IOER and Banks' Demand for Reserves, Yet Again

ank lending, demand for reserves, IOER, net interest margin, reverse repos

In our recent American Banker opinion piece, Heritage's Norbert Michel and I argue that, if the Fed is really serious about shrinking its balance sheet, it had better quit paying interest on banks' excess reserves (IOER) as well. How come? Because the current, relatively high IOER rate  is contributing to a strong overall demand for excess reserves, while a shrunken Fed balance sheet will mean a reduced supply of reserves. Reducing the supply of reserves while doing nothing to reduce banks' demand for them is a recipe for demand-driven deflation, which is a monetary policy no-no.

Predictably (because it has happened every time I write on this topic) our article generated several comments to the effect that we didn't know what we were talking about, because banks couldn't possibly prefer the meager 100 basis points they can earn by holding reserves (or something less than that, if they are obliged to pay FDIC premiums) to the far greater amount they can earn by making loans.

The remarkable thing about these criticisms is that they all appear to deny that banks (or some banks, in any event) are in fact sitting on large amounts of excess reserves, and that they are, to that extent, settling for a return on those reserves of 100 basis points or less, instead of swapping  reserves for other assets.

"For a bunch of 'smart guys," our first commentator writes,

these fellows don't understand how banking works. 10 times out of 10 a bank would rather make a loan than have the funds parked at the Fed. Of course the loans have to be of the quality that the bank would expect the borrower to be able to repay the loan.

So far as the evidence up to October 2008 is concerned, our commentator is on solid ground, for until then banks did in fact prefer making loans to holding reserves 10 times out of 10. But that has manifestly not been the case since October 2008, which happens to be when the Fed started paying IOER. Since then, as the figure below, comparing commercial banks' loans and leases to their total deposits and Fed reserve balances, shows, the odds that a bank would rather make a loan than park funds at the Fed have been closer to 8 to 10:

Although our friend Chris ("r.c.") Whalen, a highly-regarded bank consultant, is at least aware that reserves now make up a substantial share of commercial banks' assets, he denies that this has anything to do with the fact that those reserves now yield a positive (if seemingly modest) return.

Whalen's brief comment consists of two parts. The first, declaring that "The Fed is not paying banks not to lend. It prices the rate for excess reserves and Fed funds at a margin designed to preserve balance," strikes me as nothing more than an exercise in empty semantics. Whatever "balance" the Fed may be trying to strike, the fact remains that it involves a substantial increase in banks' overall demand for excess reserves. And if paying 100 basis points instead of zero doesn't make reserves more desirable, and all the more so when rates are generally low, then it is time for us economists to toss-away everything we thought we knew about the workings of supply and demand.

The rest of Whalen's comment is more substantial. "Even if you ended paying interest on excess reserves," he observes, "the totals would not move because they are ultimately tied to a purchase of securities by the FOMC." More substantial, but still wrong. As I tried to explain in a previous Alt-M article, although the Fed's security purchases largely determine the total outstanding quantity of bank reserves (and currency), those purchases  don't determine the outstanding quantity of excess reserves, or the outstanding ratio of bank loans to deposits, both of which depend on what banks choose to do with reserves that come their way.

The point is perhaps best illustrated by looking at statistics from before 2008. Back then, banks hardly held any excess reserves; yet ongoing Fed security purchases (and sales) caused the total quantity of reserves to vary considerably, at least by pre-2008 standards:

Again, for emphasis: the size of the Fed's balance sheet determines the quantity of total, but not excess, reserves. If excess reserves increase along with total reserves, as they have tended to do since the fall of 2008, that's because banks have found it worthwhile to accumulate excess reserves, and not because they could not possibly get rid of them.

The last comment on our piece is by Wayne Abernathy, the ABA's Executive VP for Financial Institutions Policy and Regulatory Affairs. In full it reads,

A major problem with the authors' theory is the assumption that banks prefer to place money at the Fed rather than lend it out. In fact, banks would rather receive the 3.19% margin that they get on loans than the net 60 or 70 basis points that they get from the Fed. Loan demand, while growing, is not yet vigorous enough to absorb the flood of deposits that banks are still receiving. The banks' choice is place the excess deposits with the Fed or tell their depositors "no thank you."

In referring to "60 or 70" rather than 100 basis points as the net return on reserves, Abernathy evidently has domestic U.S. banks in mind, since U.S. branches and agencies of foreign banks, being exempt from FDIC charges, earn their 100 basis points free and clear. Pointing this out isn't nit-picking, because  foreign banks have been holding a very large share of all outstanding excess reserves, in part precisely because reserves yield more to them than to their domestic counterparts. But the more important point is that, so far as both these foreign banks and the (mostly very large) U.S. banks holding large amounts of excess reserves are concerned, holding Fed balances is in fact more profitable, at the margin, than lending the funds those balances represent.

It must then be the case that relevant opportunity cost for banks that have chosen to hold substantial excess reserves isn't 3.19%. So what is it? First of all, margins for the largest U.S. banks and foreign bank branches and agencies, which are the ones holding most of the reserves, are much lower than that for U.S. banks as whole. Although the FRED database doesn't supply separate net interest margin data for the very biggest U.S. banks (instead it gives the margin for banks of over $15 billion in assets, which is not high enough for the purpose), it does report the margin for New York banks, which is a better though still rough proxy.  Here's a chart comparing that measure to net interest margins for U.S. banks as a whole, and also to margins for banks in the Euro area, which are available in FRED only until 2014:

Evidently, if you are a New York bank, or a branch of a European bank, your idea of a decent net lending margin is, not 3.19%, as it might be for a "typical" U.S. bank, but something closer to 2% or (for the foreign banks) 1.5%.

In fact, many foreign banks found it profitable to acquire and retain excess dollar reserves for the sake of earning the even more minuscule spread between the risk-free IOER rate and lower effective Fed Funds and private repo rates. We know that, because they've been arbitraging that difference for some time. Foreign central banks, in the meantime, have been parking money at the Fed through its reverse-repo facility, which allows them to arbitrage the spread between what the facility pays and rates on short-term  T-bills. Before the Fed began paying banks to keep balances with it, these arbitrage opportunities simply didn't exist.

Why would any bank settle for a return on assets below its net interest margin?  In fact, so long as bank loans are heterogeneous, and banks can set different prices for loans of differing quality, any bank that didn't would fail to maximize its profits. Here it's worth keeping in mind that, whereas the demand schedule for bank reserves (that is, the Fed's demand for a bank's callable loans to it) is, in effect, a horizontal line at whatever rate the Fed is paying, the  demand schedule for loans slopes downward. Profit-maximizing banks will expand their loan portfolios to the point where the net marginal return on loans, which is necessarily below the net interest margin (which is a measure of the net return on banks' overall loan portfolios) equals the IOER rate. Beyond that, reserves dominate loans. In equilibrium, in other words, parking another dollar at the Fed pays more than lending it does. Were IOER reduced to zero again, on the other hand, banks would once again find lending more profitable than reserve-hoarding, and they would continue to make loans until the net margin on them (the marginal net margin, that is!)  itself approached zero.

In case it helps, here is a picture of what I just said:

In the picture, the blue line is the (downward-sloping) demand schedule for bank loans, while the orange and grey lines are the Net Interest Margin for all bank loans (which, with deposit rates near zero is approximately equal to the gross interest return) and the IOER rate, respectively.* The picture assumes a given level of total bank deposits, here set equal to $10 trillion. The vertical red line shows equilibrium quantity of bank loans with IOER=1, while the vertical green line shows the equilibrium quantity with IOER=0.  The numbers are, of course, only meant to be suggestive.  And the IOER = 0 equilibrium shown here assumes that the Fed "mops up" the excess reserves that banks no longer wish to hold, so as to prevent any further increase in total bank deposits. Otherwise the reduced IOER rate would result in a substantial  increase in total bank deposits, along with a correspond, further increase in bank lending.

For those who continue nonetheless to doubt that the IOER rate has much bearing on banks' demand for excess reserves, I offer, as a final exhibit, and without commentary, one last chart, this time comparing the difference between the IOER rate and the LIBOR rate, which I treat as a measure of the relative yield on reserves, to the overall ratio of reserves to commercial bank deposits:


*Strictly speaking, the loan "demand schedule" is actually a marginal revenue schedule. The two are the same if either (1) banks are perfect price discriminators or (2) the loan market is perfectly competitive.  Note that perfect competition still allows for differential pricing of heterogeneous loans, and that the presence of some riskier loans means that the overall NIM will exceed the marginal lending rate. The only case in which the marginal rate will equal the overall net interest margin is that of perfect competition with homogeneous loans.


  1. Good discussion. IOR is what is driving and distorting Fed monetary policy. Normalization requires that the Fed also unwind IOR along with removal of excess reserves. The greater problem is that global central banks have adopted the same QE monetary policy, or worse. The ECB and BOJ are still monetizing $200b/mo. Their balance sheets are worse than the Fed's. China's currency is linked to the dollar, so the PBOC will import as policy the Fed's normalization. I don't think the Fed has any real idea of how to get out of the global monetary mess they created. Funds rate hikes are "tightening" theater while the Fed's balance sheet remains unchanged at $4.4t.

    Through IOR, regulatory control, and a flatlined economy the Fed has gotten away with massive excess reserves not causing runaway inflation (there is inflation, but not runaway)–so far. For these reasons depository institutions, including the U.S. subsidiaries of foreign banks, are willing to accept IOR over lending–as the author illustrates.

    Normalization really requires the Fed to squeeze the toothpaste back into the tube. This is the Fed's problem. The one thing the Fed can't do is lose control of CB manipulated interest rates. That's why the Fed is gingerly, tippy toeing with extreme dread into the normalization water.

    I wrote about the normalization, excess reserve, IOR problem here.

    1. Thanks, Michael, for your supportive comments and the link to your own post, which I will read shortly.

      You and I are in what appears to be a small minority on this topic, for, as Chris Whalen has reminded me in a tweet this morning, not a single banker he knows thinks that IOER has any bearing on bank lending.

      1. Would we have had the great inflation of the 70s if the Fed paid IOR then? It is important going forward to examine the economic differences between the 70s and today.

        The Fed in the 70s was not creating a liquidity flood due to Great Depression academic theory in response to a financial crisis. Friedman’s monetarism was supposed to regulate base money supply consistent with long-term growth rates. The Fed post Bretton Woods was targeting monetarism aggregates. There was no need for IOR in the 70s according to monetarism theory because theoretically there was no excess base money creation. Another academic theory that results in economic regression when put in practice in the real world. Friedman eventually issued a mea culpa in the FT late in his life for the disastrous results of his quantity theory of money, "I'm not sure I would as of today push it as hard as I once did."

        Post 2008, IOR allowed the Fed to enact its liquidity flood without the great inflation of the 70s. The market initially believed a repeat of 70s inflation was in the offering with gold bid up to $1900/oz. I admit that for a very long time I did not think the Fed would get away with it. (They still haven't gotten away with it as long as excess reserves remain.) I assumed at the beginning of QE that banks would choose to lend over holding excess reserves and receiving a paltry .25 percent return from IOR. A couple of factors are now apparent. U.S. subsidiaries of foreign banks hold a significant portion of excess reserves, probably in response to conditions in the EU. Large U.S. banks hold the majority of the remaining. Congress gave the Fed new regulatory power after the 2008 crisis. One can assume that after the Fed bailed out the TBTFs, they are holding excess reserves in compliance with the Fed's regulatory wishes.

        Many believe today is a repeat of the 70s. It is only a matter of cutting taxes and the economy will boom again, the same as the 80s. The difference is that market forces set interest rates in the 70s. Today the Fed and CBs centrally command the price of interest. This unwind and new monetary stability will have to occur for a return to fundamental growth. IOR is at the center of it.

  2. Every time the Fed issues a dollar of reserves, the fed gets a dollar's worth of bonds or other assets, and those assets are obviously enough to buy back all the reserves the Fed has issued. So just as a corporation that expands both sides of its balance sheet equally will see no change it its stock price, the Fed can expand both sides of its balance sheet by equal amounts without affecting the value of the dollar.

    1. Whenever the Fed issues reserves, the interest charges change and if the Fed thinks nominal assets back the reserves, the Fed is mistaken. Assets do not balance reserves until depositors react to the issuance of a new loan, and after depositors react, then the balancing assets are mostly correct. The whole idea of a loan is to make a change in the state of the economy, and cause changes in interest, and if depositors are not allowed to react, pricing of assets goes screwy.

      1. When the Fed issues a dollar of reserves (which should be called federal funds, actually), it gets a dollar's worth of assets in exchange, so the value of the dollar is unaffected. If a private bank then lends a dollar to some customer, then that bank gets a $1 IOU in exchange for the lent dollar. This leaves the assets (and net worth) of the private bank unaffected. The net worth of the borrower of that dollar is likewise unaffected. All we get is offsetting changes in balance sheets, with no change in the amount of backing per dollar, and therefore no change in the value of the dollar.

        The Law of Reflux should be mentioned at this point, otherwise we are "reasoning from a loan change". If we simply suppose that a mint issues more coins and lends them, then the LOR tells us that nothing important will change, since other money-issuers will react to the new coins. The same is true of a bank issuing new dollars on loan.

        1. If the Fed issues a dollar in reserves the value of the dollar has definitely changed. The issuing of reserves is done precisely because some member banks detect a potential productivity improvement, with knowledge the other member banks do not have. Otherwise, why take out a loan? The point of a loan is that past interest charges are less than possibly investment gains. That means the borrower is actually intending to snooker the other member banks with inside information and if correct, he gains as the currency value changes. It it the mechanism used to price inside information and monetize it.

          The whole point of currency banking is to spread the currency risk fairly to all participants, and the currency risk is real in the sense that the currency banker is a market maker and has to gain or lose currency to do its job.

          So the idea is first, a member banks takes a loan, second all other member banks adjust deposits to cover the loan as they wish. It they do not cover the whole loan, then the currency banker makes the market, and absorbs the loan risk.

          1. "he gains as the currency value changes"
            You presume the correctness of your own conclusion. Only on quantity theory principles does a newly-lent dollar cause existing dollars to lose value due to the money supply outrunning the quantity of goods.

            On backing theory principles, the new dollar is backed by a dollar's worth of assets acquired by the issuing bank, and the value of the dollar is unchanged. By analogy, GM can issue a new share of stock, but it simultaneously gains new assets equal in value to the new share, so share price is unaffected, even thought the quantity of GM shares rose relative to the economy's production of goods.

            When you say "cover the loan", I presume you mean that as one bank issues a new dollar on loan, other banks reduce their lending by $1. But on backing theory principles this is irrelevant. The issuing bank backed its new dollar with a dollar's worth of new assets, so its dollars will hold their value regardless of what other private banks do. The other banks, for their part, will also back their new dollars, so their dollars will hold their value as well.

          2. The trade is not statistically stationary, meaning as the bets pile in the odds change. Hence, when a bank issues a loan, the odds have changed, it is the depositors turn to respond.
            Consider the depositors and lenders as two queues. The bank issues a loan, then immediately becomes first in line to issue the corresponding deposit. That process adds tradebook uncertainty, the borrower gets a double peek and trade on the trade book while it is effectively hidden from the depositors. The result is that currency risk is transferred to the central bank, which then transfers it to government.

          3. 1) There is no currency risk as long as each $1 lent into being is matched by $1 of new bank assets. The loan will not cause the value of the dollar to go up or down.
            2) A bank ends up owing $1 to one person and being owed $1 by another, so even if the bank's action did change the value of the dollar (and I say it wouldn't), a change in the value of the dollar would not change the bank's wealth.

  3. How about the role of post-crisis Fed-imposed liquidity and leverage requirements like the LCR? Given minimum safe asset holdings under these regulations, it would seem the opportunity cost for excess reserves is not determined by normal bank loans but by other compliant safe assets like Treasuries. Thus even if you eliminate IOER, you probably still have a safe asset shortage (due to the regulator's tight definition of compliant assets).

    1. Eric, there's no doubt that the LCR requirements have increased banks' demand for safe liquid assets. However, in the absence of IOER, that demand would almost certainly be fulfilled by Treasuries rather than by (zero-interest) reserves. As for Treasuries being in short supply–well, the Fed has something to do with that as well!

  4. George, you write:

    "…the Fed must shrink its balance sheet and stop paying banks not to lend."

    "…the Fed needs to combine its plan for shrinking the balance sheet with one for phasing out interest on excess reserves."

    Reading the original post, it's not entirely apparent what you are arguing for.

    On the one hand, it could be interpreted that you're arguing that any shrinking of the balance sheet needs to be twinned with a reduction of IOR to 0%. (And maybe a few years later, when condition merit, IOR will be brought back up.)

    Or you're arguing for abolishing IOR altogether. I'm pretty sure this is the point you're meaning to make. If it phased out IOR, don't you think the Fed will be giving up an important tool that gets it closer to implementing the Friedman rule?

    1. JP, thanks for asking. I'm for eliminating IOER altogether. I have no problem with the Fed paying interest on required reserves, which is all it has to do for the sake of minimizing the reserve 'tax." IOER makes no sense for that purpose, because it means encouraging banks to hold reserves they wouldn't hold otherwise.

      Better yet, reserve requirements should be abolished.

      1. In Canada we don't have reserve requirements, but banks still keep small amount on deposit overnight. The Bank of Canada does pay interest those balance. Should it stop paying interest? In Canada banks don't pay a reserve tax, but they do pay a monopoly penalty of sort… i.e. they can't turn to a private alternative for clearing and settlement. Doesn't IOR minimize this?

        1. In principle JP, where there are no legal reserve requirements as in Canada, it is beneficial to pay IOR, so long as it is at or slightly below the equivalent market rate.

          In fact that was the intent of the U.S. law. However, the Fed has simply ignored the law, as I pointed out in the WSJ some time ago ( That's why I favor abolishing IOER altogether, and allowing only interest on required reserves.

          Perhaps the Bank of Canada is more trustworthy.

          1. "In principle JP, where there are no legal reserve requirements as in
            Canada, it is beneficial to pay IOR, so long as it is at or slightly
            below the equivalent market rate."

            Makes sense to me.

            "Perhaps the Bank of Canada is more trustworthy."

            In Canada we don't have the same GSE problem that U.S. has… i.e. we don't have government run corporations that can't receive interest on central bank balances. So the BoC's deposit rate, unlike the Fed's IOR, sets an impervious lower bound to interest rates… there is pretty much no way for the BoC to pay more than the equivalent market rate. This seem to be more a matter of market microstructure than the central bank being trustworthy or not.

          2. Perhaps. But in any event, I've seen and heard enough about how Fed officials operate, including a lot of shocking revelations from persons who worked in the system, to regard them as untrustworthy nonetheless!

  5. George, I think looking at this from an investment/business perspective can further clarify why you are correct on this big issue. Let's look at BB&T's financial performance as a proxy for a typical bank (chosen at random, using Bloomberg as my data source). 2016 ROE was 8.8%. Its 1Q17 operating leverage (total assets/total equity) is 7.35x. Thus, earning 100 bps of IOER equates to a 7.35% ROE – not a bad investment proposition for a risk-free (free from credit and duration risk)/capital charge-free investment with immediate liquidity, and (as you pointed out) at higher rates found in other short term securities (i.e. LIBOR and T-bills). This is even more powerful when thought of in marginal terms (i.e. as a marginal dollar invested). Any analysis considering only the possible spread is cursory, and ignores the other considerations inherent in bank lending (like credit risk, ALM practices, capital charges, liquidity, etc.).

    1. Seth, thanks for sharing your alternative and insightful perspective. It is always helpful to look at things from very different angles; and it's reassuring to see how in this case two of those different views reveal a similar conclusion.

  6. The fascinating thing about this discussion is that as we talk about the Fed attracting deposits away from lending via IOER, the competition for deposits among regional and community banks is intense and growing. These smaller banks don't really play in the world of institutional funding or securities, and historically tended to be net sellers of funds to the larger banks. I cannot see any real connection between how the big banks manage their liquidity and securities portfolios (mostly institutional) and the smaller banks, which focus on retail core deposits and then FHLB advances for liquidity.

    The large proportion of foreign banks in the IOER equation seems very relevant. How does it impact a discussion of lending in the US? Foreign banks tend to focus more on investing activities in the US than lending. The chart on LIBOR vs IOER is quite interesting in this regard, but is the movement coincidental? If not, perhaps the BIG impact of the Fed's payment of IOER was to tighten the already thin market for Eurodollar funding as the chart suggests. But this would also suggest that the foreign banks were the predominant sellers of securities to the Fed during QE. To suggest that the increase in excess reserves by foreign banks was an active asset allocation choice determined primarily by the yield on IOER is interesting given the rising bid for LIBOR. There may be a larger story here to support George's thesis.

    As per my email to George, I am going to do some work on the rest of the story, namely how leverage limits and other regulatory factors impact lending. If shifting assets into excess reserves is really an active decision by a bank (apart from a securities transaction with the Fed) where the risk-free rate offered is more attractive than a private loan, then we need to build out this part of the analysis to complete the picture. There certainly are times where banks reach the hard leverage limits on risk assets and would then prefer to hold excess reserves or some other risk free asset. The supply demand equation that applies here is more complex than merely a question of price. But IF a domestic bank really has an active choice of whether to or not to hold excess reserves (as opposed to merely being an incident to a securities sale to the Fed), then the IOER less the relevant FDIC insurance premium seems to be the right yardstick for comparison.

    Regardless of how we view the relative attractiveness of IOER vs other asset allocation choices by banks, the fact of QE clearly seems to be the driver for the accumulation of the very large excess reserve balances. If the Fed were to end reinvestment of securities and even sell assets, the excess reserve figure would presumably fall as principal repayments and/or outright sales were credited to bank reserve accounts. Referring back to the work of our friend Bob Eisenbeis, it is important to remember that the Fed is the alter ego on the Treasury in financial terms. QE is a subsidy for the Treasury equal to the coupon on the Treasury debt less the cost of the IOER. In that sense, the Fed paying IOER is no different than the Treasury paying interest on T-Bills. But obviously the 1% on IOER is much higher than other risk-free alternatives as George points out.


  7. George, another great piece of analysis and you are not alone on this issue.

    Remember, Greenspan and Richard Fisher (Dallas Fed) have made the same argument as you have detailed above, in a number of conferences and appearances on tv. I would add a comment and a question:

    1) many of the banking commentators don't seem to be factoring into bank's decision making process the marginal net interest margin that you have clearly demonstrated in the above charts. In addition to this element is that return on equity calculation, which is driven more so by the capital requirements for the type of loans a bank makes. Holding excess reserves and having to set aside 0.0 capital against the balance is more attractive to banks than lending out those balances to create a senile commercial loan, which requires 50% capital set aside by the bank. Banks would rather take the 70-100 basis points (as evidenced by their very self evident decision to park the reserves at the Fed in excess!)

    My friend who is working for Westpac (one our Australia's big 4 banks) in New York in their international wholesale finance division has told me as much – Australian banks would much rather park their excess reserves (and have been doing this since 2008) in the US Federal Reserve system than at our own Reserve Bank of Australia, where they receive 0% interest. Currently Westpac (and he is the one managing this process) is parking ALL excess reserves in the US. Meanwhile, commercial loans by the Austrlian big 4 have only now surpassed the 2008 nominal levels. IOER is an incredibly tight monetary policy and it is impacting overseas banks just as much as US banks. By the way, Westpac has the second highest return on equity for any major bank in the world, so it's interesting to see banking commentators think IOER has no impact on lending.

    2) I haven't seen a good piece of analysis on how the Federal Reserve actually "mops" up these excess reserves while reducing the IOER rate. If they can't mop these balance up in an orderly manner, the balances will get out into the real economy and the money multiplier will begin to move, putting pressure on the 10 year note as inflation expectations begin to move.

    Do you have an idea as to how they could structurally mop up the excess reserves, lower the IOER and still have short term rates rising?

    If they continue to raise IOER so as to not shrink their balance sheet, wouldn't this effectively be deflationary as the net interest margin will remain stable but the risk less c.150 basis points a bank can earn on excess reserves pull more deposits out of the loan market?

    Great analysis as always.

    1. Thanks, Rob, for your thoughtful, informative and encouraging comments. This has been a hard one to get across, but so long as I have some fight left in me I'm not backing down on it!

      Concerning your questions, see this other recent op-ed of mine in The Hill: There I observe that raising IOER is indeed deflationary, other things equal, while lowering it loosens credit. But Fed asset sales also tighten. So the idea is to combine those asset sales (or passive asset shrinkage) with gradual reduction of the IOER/LIBOR gap, so as to keep policy on track. In the process the Fed should treat LIBOR or other overnight rates besides the FFR as it's operating target, returning again to FFR targeting once that market revives, as it will when reserves become sufficiently scarce again.

  8. "The picture assumes a given level of total bank deposits, here set equal to $10 trillion. "

    I don't understand this point. If "loans create deposits," like the BoE and the Bundesbank say, then how can you possibly have a graph with varying amounts of lending but a constant amount of deposits? More lending should equal more deposits created, and less lending should equal less deposits created.

    1. Not so, Sam: deposits can be backed by loans, or by other banks assets, including Fed balances. In fact, as I show in the chart above, loans made up almost 100% of deposits prior to the crisis, whereas they have made up only 80% of deposits in recent years. The supply-demand diagram shows the value of loans assuming a fixed total value of deposits, and (as I explain in the footnote) relies on the implicit assumption that the Fed adjusts total reserves as banks demand for excess reserves changes to preserve that fixed total. In the equilibrium with IOER = 100 basis points, banks make $8 trillion in loans, and hold $2 trillion in excess reserves. In the equilibrium with IOER = 0, they make $10 trillion in loans, and hold no excess reserves.

      1. "…relies on the implicit assumption that the Fed adjusts total reserves as banks demand for excess reserves changes to preserve that fixed total."

        Ok but…that doesn't happen at all. That was the whole point of paying IOER, so that the Fed would have a "floor" and therefore wouldn't need to adjust the quantity of reserves as much. They could just leave excess reserves in the banking system without needing to worry about the Fed or Treasury selling bonds to drain them, and still have the interest rate hit the target. (Although this has proven to be a leaky floor, necessitating additional measures).

        Also, I'm further confused by when you say "banks make $8 Trillion in loans and hold $2 trillion in excess reserves" and "make $10 trillion in loans and hold no excess reserves." You seem to be implying that the amount of reserves that convert from excess reserves to required reserves when the banks create deposits are numerically equal to the size of the loans. Or am I interpreting you wrong?

    2. Sam, you can say, if you prefer, that there are no more deposits and loans in aggregate, only a change in the borrower. Whereas when IOER = 1 the banks lend $2 trillion to the Fed, and when the IOER = 0 they lend the $2 trillion to the general public.

      1. Either I'm misreading you, or you've got the accounting seriously wrong. Banks mostly do not lend to the Fed. If anything, they are often in debt to the Fed.

        1. This isn't correct. Banks lend more these days to the Fed, by choosing to hold excess reserve balances with it (for which

          it pays a relatively attractive return) that hey owe to it, In fact, the discount window hasn't been active for some time, so they owe it practically nothing.

          As with all matters economic, one must think in terms of real, and relative, magnitudes. The real demand for reserves, beyond mandatory requirements, measures banks' voluntary lending to the Fed.

          1. Hello George, two points.

            First of all, interesting of you to mention the discount window without also mentioning Fed overdrafts. The Fed also lends to banks through automatic overdrafts, and these are usually much larger than the discount window, except perhaps during crises. Because of all the excess liquidity, both are indeed down substantially, but overdrafts (at a peak of around $5 Billion daily) have not gone down to nothing in the way that the discount window has (which averages around $30 Million outstanding). Data here:

            But furthermore, referring to banks holding reserves as "lending to the Fed" makes no sense to me. Reserves don't mature ever, and they're only redeemable into other other Fed liabilities (currency, in particular). It doesn't help the Fed at all for the banks to hold reserves in the same way that it helps a debtor for a creditor to hold his promissory note (the Fed's ability to spend or finance purchases is unchanged regardless of the level of reserves). So, if this gets to be called "lending to the Fed," then, arguably, every American who has a $1 bill in their wallet is also "lending to the US government." The only relevant difference between holding reserves and holding currency is that one of them pays interest and the other doesn't. (And there's no real reason that they both couldn't pay interest.)

            Thanks as always!

  9. I always enjoy George Selgin's writings. I may not always agree, but I always learn, and Selgin is one of the few who approaches monetary policy without an ulterior motive or agenda.

    BTW the Bank of Japan pays interest on excess reserves–negative interest, that is.

  10. I think you need a better description than just saying lending.

    From the other post:

    "Well, banking would indeed be a marvelous business if it worked as our expert at Forbes assumes. Alas, it is not so marvelous as that. For despite what Ms. Coppola claims, banks do, in effect, lend "reserves" to customers no less than to other banks. The lending of "reserves" is more apparent in the overnight market simply because it is reserves per se that borrowers in the market are after, for they need extra reserves to avoid shortfalls that would otherwise subject them to penalties, or to what amounts to the same thing: a visit to the Fed's discount window.

    If, on the other hand, a businessman borrows $500,000 from a bank, it isn't cash itself that the businessman wants, but other things that can be got for the cash. But as soon as the proceeds of the loan, originally received as a deposit balance, are drawn upon for the sake of acquiring these other things, the withdrawals, whether by check or draft, lead quickly to redeposits in other banks, and thence to a $500,000 adjustment to the pattern of interbank clearings and settlements at the expense of the lending bank and in favor of rival institutions compared to what would have been the case had the loan not been made."

    Lending of what I call central bank reserves happens in the fed funds market. A bank lends demand deposits to customers. The customers don't borrow central bank reserves. That I believe is what people mean when they say banks don't lend reserves to customers.

    A better model would be to assume there is only one commercial bank. Get that model correct and then add other commercial banks.

    The loan process and the payment process should be separated.

    1. All I can say is that you obviously cannot have understood what the paragraphs by me that you reproduce actually say, for they explicitly acknowledge and address every point you raise! You simply don't take into account what bank borrowers _do_ with deposit credited to them, which is to _spend_ them, in relatively short order–in bankers' lingo, lent deposits are for the most part not "core" deposits. The spending results, other things equal, in an equivalent increase in the lending banks' clearing losses, which are settled with reserves. Although banks can and do cover such losses by borrowing in the fed funds market, their willingness to do so is limited, and in full equilibrium they do not wish to further increase their persistent borrowing on that market. Hence, although at first L and D go up, eventually D and R go down by the same amount. The overall, final balance sheet adjustment is therefore L up and R down, by the same amounts. Hence the bank in effect lends its reserves. QED.

      Tracing the transaction further back, on the other hand, we can imagine a prior increase in ("core") deposits that supplied the bank with additional cash resources in the first place, thereby allowing it to lend more. We thus end up with a complete picture of a bank acting as a funds intermediary.

      And no, assuming a monopoly doesn't help. On the contrary: that is the exceptional case in which "lending deposits" isn't equivalent to lending (parting with) reserves. Analyzing a single bank in a competitive system as if it were a monopoly is a very serious analytical error.

      1. "Analyzing a single bank in a competitive system as if it were a monopoly is a very serious analytical error."

        That is not what I said/meant.

        "And no, assuming a monopoly doesn't help. On the contrary: that is the exceptional case in which "lending deposits" isn't equivalent to lending (parting with) reserves."

        Go on. I need some more detail there.

        "The spending results, other things equal, in an equivalent increase in the lending banks' clearing losses, which are settled with reserves."

        Let me try the single commercial bank point/separate loan from payment point this way. Apple and I "bank" at the same commercial bank. I borrow $500 of demand deposits to buy an iPhone. There is an asset swap of demand deposits and the iPhone. Apple saves the demand deposits at the same commercial bank.

        Does that help?

        1. Yes, it helps me to see where you are coming from, but not to agree with you. The usual assumption is that the checks a bank's borrowers write end up deposited, not in the same bank, but in some rival bank. Given that there are 6800 banks in the U.S., that's a lot more reasonable than the assumption that bank borrowers write checks exclusively or mainly to people who happen to be at the same bank!

          Moreover (though the explanation is more involved), even in your case, so long as the recipient of funds is managing his or her balance, instead of allowing it to grow willy-nilly, the long-run tendency is the same as if there were no same-back transfers of the sort you describe. For details, see C.A. Phillips' Bank Credit or Alex Macleod's Principles of Financial Intermediation. The latter is a favorite of mine, because he even considers the case in which banks issue their own notes!

          1. "The usual assumption is that the checks a bank's borrowers write end up deposited, not in the same bank, but in some rival bank."

            You were writing about aggregates, why banks don't lend or lend less. If deposits end up in rival bank then rival's deposits end up in your bank. If there is a deposit run from one bank then I am sure you were not talking about that here.

            But this is not even important here, banks have expenses associated with managing reserves, I agree. They lend when they see a profitable lending opportunity, they make money on spread. The causation is opposite to the textbook explanation from reserves to loans. It is from bank loans to reserves. That is how every Post-Keynesian like Coppola is thinking. They are not quantity theorists.

            Fed is a monopoly supplier of reserves, monopolist can fix price or quantity but not both at the same time. Most CB-s have chosen to fix price and let the quantity float.

            Your logic about IOER reducing bank lending disappears when total quantity of reserves is not fixed

          2. "Most CB-s have chosen to fix price and let the quantity float." I know this is what the MMT's believe; but it is a superficial perspective, even w.r.t. the pre-crisis Fed regime to which it appears most applicable.

            In that regime, the Fed sets a funds rate target, and supplies reserves to maintain it. So far, so good for the "horizontalist" view. But the Fed's rate target is conditional: it remains in effect only so long as it is regarded as consistent with the Fed's ultimate objectives, including its inflation target. For that reason, the long-run reserve supply schedule cannot regarded as horizontal.

            For example, suppose the Fed targets inflation, and suppose an initial steady-state with ffr(1) = ffr* and inflation (in) = in*. Next imagine that an outward shift in credit demand, driven by favorable animal spirits or whatever, encourages the banks to collectively expand their lending, provided they can secure additional, precautionary clearing balances. If the Fed accommodates the banks, the eventual result will be in>in*; which will lead it to raise the ffr. Alternatively, the Fed might anticipate the need to adjust its policy rate to head-off the inflation. Either way, the effect is just as if the reserve supply schedule were horizontal rather than vertical.

            The MMT perspective would be valid were the Fed indifferent to the ultimate affects of its maintaining a fixed reserve supply schedule, that is, were it merely a passive supplier of reserves. But the Fed is clearly not that: it routinely adjusts ffr* in response to changing conditions–not always ideally, to be sure! Only a central bank indifferent to the course of inflation and other nominal variables would allow the stock of reserves to grow willy-nilly in response to growth in banks' demand for liquidity. Such cases do exist: hyperinflations certainly involve something like rigidly horizontal CB reserve supply schedules! But they are the exception rather than the rule.

          3. "The usual assumption is that the checks a bank's borrowers write end up deposited, not in the same bank, but in some rival bank. Given that there are 6800 banks in the U.S., that's a lot more reasonable than the assumption that bank borrowers write checks exclusively or mainly to people who happen to be at the same bank!"

            Sorry for the late reply.

            I want to change some of the usual assumptions to show how the system works.

            First, here are most of the steps involved:

            1) some type of "lending" by a commercial bank

            2) the lent demand deposits being spent

            3) If necessary, the demand deposits being "redeposited" in another commercial bank

            4) If necessary, the first commercial bank borrowing central bank reserves in the fed funds market

            I want to change the usual assumptions so that 3 and 4 do not happen.

            Assume JP Morgan is the only commercial bank and has only one branch. Every private entity banks online with a computer there.

            That eliminates lending in the fed funds market and eliminates "redeposited" demand deposits needing to be "payment cleared". 3 and 4 do not happen.

            Now run the model.

            "Moreover (though the explanation is more involved), even in your case, so long as the recipient of funds is managing his or her balance, instead of allowing it to grow willy-nilly,"

            I assume you mean demand deposits in a checking account. If so, that is why I picked Apple. For the most part, it does allow its checking account balance to grow "willy-nilly".

        2. He mixes up a bank making loans and losing deposits for some reason, he thinks those two are necessarily associated. So next he argues that losing deposits is associated with expenses for banks (that is true generally) and says that why should banks want that if there is IOER. I know It doesn't make any sense. even if it is so, what does IOER have to do with It? He thinks that quantity of total reserves is fixed
          he doesn't understand endogenous money.

          1. You don't seem to understand that in the case at hand, the deposits I am referring to consist of the credit balances that the bank creates in making a loan–that is, balances credited to a borrowers account! Ut is those deposits that tend quickly to be spent after the loan has been extended, for the simple reason that most people don't take out loans just for the sake of having $$$ sit around! Indeed, with lines of credit, banks don't book the loans until the credit lines are drawn upon, and only to the extent that they are! So, lending and spending of lent deposits go hand-in-hand, and with them goes as well the reserve transfer that, other things equal, must accompany the extra spending.

            As for the rest, you make such a jumble of my arguments that it isn't worth my time to try to untangle what I have said from what you say I have said. I ask my readers to kindly refer to my own statements rather than your understanding of them, which I myself can't understand.

            Finally, as for "endogenous money" of the MMT sort (if that's what you mean), I understand it well enough to think very little of it and the general beliefs that it comes with.

      2. "The spending results, other things equal, in an equivalent increase in the lending banks' clearing losses, which are settled with reserves."

        Here again you put banks making loans and the deposits (that were created by those loans) moving from one bank to another in the same pot. Atracting deposits and making loans are two different topics. I don't know why you mix those two up and assume that a commercial bank that is expanding its balance sheet is losing deposits? Then you say "clearing losses", you mean the bank where the deposit originated losing the deposit to another bank and having to compensate that to the other bank with reserve balances. Then you say that this bank would rather not do that because It can earn interest on ecxess reserves. This does not make any sense.

        You write: "Although banks can and do cover such losses by borrowing in the fed funds market, their willingness to do so is limited"

        Not more so because IOER. You tend to think that Fed controls the quantity of total reserves, that is fixed in your mind. So you think that if banks make more loans the required reserves quantity goes up and banks don't want that to happen.

        Banks know what the overnight rate is, Fed has announced that. Fed provides enough quantity to keep the overnight rate target, so for banks it is about the price of the reserves, not quantity. Fed controls quantity indirectly through overnight rate.

  11. George and Chris (Whalen),

    "The rest of Whalen's comment is more substantial. "Even if you ended paying interest on excess reserves," he observes, "the totals would not move because they are ultimately tied to a purchase of securities by the FOMC." More substantial, but still wrong. As I tried to explain in a previous Alt-M article, although the Fed's security purchases largely determine the total outstanding quantity of bank reserves (and currency), those purchases don't determine the outstanding quantity of excess reserves . . . "

    Strictly speaking that's true. However, given the tiny role played by required reserves, for all practical purposes I think Chris is right. Individual banks can affect the level of their own excess reserves through lending and security dealing, but the banking system as a whole can't. What one loses (or gains), the rest of necessity gain (or lose).

    It seems to me the primary role of IOER is simply to affect banks' comfort levels with holding these systemically unavoidable excess reserves. In the absence of a (perceived) competitive rate, there would be a powerful incentive for individual banks to try to get rid of their excess reserves by expanding their balance sheets through security purchases and/or lending. Since at a system level all these efforts must come to naught (more or less), it's not hard to imagine the Fed losing control. Indeed it still may, since as many here have noted the game is far from over.

    1. "Individual banks can affect the level of their own excess reserves
      through lending and security dealing, but the banking system as a whole
      can't. What one loses (or gains), the rest of necessity gain (or lose)."

      Basho, repeated assertions aren't an argument; and what you claim here is simply what I take pains to refute in the sentences that follow the ones you quote. If you want to counter those arguments, well and good; I am happy to entertain your counterarguments. But you don't do that. You simply ignore what I say, and so are unable to appreciate that because individual banks' continued efforts to get rid of unwanted excess reserves necessarily contribute to growth in system deposits, these do in effect serve, ultimately, to reduce excess reserves. If the growth in reserves is sufficiently large, the process, if allowed to play out leads to hyperinflation. But reserve ratios decline, eventually to however minuscule a level banks desire. This is well-tested theory.

      But now I'm repeating myself, which is pointless, so I'll leave it at that.

  12. The CFPB and GSEs seem to have imposed liabilities on some loans, some of which are vague, which probably act as a non-cash expense to lending, especially mortgage lending, so that cash based spreads are probably somewhat inflated. In other words, at the margin, commercial banks need a higher net margin on mortgages to justify marginal lending. This could explain having some interest margin even while retaining excess reserves.

Comments are closed.