Interest On Excess Reserves isn’t Keeping Inflation Low

Noah Smith has some good observations about a recent Martin Feldstein piece in which Feldstein claims that interest on excess reserves is keeping inflation low because it’s enticing banks not to lend out their reserves or make investments.  Regulars know where this is headed.   Sigh.

Noah’s explanation is pretty good.  But I think he makes it too complex.  You see, banks never lend reserves.  The amount of reserve deposits in the Fed system are determined exogenously by the Fed.  When a bank wants to make a new loan to a creditworthy customer it ensures it has the capital and simply makes the loan.  If it needs reserves then it can find them after the fact.

Now, Post-Keynesians know that loans create deposits.  That is, a bank can create new money simply by creating new loans that lead to new deposits.  It doesn’t necessarily need reserves to “multiply”.  It simply needs capital and creditworthy customers walking in its doors.  The problem with today’s environment is not that the banks are holding a huge amount of reserves and unwilling to lend or being paid to hold them.  The problem is that there aren’t enough creditworthy customers walking in the doors to demand loans.  If there were we wouldn’t see data sets like this one at levels that are reminiscent of some sort of catastrophe:



(Household Debt)

It’s worth noting that this has all been discussed broadly in Fed research papers already (literally for YEARS).  Many people at the Fed know what they’re doing and for some reason most mainstream economists simply refuse to try to understand how banking or central banking works.  Here’s the key quotes from the horse’s mouth in a paper that has been circulating for two years (via the NY Fed):

“In this note, we present a basic model of the current U.S. banking system, in which interest is paid on bank reserves and there are no binding reserve requirements.  We find that, absent any frictions, lending is unaffected by the amount of reserves in the banking system. The key determinant of bank lending is the dfference between the return on loans and the opportunity cost of making a loan. We show that this difference does not depend on the quantity of reserves.”

That’s basically all there is to it.  No need to make it more complex than that in my opinion.  If there’s a “macro war” going on the Post-Keynesians just took down one of the mainstream’s generals who proved he doesn’t understand how modern banking works.


Got a comment or question about this post? Feel free to use the Ask Cullen section, leave a comment in the forum or send me a message on Twitter.
Cullen Roche

Cullen Roche

Mr. Roche is the Founder of Orcam Financial Group, LLC. Orcam is a financial services firm offering research, private advisory, institutional consulting and educational services. He is also the author of Pragmatic Capitalism: What Every Investor Needs to Understand About Money and Finance and Understanding the Modern Monetary System.

More Posts - Website

Follow Me:

  • FrankH

    What an embarrassing commentary by Feldstein. I thought he might be one of the good guys.

  • SS

    How could Feldstein make such a basic error?

  • D

    Keep fighting the good fight Cullen.

  • Greg

    So according to his logic the fed just needs to keep raising interest on reserves so we can keep those inflation causing reserves from infecting our bank accounts via loans. After all low inflation is the only metric that matters

  • Sant Manukyan

    Hi Cullen,

    What he says is basically what NY FED says. Obviously high excess reserves does not mean no or weak lending but what Feldstein says about inflation is not 100% wrong.

    “Paying interest on reserves breaks this link between the quantity of reserves and banks’
    willingness to lend. By raising the interest rate paid on reserves, the central bank can increase
    market interest rates and slow the growth of bank lending and economic activity without
    changing the quantity of reserves. In other words, paying interest on reserves allows the central
    bank to follow a path for short-term interest rates that is independent of the level of reserves. By
    choosing this path appropriately, the central bank can guard against inflationary pressures even if
    financial conditions lead it to maintain a high level of excess reserves.”

    “If inflationary pressures begin to appear while the liquidity facilities are
    still in use, the central bank can use its interest-on-reserves policy to raise interest rates without
    necessarily removing all of the reserves created by the facilities.”

  • Geoff

    “The amount of reserve deposits in the Fed system are determined exogenously by the Fed.”

    I was under the impression that reserves were simply a facilitating feature, and therefore mainly endogenous? I thought the amount of reserves in the system were determined mainly by the needs of the banks. The Fed supplies reserves mostly in “reaction” to banking activities.

    QE operations would be an exception. When the Fed “actively” enters the market to swap reserves for other securities, the resulting excess reserves in the system would, I suppose, be considered exogenous.

  • Geoff

    F-ck, I hate making grammar mistakes. Does this new site have an edit feature?

  • JP Koning

    I don’t think inflation would explode if the IOR fell to 0% (or even some lower number like -.2%) but it would surely increase. If starting from a stable equilibrium an asset’s return is reduced relative to all other asset returns, then that asset’s price must fall to some lower price so that subsequent investors are sufficiently rewarded for holding it. If the asset suffering a decline in its expected return happens to be the unit of account (through a drop in the IOR), then the entire price level must rise (ie the value of the currency must fall) in order to coax investors to hold it again. Seems pretty uncontroversial to me that reducing IOR to 0, ie. taking it away, would result in inflation.

  • NMW

    My understanding is that the main purpose of IOER is to put a floor on the fed funds rate and the reason we are not seeing inflation is not (necessarily) because of IOER, but instead because banks are not seeing the same demand for loans (consumers deleveraging) or creditworthy borrowers (stricter regulations). This may be oversimlplifying a little but I’m interested if this is the take that other people have.

    I also think that the scarcity of safe collateral fits into it but I’m still formulating that line of thought.

  • Romeo Fayette

    I get why Feldstein’s argument is wrong, but I think he’s confused himself with the terminology (and a poor understanding of the system). Phrased another (more correct) way, couldn’t I say that ZIRP has enabled banks to take their surging deposits and invest them up the curve in something like Treasuries/MBS instead of creating loan assets for the real economy? The NIM on a loan may not adequately compensate a bank for the credit risk. (Excess deposits–not reserves–were after all what enabled JPM’s CIO to make the CDX investments that precipitated the London Whale.)

    Further, were IOER to offer a zero yield, excess reserves would drag on ROA, et al for banks. Given that drag, banks would be encouraged to make more loans, creating more deposits, against which they already have the reserves to meet regulatory requirements. (Already having the reserves helps their margins, because otherwise they’d have to acquire the necessary reserves at a cost via Fed Funds or the interbank market.)

    Yes, banks don’t lend reserves, but in such a way as described herein, IOER affects the decision to lend. What say ye?

  • wh10

    Cullen, I don’t think the authors of that Fed paper have the same general banking model in mind as you do, even if you agree at a high level with them on how the IOR banking system works. For example, they write “Historically, the quantity of reserves supplied by a central bank determines the amount of bank
    loans. Through the money multiplier, banks expand loans to equal the amount of reserves divided by the reserve requirement.” Or “Historically, central banks used a reserve requirement ratio in order to create a demand for reserves that were not paid interest and to control the amount of bank loans through the money multiplier.”

  • narayankpl

    “The problem with today’s environment is not that the banks are holding a huge amount of reserves and unwilling to lend or being paid to hold them. The problem is that there aren’t enough creditworthy customers walking in the doors to demand loans”

    It could also be that there are not enough businesses or people wanting loans. Why would businesses or people not want loans? Businesses because they do not have customers and people because they do not want to or cannot borrow?

    So it is a demand problem (either from business or creditworthy customer). When there is a demand problem how is the Dow and S&P making new highs?

  • Tom Brown

    Scott Sumner makes some comments regarding Noah’s post too:

    particularly this part:

    “The best way to think about monetary policy is through the lens of the expected hot potato effect. Any Fed action should be judged in terms of its impact on the long term trends in monetary base supply and demand, not interest rates. Policies that make the expected future base rise relative to expected future base demand are expansionary, and vice versa. That approach doesn’t tell us that IOR was highly contractionary, but given that the IOR rate is higher than what banks earn on T-bills, it certainly might have sharply increased base demand.”

    Cullen, how would you respond to what Scott wrote here?

  • Tom Brown

    wh10, the key word in both the quotes you provide is “historically.” Take another look at the Fed paper. They are saying that’s how it USED to be, but no longer is. For example, just prior to your first quote from the Fed paper is this sentence:

    “The current banking system in the United States and worldwide no longer re-
    sembles the traditional textbook model of fractional reserve banking”

    They then go on to describe that “traditional textbook model” which our system “no longer resembles.”

    After the second quote you provide the Fed writes:

    “Our model of bank lending, with interest on reserves and no meaningful reserve requirement, shows that the money multiplier is no longer relevant”

  • narayankpl

    No edit provision. Consider this sentence deleted in the above post of mine.

    “It could also be that there are not enough businesses or people wanting loans”

  • Geoff

    Hey Tom, you should simply respond with this:

    “The hot potato effect is a load of crap”

    That should suffice.

  • Tyler

    Does anyone else see a new website format? lol. Seems to change every time I visit.

  • Tom Brown

    Geoff, … I guess I’m not so interested in refuting Sumner, but getting a bit more information about how Sumner’s world view relates to MR…. especially this new “twist” (to me anyway) on summing up MM. What follows the “expected hot potato” bit is more interesting to me… perhaps it serves as a nice succinct summary of his views… and maybe what lies behind his constant refrain of “Never reason from a [price change/interest rate change/etc].”

    I don’t see why Noah’s response to Feldstein, for example, makes Scott “want to pull his hair out” except that it implies that this whole concept of the expected supply vs demand of base money is not relevant some how. There’s something about Noahs straightforward explanation here that Scott finds very objectionable. This was surprising to me at first… but maybe that’s it! Maybe it’s a threat to this long term differential supply vs demand of base money concept that lies at the heart of what Scott thinks is important. BTW, just as frequently Scott want’s to pull his hair out over something Feldstein writes BTW… especially in regards to QE being dangerous or destabilizing.

  • Cullen Roche

    Someone’s working on it. Might be a day or so before things are settled. Sorry.

  • Geoff

    I’m not sure why Scott is frustrated either. You are the MM expert around here :) but it seems to me that Scott’s focus on base money means that he firmly believes in the money multiplier myth.

  • LRM

    Site looking good, Glad you added the comment # beside post to see what is the daily hornets nest!!

  • LVG

    I like the new layout. It’s more like the old layout, but a little cleaner looking. Thank your tech guys for us!

  • wh10

    Yeah, but I’ve been following Cullen for a long time, and am pretty confident Cullen doesn’t think the banking system historically (i.e., prior to IOR) used to resemble the traditional textbook model with the money multiplier.

  • Cullen Roche

    Sure, if they raised the rate on reserves to such an onerous rate that all benchmark lending rates collapsed then they’d certainly have a huge impact on lending in general and the spread at which banks make a profit. But we’re at zero so that logic is not what Feldstein is implying….The IOR is the de facto FFR now. And the de facto FFR is at zero (essentially) so it’s not hurting bank lending in any meaningful way.

  • Cullen Roche

    Think of the whole reserve system as being exogenous to inside money. If banks had it their way there’d be no reserve system and they’d monopolize the system under one bank. Then there wouldn’t even have to be a reserve system. Since we have a pvt banking system with competitive banking the reserve system is a must and only exists because the Fed makes it exist via reserve settlement and reserve requirements.

  • Cullen Roche


  • Cullen Roche

    I think banks would be in this position whether the Fed was paying IOR or not. It’s the demand. Since the demand and spreads on boring old lending are so thin they’ve looked to other more intricate ways to make money. That was going on when the FFR was 5% so I think that further proves that banks are just getting more innovative in their profit endeavors. I think IOR has a lot less to do with all this than some think.

  • Cullen Roche

    Thanks. I just googled something fast since I know that there are Fed papers out there debunking the multiplier. I may have read that one in haste. This paper’s better:

  • Cullen Roche

    His “hot potato” amounts to the equivalent of changing someone’s saving account to a checking account and assuming that someone is more likely to spend just because they have something of higher moneyness. Something of higher moneyness DOES NOT mean more people will spend it.

    If you took every govt bond in the world and swapped it with cash, would spending suddenly rise? Of course not. Why? Because spending is a function of expected income relative to desired saving. If you lose a govt bond your income actually goes DOWN and you saving remains the same. Why would anyone in their right mind go spend more just because they have cash? No one says “oh gee, this cash is burning a whole in my pocket because of inflation, I should probably spend it”.

  • Tom Brown

    BTW, I was surprised to see this statement from Scott:

    “Saturos asked me for areas where my views have been changed by bloggers. I’d rather talk about bloggers who have influenced me. MR is probably my favorite blog…”


    …. “MR?” Really??!! …. then it occurred to me what he probably means: Marginal Revolution!


  • Tom Brown

    Right… but perhaps Sumner’s definition of “expansionary” monetary policy has been a moving target here to some extent. It sounds to me like he’s abandoned the most simple minded view of the “hot potato” effect and now ascribes the magic powers to something else: specifically to the “expectation of base money supply vs base money demand.”

    Why THAT has magic properties is not clear. So it’s no longer simply about putting more deposits in people’s hands (rather than bonds)… it’s about what does the market EXPECT the future (long term) supply of “base money” to be vs the demand for base money. According to Scott, the large inventory of excess reserves on bank balance sheets isn’t doing much (in part) because IOR has increased bank DEMAND for “base money.” So the effects of increasing the supply of base money (excess reserves) with QE has been nullified. (Here I’m in danger of conflating “supply” with “stock”… but I’m not sure how else to write it.)

    So again, I don’t agree with Scott here, because I still think he’s ascribing to base money some special powers which I don’t think it actually has. What he thinks those special powers are is what I don’t understand. WHY does he think these expectations about base money have such an effect? Is this a new kind of money multiplier he’s come up with?

  • dw

    thinking the low inflation rate is caused by consumers (aka workers) incomes not being able to keep up what little inflation we do have. and they aren’t interested in making up the shortfall any more with credit

  • Mr. Market

    Right. Banks are too fearful and don’t want to lend that money. So, they’re keeping that money on at the FED. And then the law of supply & demand kicks in. A lot of demand and (relative) little supply. That pushes the price for T-bills higher & higher and that means that rates go lower and lower.

    We DO have inflation !!! Just look at what has happened in the stockmarkets, bondmarkets and the real estate market since 1981. Falling rates = bondprice inflation.

    @Cullen Roche: I llike the current lay out much better than the previous one. But what I don’t like is my name in italics. I would prefer to see my name in bold characters.

  • Cullen Roche

    Like most people, Sumner doesn’t understand what money is, but thinks that spendable money is somehow likely to make people spend more if they have it. As if we all sit around with cash trying to get rid of it at any cost. No, someone has to hold the cash. And most of us are just fine holding big chunks of cash or deposits earning nothing. The reason why banks love depositors is because they realize that people don’t generally mind holding their deposits in a bank account earning nothing so long as they know it’s safe. Sumner is clearly assuming some other exogenous force. But people don’t spend more just because they have cash. That’s totally illogical and makes me wonder if Sumner lives on planet Earth.

    Of course, this is the same guy who wants to eliminate banking from macro. Does he even realize that most of the “hot potatoes” are created by banks????

  • Rob Jones

    Banks have accumulated about $1.8 trillion in excess reserves since Bernanke started paying interest on them. They had practically none before. And the huge increase in excess reserves goes a long way towards counterbalancing all of the money that the Fed had injected into the economy via QE.

    Whether the banks are accumulating the excess reserves because of the interest paid on them, or whether it is due to a lack of good lending opportunities is difficult to say. Right now, the interest on 1 year treasuries is only 0.12%. So excess reserves are a better place to park short term money than treasuries now. If short term treasury interest rates went above 0.25%, I think banks would switch over to them.

  • Tom Brown

    Rob Jones, you write: “Whether the banks are accumulating the excess reserves because of the interest paid on them, or whether it is due to a lack of good lending opportunities is difficult to say.”

    I think they’ve accumulated them because there’s not too many places those reserves can actually go:

    They don’t get lent out. A small amount gets converted from excess to required status with each loan, but the overall amount does not increase with lending.

    Your point about Tsy rates is interesting, but I think you have to think about it in a macro sense. Where would they obtain those Tsys? If they obtain them from Tsy directly, then they do swap reserves for those. If they obtain them from other banks, this changes nothing in aggregate. If they obtain them from private non-banks, then this doesn’t change reserve levels either… but it does increase the amount of bank deposits that non-banks hold (imagine all the banks aggregated together as one big bank… then since non-banks can’t hold Fed deposits, the bank can only buy Tsy’s from the non-banks by crediting their bank deposits w/o any effect of the banks’ reserve levels). Crediting non-banks’ bank deposits will cause a small % (~10%) of the ER to convert to RR.

  • bart

    Excess reserves may be used as collateral or rehypothecated.

  • Romeo Fayette

    Loan demand is not soft:

    IOER of 25bps provides a risk free alternative (“opportunity cost”) for banks with capital, so reserves against deposits are no longer an economic expense (dead weight), but a revenue generator.

    Deposits come at a weighted avg 10bps cost, more than offset by 25bps IOER, which allows banks to invest the deposit liabilities in something like a risk free Treasury, not needing to take on risk to generate profits.

    This is a very important realization for policy…

  • JKH

    Smith’s piece is actually pretty good for refuting the market monetarist claim (led by Sumner) that 25 basis points has been the issue, compared to zero, or even negative as originally proposed by Sumner. It’s a common sense differential rate argument. 25 basis points is peanuts, no matter how you analyze system behavior.

    Your explanation is more comprehensive, and correct, as to why any interest rate on reserves is not really the issue.

    Slight difference in the targeted error of thinking.

    Smith gets to second base. You get home.


  • JP Koning

    But if the IOR is dropped from 0.25% to 0%, the price level has to rise a little bit, right?

  • Tom Brown

    Now Krugman get’s into the mix backing up Smith’s analysis… which prompts Sumner to scold him too!

  • JJTV

    15bps is not an incentive to hold money in excess reserves. Banks do generate income from their bond portfolios and they certainly won’t hold excess capital in that form. Assume $10bn in investestable capital and your looking at $1.5M…banks are going to be looking for creditworthy borrowers. At the very least they will be using that capital to build bond portfolios that offer higher levels of interest.

    IOER is important because it allows the Fed to increase the deposit rate/overnight rate without selling securities. Thus rates can increase, when the fed wants them too, independent of their asset/liability positions. Banks don’t care very much about their reserve positions. In fact, ask head of lending for a very large bank about their reserve positions, at best you will get a quizical look and a statement like “why do you care? That’s something we always settle last minute and it doesn’t matter from a lending position”. Same with IOER…quizical look, no understanding of why that matters.

  • flow5

    Just like Alton Gilbert’s Requiem for Reg Q. Nobody understands M&B. What was the CB vs. NB cost of funds differential in 1966?

  • Tom Brown

    It’s true that Sumner is upset by Smith’s argument… although he goes on to claim that he’s not so upset with his conclusion as his line of reasoning. Sumner claims he could also make an argument against Feldstein on this point (whom he has a BIG problem with anyway since Feldstein is anti-expansionary monetary policy) but without invoking the Smith/Krugman (Krugman has adopted it too now) line of attack. I have no idea what Sumner’s line of reasoning would be though against Feldstein! … (to be fair, Sumner claims he could argue it *either* way… for or against Feldstein’s hypothesis).

    Today, when criticizing Krugman for adopting Smith’s line of attack on this issue, Sumner writes:

    “Debating the conservative inflationistas [Feldstein, et al] is like shooting ducks in a barrel for a brilliant economist like Krugman. He can do better.”

    But again he fails to demonstrate what that argument would look like.

  • DOB

    Hi Cullen,

    While everything in your post is correct, I humbly–but strongly–disagree with its title.

    IOR is currently the opportunity cost of funds (that’s because base was massively increased to push fedfunds against IOR).

    You (and MMT and MMR) point out that funds are always available to banks in an unlimited quantity (or at least in a non-binding way). While that is correct, those funds aren’t available for free. Lowering IOR will, pretty much bp for bp, lower cost of funds, which is absolutely inflationary (happy to be more explicit on that point if it’s not self evident).

  • Tom Brown

    Mr. Market, you write:

    “Right. Banks are too fearful and don’t want to lend that money. So, they’re keeping that money on at the FED.”

    But banks can’t lend out the money they have at the Fed except to other banks. Thus banks in aggregate can’t lend out that money at all. That money is in the form of Fed deposits and only certain entities can have Fed deposits. Private non-bank businesses and individual people cannot have Fed deposits.

  • Cullen Roche

    It’s only inflationary if there is demand for loans, which is my point. It’s like saying that the relative cost of an iPad should rise if you reduce the price. Well, that assumes either stable or rising demand, right? Same basic thing here. Demand for debt is rising now and banks are much looser in lending than they were so I think monetary policy and interest rates are becoming more important than they were. The IOR is the de facto FFR now. But I think the demand function is more important here. Banks, in theory, have unlimited access to the supply of loans so it’s more about demand than the cost of supply….

  • DOB

    So I’m going to continue to disagree :)

    First of, demand for loans isn’t a yes-or-no binary function. It’s reasonable to assume that it’s smooth as a function of interest rates (and many other factors).

    Here’s one way to look at the effect of lower IOR on higher price level via the private credit standpoint.

    1) My project was unprofitable when financed at X%, now profitable at X% – epsilon. I know that so I’m more likely to ask for the loan. My banker knows that too so he’s more likely to give me the loan.

    2) The prior effect is true for me but also many of the people who will buy my stuff which reinforces (1)

    But all of this holds true even without a banking system which gives out loans. Even if the central bank was the lone supplier of medium of exchange/unit of account and only loaned it against government bond collateral (therefore no direct channel to private credit), interest rate would *still* be the primary determinant of the price level.

    I could totally imagine a scenario where you have steady and health inflation (or NGDP growth), interest rates falling, and private credit contracting. In fact that’s what should have happened in 2008/2009 had the CBs done their jobs and lowered rates below 0%.

  • Adams

    Thanks for all of your work, I thoroughly enjoy reading and trying to understand the monetary system. I keep up with pragcap and MMR and have read the educational links, but can you explain a bit further the idea of Loans creating deposits and banks not being reserve constrained? Can banks basically print money to create a loan for a worthy customer, or must they have the deposits in the bank? Maybe I am thinking too hard, but this article confused me a bit!

  • John Daschbach

    I think Cullen’s statement is brilliant and captures QE in a nutshell to those willing to think. Certainly some “smart” money has wagered that QE has an effect (lower yields, even if yields have risen during QE periods) and opted to swap the current yield for zero yield on cash with the hopes of buying back the nominally same securities at higher yields (lower prices). But it’s closer to arbitrage than most people think. It’s trying to make money on the very small impacts QE technically has on yields.

  • Tom Brown


    First off, it’s just plain “MR” not “MMR.” I know there are instances of vestigial “MMRs” around here and there (especially at

    I’ve drawn some simple examples out on spreadsheets if that helps. Try starting with this one:

    Now we can add in reserve requirements:

    And finally capital requirements:

    So to answer your questions:

    “Can banks basically print money to create a loan for a worthy customer, or must they have the deposits in the bank?”

    Technically they can’t “print money” but they can do the electronic equivalent: type balances into your bank deposit. So no, they do NOT have to have deposits in order to do this. It’s possible for a bank to make tons of loans and NEVER have a deposit on it’s books! For example, suppose a bank does nothing but make loans to people that use the money immediately to make purchases from sellers who bank at other banks. This is basically what happens when a credit card is used (assuming the buyer and seller have different banks). The lending bank would NEVER keep a deposit… they’d always be transferred immediately… along with the backing reserves. So what would that bank’s balance sheet look like? Something like the Bank A’s final balance sheet in the following example:

  • Tom Brown

    So your “Most Popular Posts” box still indicates the number of comments that were present prior to disqus, but now with disqus all the comments are not visible?

    I hope my old links to some of my favorite comments are still active somewhere… I haven’t checked yet, but that would be great if they were.

  • Cullen Roche

    The old comments should be importing at some point.