 or 5 or 100%).

      IIRC, we've already discussed this somewhere. It's just the Friedman rule.

  13. JP, Thanks for the quick catch-up. We probably did discuss this---without conclusion. Conclusions are hard to reach in economics discussions!

    I think the fundamental problem (currently discussing) is gift certificate capture. The certificates have a way of concentrating into the ownership of only a few hands. This leaves geographical regions and ownership blocks short of sufficient certificates.

    IMHO, more certificates won't solve that problem(s). It requires taxes or other redistribution measures.

    I guess IOR applied in regions-experiencing-shortages could encourage owners to keep their certificates near home (which would only help if 'home' was in the region). OTOH, interest payments to certificate-holding regions would only exasperate the problem faced by certificate-short regions.

    Or maybe I am not understanding the exact problem needing solution.

  14. It seems that this discussion will become less theoretical and more important as the Fed reduces the size of its balance sheet over the coming 2-3 years and slowly eliminates the excess reserves liabilities on its balance sheet.

    I will be interested to see the interplay between the shrinking supply of excess reserves and how demand for those reserves changes as the FOMC gradually raises rates and the liquidity coverage ratio keeps demand high.