Bank Lending and Bank Reserves

By Dan Kervick

Frances Coppola has a very nice piece in Forbes that takes on some of the continuing confusion over commercial bank reserves, central bank payments of interest on reserves and the relationship of both to commercial bank lending. She concludes with a ringing rejection of the frequently voiced claims that the Fed’s policy of paying interest on reserves inhibits bank lending, and that high excess reserve levels are an indicator of sluggish bank lending or bank hoarding:

The volume of excess reserves in the system is what it is, and banks cannot reduce it by lending. They could reduce excess reserves by converting them to physical cash, but that would simply exchange one safe asset (reserves) for another (cash). It would make no difference whatsoever to their ability to lend. Only the Fed can reduce the amount of base money (cash + reserves) in circulation. While it continues to buy assets from private sector investors, excess reserves will continue to increase and the gap between loans and deposits will continue to widen.

Banks cannot and do not “lend out” reserves – or deposits, for that matter. And excess reserves cannot and do not “crowd out” lending. We are not “paying banks not to lend”. Positive interest on excess reserves exists because the banking system is forced to hold those reserves and pay the insurance fee for the associated deposits. It seems only reasonable that it should be paid to do so.

I wholeheartedly agree with the bottom line moral Coppola draws from the operational mechanics of bank lending, but I do think some additional clarity can be had on the question of whether or not commercial banks lend their reserves. And I also have some reservations about the justification Coppola cites for the policy of paying interest on reserves in the first place.

First a clarification of terms: Commercial bank reserves come in two forms. One of these forms is the physical cash held by banks in their vaults (and their ATM machines, tellers’ drawers etc.) The second form of commercial bank reserves consists in the deposit balances held by commercial banks at the central bank. Physical cash reserves are held by banks to satisfy their depositors’ demand for cash withdrawals. Reserve balances at the central bank, on the other hand, are used to make payments to other businesses that have their own Fed accounts – primarily other banks.

The traditional terminology of “reserves” can be misleading, especially as applied to reserve balances at the central bank. That terminology suggests that these assets are held “in reserve” as some sort of backup against unforeseen contingencies. But the fact is that banks need their reserve balances to conduct routine, daily business. They need them because in the everyday course of business they must make payments to other banks in the commercial banking system. Whenever a depositor at one bank makes an electronic or check payment to someone who banks at a different bank, then in order for the payment to be completed, the first bank will have to settle with the second bank.  Net daily interbank settlements are processed by flows into and out of reserve accounts. These reserve balances could thus more accurately be called “payment assets” or “clearing balances”.

Do banks lend their reserves? A compete answer requires being clear about which question we are asking.  If the question is, “Does an individual commercial bank lend its reserves?” then the most accurate answer is “Sometimes.”  If the question is “Does the commercial banking system in the aggregate lend its reserves?” then the most accurate answer is “No.”

Let’s consider the mechanics of a single bank loan. A bank loan is an exchange: The borrower receives some asset from the bank, and the bank gets some asset from the borrower. The asset the bank receives from the borrower is a promissory note specifying the amount to be paid by the borrower to the bank and the schedule on which the payments are to be made. At the time of the loan, the borrower may request the entire borrowed amount in cash. In that case the bank clearly does lend out part of its reserves. The asset the borrower receives is cash, and the bank has reduced its cash reserves by surrendering part of those cash reserves to the borrower. In exchange, the bank gets a promissory note. It has financed its purchase of the promissory note by paying cash.

But it is more likely that the borrower will not take cash on the spot. Instead the bank will credit a sum of money to the borrower’s demand deposit account at the bank. In that case the asset the borrower receives is not a cash asset that the bank surrenders to the borrower, but a liability of the bank. The bank has in this case financed its purchase of the promissory note with debt. Yet once the borrower begins to make use of that deposit balance by making payments to various other people and businesses, the bank will have to begin surrendering assets from its reserve account – assuming that many of those payees bank at other banks.  And the bank will surrender cash reserves to the borrower if the borrower decides to makes cash subsequent withdrawals from the account. So when an individual bank makes a loan, a portion of its reserves do then leave the bank as a consequence, whether that consequence is immediate or takes place over time.

But the important thing to note is that whenever these reserve movements take place, the reserves don’t leave the banking system as a whole. An interbank payment results only in reserve balances moving from one bank’s reserve account to another bank’s reserve account. Similarly most cash withdrawals proceeding from a loan are used very quickly to make cash payments to a business, and as a result are quickly re-deposited back at some bank in the commercial banking system. Thus, an increase in bank lending only results in more rapid movement of reserves from bank to bank, but has negligible net effects on the total volume of reserves held. If an analyst is attempting to gauge changes in the pace of bank lending, and is looking for a fall in reserve balances as an indicator of such a change, then that analyst is looking in the wrong place.

In fact, in a more normal, pre-QE environment in which banks are not carrying excess reserves, an aggregate increase in bank lending would result in an aggregate increase in reserves. An expansion in the consolidated balance sheet of the commercial banking sector will almost certainly be accompanied by a corresponding expansion in the volume of interbank payments and an expansion in the velocity of cash withdrawals and re-deposits. If banks were previously carrying just enough reserves to manage their liquidity needs relative to the previous size of the consolidated banking system balance sheet, they will likely need to expand their reserve holdings to manage their newly increased liquidity needs. Banks will either borrow the additional reserves from the Fed directly or sell other assets to the Fed to get those reserves. There is no other way to get them. An individual bank can increase its reserve holdings by attracting deposits from other banks. But the commercial banking system as a whole cannot do this, since there is no significant source of deposits outside the banking system as a whole.

So now let’s turn to the question of interest on reserves.  First, we should note that the Fed currently pays interest on all reserves balance, not just excess reserves. Currently the interest rates for these two types of balances are the same: 0.25%. The Fed has Congressional authorization, however, to pay different rates for excess reserves and required reserves, including a rate of 0% if it so chooses.

But why pay interest on reserves in the first place? Coppola suggests that it is to compensate banks for the insurance fees they must pay to guarantee their deposits:

Positive interest on excess reserves exists because the banking system is forced to hold those reserves and pay the insurance fee for the associated deposits. It seems only reasonable that it should be paid to do so.

But I don’t really understand that rationale. For one thing, there may be no “associated deposits”. Coppola’s article is about the recent divergence between bank deposits and bank loans, and in the course of the discussion Coppola ably explains that in the QE era, central bank asset purchases can lead to higher reserve balances that are not driven by any increases in bank lending. Also, banks have been paying insurance fees for insured deposits for many decades, but the payment of interest on reserves is a relatively new policy – only in effect since 2008. Did the Fed suddenly decide that it wasn’t being reasonable?

Banks exist to serve the public interest, and while there is a public interest in ensuring the overall health of the banking system, it is no part of public policy to make sure banks earn the maximum amount of possible profit. Banks do not need to be compensated for holding one portion of their assets in the form of reserves, a substantial portion of which they would probably choose to hold anyway to manage liquidity needs. The Fed requires banks to do a lot of things, many of which incur costs, but it doesn’t pay for all the things the banks are required to do. Also, holding cash reserves in bank vaults is another source of security and insurance costs for banks. But the Fed does not pay interest on physical cash reserves – only reserve account balances. If payment of interest on reserves was motivated by the desire to compensate banks for costs associated with reserves, why doesn’t it make payments on all reserves, and not just deposited reserves?

The fact is that the Fed has told us many times why they pay interest on reserves, and in this case there is no reason at all to be skeptical of the explanation. Part of modern central bank policy consists in targeting the interbank lending rate, which in the US is known as “Fed Funds rate”. Prior to the decision to pay interest on reserves, the interbank rate had shown more volatility than Fed policy makers were comfortable with. By paying interest on reserve balances, the Fed hoped to establish a floor to the Fed Funds rate, because – in principle – no bank will lend to other banks at a rate lower than the rate they can earn by holding the reserves in their Fed account.

However, there are non-depository institutions that possess Fed accounts and have access to the Fed Funds market, but which are not eligible for the interest payments the Fed is authorized to pay to depository institutions. So the decision to pay interest on reserves has not had the effect of creating the hard floor on interbank lending rates that the Fed might have initially hoped it would have.  The Fed has recently attempted to address that issue by experimenting with a fixed rate overnight reverse repo facility that could – again in principle – re-establish a hard floor for short-term interest rates by offering these non-depository institutions reverse repo contracts at a fixed positive rate determined by Fed policy-makers.

However, in her confirmation hearings, new Fed chief Janet Yellen indicated that the Fed is considering reducing the payment of interest on excess reserves.  Some have suggested that the Fed reduce interest payments on excess reserves as a kind of cosmetic “signal” to be sent in conjunction with the tapering of QE, so as to reassure markets and the public that despite the end of QE, the Fed is committed to a policy of easy monetary conditions. Some economists go as far as Alan Blinder in suggesting that eliminating the payment of interest on reserves is necessary to restore the phenomenon of “high-powered money” and allow the Fed to stimulate economic expansion via quantitative means. But for the reasons Coppola has ably adduced in her article, whether or not the Fed pays small amounts of interest on reserves has almost nothing to do with bank lending decisions. The theory of high-powered money, and the money multiplier and loanable funds models that go with them, are pieces of textbook economic lore that were always wrong, and should be disposed of for good.

Cross-posted from Rugged Egalitarianism

Follow @DanMKervick

41 Responses to Bank Lending and Bank Reserves

  1. The notion that deposit insurance premiums are paid by banks is wrong. Like any cost of any business, it is the customer who pays, not the business. Interest rates on insured deposits are lower because of FDIC insurance.

    When my employer “unbundled” some services long ago, my customers complained about paying for things they used to get “for free”. I told them that they had always paid for everything, because my company never had any money to pay for anything, except the money we got from our customers.

    The words don’t apply as well to banks, but the principle is the same.

  2. “The volume of excess reserves in the system is what it is, and banks cannot reduce it by lending.”
    That is a true fact if banks hold/own all money and none is held in the economy. However, if money is loaned into the economy and held by the economy then the excess reserves will decrease as loans are made.

    • I think you should re-read the Coppola piece and this piece. The point is that when banks loan money into the economy, they don’t do it by decreasing their reserve holdings in the aggregate. They do it by expanding their balance sheets. Money is one kind of entry on the liability side of commercial bank balance sheets.

    • What sorts of activities might be going on in that economy if it chooses to hold on to that money? I’d guess either the volume of economic transactions has increased or the price level of existing transactions has increased or some combination. Just wondering if I missed other possibilities.

      • A legitimate need for cash or a loss in confidence in “computer money” and credit cards would result in a greater demand for cash money held in the economy. Black market economies/drug trade is an example of the first and we may be only one smart hacker away from the second.

  3. financial matters

    I agree with the money multiplier and loanable funds theory being wrong but think high-powered money still stands at the top of pyramid. Sovereign currency such as Treasuries and US dollars. Then bank loans and then non-bank loans.

    I would be interested in your take on this statement..

    ‘“”Essentially, the new reserves provided by the purchases program enabled the banking system to fund the repayment of about $1 trillion of various forms of advances to financial institutions under the emergency lending program. The emergency lending program ended, but quantitative easing replaced it.””

    http://www.levyinstitute.org/pubs/rpr_4_13.pdf

    In his book ‘Extreme Money: Masters of the Universe and the Cult of Risk’ (2011) Satyajit Das talks about the ‘Liquidity Factory’.

    On an inverse pyramid at the bottom little pinnacle are the central banks, 2%, then there are bank loans, 19%, then securitized debt, 38% and then derivatives 41%.

    We’ve seen what can happen when the Fed adds securitized debt to its balance sheet. I’m wondering what will happen with the derivatives problem.

    Basically the Fed seems to have gotten into this problem by dealing with the GFC as a liquidity crisis rather than taking the steps toward resolution of fraudulent institutions.

    • Some people might just use the term “high-powered money” as a synonym for “monetary base.” In that case I have no objection to it, and I agree that it is at the top of the pyramid of banking system liabilities.

      However, I believe that the term “high-powered money” was originally introduced to capture the belief that the introduction of this kind of money into the economy carried a greater impact than the introduction of other forms of money, because it was associated with a predictable money multiplier, and so incremental additions of monetary base generated predictable incremental increases in broad money.

  4. The theory of high-powered money, and the money multiplier and loanable funds models that go with them, are pieces of textbook economic lore that were always wrong, and should be disposed of for good.

    I was not an econ major, but did take a couple econ classes as an undergrad, one of which was “Money and Banking.” I was taught these things and believed them. But it didn’t take too much convincing for me to see that they weren’t true (which it shouldn’t have, because they aren’t).

    So, where did they come from? Why? I mean, I can understand why something can become entrenched over time and, therefore, hard to get rid of. But I have a hard time understanding where these ideas came from in the first place – how they originated, such that they could become entrenched at all.

    I’m hopeful the chinks in the armor may be getting more obvious now, at least partly thanks to the internets.

  5. Dan, can you explain why interest paid on reserves has not established a hard floor under the Fed Funds rate for non-deposit institutions? Is it because deposit institutions can make loans to non-deposit institutions regardless of their reserve balances? That .25% interest rate looks pretty puny as it is. Would (do) banks make loans at even lower rates to non-deposit institutions? Thanks, James Cooley

    • In one of the NY Fed pieces I linked to this is explained, James, and the explanation actually goes the other way around. There are non-depository institutions that have deposit accounts at the Fed and also have access to the interbank market for both borrowing and lending, but don’t receive interest payments on their Fed deposits. If a depository institution is receiving 0.25% on its Fed deposits, then it will not want to lend those deposits in the interbank market for less than 0.25%. But if a non-depository institution is receiving 0% on its deposits, it will be willing to lend them at any positive rate of interest. This has resulted in downward pressure on the Fed Funds rate, often bringing it down below 0.1%, since there are players in that market who are not receiving IOR and are thus willing to compete to lend at these ultra-low rates.

      The idea behind the new experiment in fixed rate reverse repos is that if you offer these non-depository institutions a reverse repo contract for greater than 0.25%, then if they have extra cash on hand they will go to that market to take that rate, and not lend to other banks at a lower rate.

      • Thanks for the explanation. That makes sense now. On the other hand, if reserves are not the source of loans, couldn’t a non-deposit institution place all of its excess funds in repo contracts and still make loans at less than the IOR rate?

    • financial matters

      I see this in a slightly different light. I think the Fed is giving these non-depository institutions, mainly money market funds, access to ‘high-powered’ money so they don’t have to go to the bank repo markets to try and keep from breaking the buck.

      The Fed set up a few facilities to address this ‘commercial paper’ problem..

      “4.4.2 CPFF

      Like AMLF, the Commercial Paper Funding Facility (CPFF) was also directed toward credit markets and had nearly the same operational duration, running from October 2008 to February 2010 with 1,159 transactions. Unlike AMLF, this facility was designed to support the commercial paper (CP) market rather than MMMFs. Because the CP market saw its funding move into safer securities after Lehman’s collapse—primarily government treasuries—issuers faced rollover risk and a plummeting issuance rate.64 The CP market would shrink by as much as $300 billion by the end of October 2008 (a month after Lehman’s bankruptcy) with 70 percent of this due to reductions in the issuance of financial CP and 20 percent from ABCP reductions.65 A new source of funding was required to drive new issuances of CP. Because purchasing CP by issuers was outside the operating framework of the Fed, the creation of an SPV to buy CP was a necessary step. Providing funding to issuers of CP not only drove new issuances and decreased the interest rate issuers would have to pay to borrow funds but it also decreased the level of asset sales by those that found themselves unable to raise cash and decreased the pressure on credit lines by commercial banks.”

      http://www.levyinstitute.org/pubs/rpr_4_13.pdf

  6. Dan,

    Thanks for this. It is a far more detailed analysis than would have been appropriate for a Forbes audience!

    I disagree with your argument that individual banks “lend out” reserves. The mechanism you describe is deposit withdrawal, not lending. Withdrawals from lent funds (deposits) by borrowers are indistinguishable in terms of reserve movements from any other sort of customer payment. It’s not helpful to confuse the creation of a loan, and associated deposit, with the subsequent disbursement of those funds by the borrower.

    I did also explain that deposit withdrawals in physical cash reduce reserves. This applies both to settlement of lending in physical cash and to other cash withdrawals, for example from wages. Again, it’s not helpful to single out physical cash disbursement of lent funds.

    Regarding excess reserves and associated deposits, I explained in the post that asset purchases from non-banks create a deposit/reserve pair. There are no associated deposits when asset purchases are made directly from banks – but that is not what we were discussing.

    • Thanks Frances.

      Quickly on the last paragraph, and then I have to drive home from work. It seems to me that the fact that asset purchases directly from banks do not result in the creation of a commercial bank deposit is relevant to your (admittedly brief) comment on why a central bank might want to pay interest on reserves in the first place. If there is no additional deposit created, then there is no additional insurance cost to the bank for insuring deposits. So I don’t think it is plausible to say that interest on reserves is justified as a payment to the bank for those insurance costs.

    • Also, it seems to me that distinguishing loan operational mechanics in which (i) a borrower takes a loan, has the loaned amount momentarily credited to a deposit account and then immediately withdraws all of it from the account on the same visit to the bank, (ii) a borrower borrows cash with no amount ever credited to a deposit account, or (iii) the borrower has an open overdraft privilege and line of credit at the bank and borrows cash by withdrawing it from an ATM machine, is to make a distinction without a difference. In all three cases, the bank’s total reserves are immediately reduced at the time of the loan, and so it seems impermissibly fussy, at best, to say the bank has not loaned its reserves.

      And even if the borrower takes the loaned amount on account initially, and only subsequently begins withdrawing cash from that account and issuing payment orders against that account, it seems to me that an important thing to stress is that a bank deposit balance is a liability of the bank, a debt payable on demand. When people take loans they receive a certain kind of claim on the banks assets, and they quickly begin to assert that claim by directing the bank’s payment assets – its reserves – in one direction or another. So the sense in which it is quite true that “banks don’t lend their reserves” is mainly in the aggregate, not at the individual bank level.

      It is also important to stress that when banks want to expand their balance sheets and give loans in the form of credits to deposit accounts, they can usually begin to do this right away, and address any needs for additional liquidity subsequently. Nor are central bank attempts to substitute liquid reserves for slightly less liquid, longer term interest-bearing assets on bank balance sheets likely to have any impact on bank lending decisions which, as you note, are mainly determined by the demand for loans at various rates of interest and levels of repayment risk.

      • I’m afraid I think the operational mechanics matter enormously.

        Take a UK auto loan, for example (typically these are not securitized so they are a good example of simple unsecured lending). On Monday, customer goes to his bank and borrows £5,000 to buy a used car. He’s seen a car he likes but he wants to get a loan in advance of expressing an interest because he knows he will get a better price if he can buy the car for cash. That money goes straight into his checking account (we call them current accounts here). Once the money is in his account, he takes the car for a test drive. He likes it, but it is a used car and everyone knows about used car salesmen, so he arranges for an independent mechanical review. This is done on Wednesday and the results come back to him on Thursday. The car is sound apart from a few minor faults. On Friday the customer pays a deposit to the car seller out of the £5,000, retaining the rest until the faults are fixed. The car seller takes a week to fix the faults. Customer picks up the car the following Friday and pays the rest of the £5,000. Clearly, there is only ONE loan but TWO payments, neither of which is made at the same time as the loan.

        The fact that a deposit is a liability of the bank is a red herring. Yes, the payments are from a current account which is a liability of the bank, but also in that current account are the customer’s wages, which were paid on the first Friday (the day he paid the deposit on the car) and from which he paid his mortgage the following Friday (the day he paid the balance on the car). How do you know which payment was made from the loan funds and which was made from the wages? They are all in the same account, they all look exactly the same, and they all cause reserve movements. You can’t tell them apart. It is payments, not lending, that drive reserve movements.

        • Oh, and how about loans that never trigger payments at all? Let me give you an example.

          Suppose our car-buying customer receives a bonus from his work that is paid directly into his checking account. He pays the car seller £5,000 from his current account for the car. Suppose he then decides that actually he would rather keep the bonus money for other things: he doesn’t have any particular plans, but it’s nice to have some money in reserve, isn’t it? So he RETROSPECTIVELY borrows £5,000 from the bank, which is paid directly into his checking account. That loan is never settled, is it? No reserve movements are ever triggered as a direct consequence of that loan. The reserve movement that it might have triggered has already happened, and how on earth are you going to determine which of all the future payments from that account are triggered by the loan?

        • Frances, it seems to me that you are just articulating some of the reasons why I gave the answer “Sometimes” to the first question I posed in the piece.

          Yes, there can be cases of loans by a bank that don’t trigger any associated reserve outflows at all. And there can be cases where there is a reserve outflow triggered by depositor payment orders following a loan to that depositor, but where there is no principled way of identifying the amount of the outflow attributable to the loan and the amount attributable to other sources of the total balance in the depositor’s account.

          But in a case in which a borrower goes into a bank to borrow £5,000 and walks out with 100 £50 notes in a satchel, it seems perfectly obvious that the bank has loaned some its reserves. The bank now has a new promissory note among its assets on the asset side of its balance sheet, the borrower has £5,000 in cash, and the bank has £5,000 less in reserve assets than it had before the transaction. The very notes in the borrower’s satchel were included among the cash reserves in the bank’s vault just moments before, and now they are not.

          And even in the more mixed cases, bank liquidity managers are charged with making rational professional estimates of increases in the rate of reserve outflows that are likely to result from an expansion of the bank’s balance sheet due to increased lending. Increased lending triggers increased reserve outflows.

          What I’m trying to guard against here are some extreme forms of the endogenous money thesis that some commenters have occasionally vented here at NEP and elsewhere, and that they read into the “from thin air” language. According to these erroneous views, deposit balances are not any kind of genuine bank liability at all. The defenders of these views imagine that not only do banks create deposits “from thin air”, so to speak, but that they can create their own assets from thin air as well. In other words, banks make money by pure seigniorage – by manufacturing it at will.

          On this view, if a bank wants to buy a fleet of cars, it can simply create a deposit account for the car dealer and credit it with the necessary total, without creating any sort of claim against its other assets that it must service by surrendering some of those assets. That’s a dotty view. Certainly a bank can create a deposit account for an auto dealer, and credit it with any amount of money it chooses in payment for the cars, but as soon as the dealer begins either withdrawing money from the account or issuing drafts against that account the bank begins losing reserve assets. The bank hasn’t manufactured assets through seigniorage; it has only issued a binding liability – a claim against its assets – that it must then be prepared to make good on.

          I wrote about this in a few earlier pieces here, including this piece:

          Hyper-Endogeneity

          • What I’m trying to guard against here are some extreme forms of the endogenous money thesis that some commenters have occasionally vented here at NEP and elsewhere, and that they read into the “from thin air” language. According to these erroneous views, deposit balances are not any kind of genuine bank liability at all. The defenders of these views imagine that not only do banks create deposits “from thin air”, so to speak, but that they can create their own assets from thin air as well. In other words, banks make money by pure seigniorage – by manufacturing it at will.

            Hello Dan. Good point; in fact, that’s always been my problem with what you call “extreme forms of the endogenous money theses”. But I can’t reconcile this with the fact that reserves amount to a fraction of the overall money supply. If, as you say, “increased lending triggers increased reserve outflows”, how is it possible that – in the eurozone, for example – bank money accounts for 91% of the entire money supply? Thanks.

            • Thomas, again we need to distinguish individual banks from the banking system as a whole. If an individual bank increases its lending there is going to be an increased outflow of reserves from that bank in the immediate and short term, but there will be inflows later on as the loans are repaid. If all banks increase their lending at the same time, then even in the immediate term all banks will see an increase in both reserve outflows and reserve inflows. In other words there will be an increase in the total volume of flows per week, month, etc. But the banking system as a whole will not experience a net outflow, because there is no out where the money can flow.

              The banking system’s total reserves only need to be a fraction of bank deposits in order to handle depositor withdrawals and interbank payment needs. But if total deposits go up, then total reserves will have to go up commensurately to handle the larger volume of business transactions.

              • Hi Dan. Thanks for your reply (and sorry for my late reply). A couple of points. I appreciate the fact that, as you say, “the banking system’s total reserves only need to be a fraction of bank deposits in order to handle depositor withdrawals and interbank payment needs”, but I think a lot of people have a hard time understanding exactly how that is possible. While it is obvious that we only take out in cash a fraction of what’s deposited in our accounts (which allows the bank to “keep only a fraction of bank deposits” in cash), it’s not at all as clear why that would be the case with loans. After all, in most cases, if I ask a bank for a loan, it’s not to keep it sitting in my account, but to make an instant bulk payment to another bank. In that case, wouldn’t my bank need to have the full equivalent amount in reserves at hand? And if so, how can it afford to only keep “a fraction of the bank deposits it just created” in reserves?

                Moving on to the wider issue of endogenous vs. exogenous money, and to the fact that, as you note, the truth is somewhere in the middle: I guess it all boils down to the extent to which central bank monetary policy is able to indirectly influence the money supply, right? In your article about “Hyper-Endogeneity” you write that “latter forms of narrow government money usually play a foundational role in constraining and underpinning the broader forms of money” but you don’t delve much deeper into the issue. I’d like you to kindly expand on this point a bit, and especially on the extent to which, in your opinion, interest rate adjustments are able to influence the credit-based money supply, in both leveraging and deleveraging contexts. While it can be argued that the relationship between the two has been rather stable throughout modern history, it seems clear that from the ’70s onwards we have witnessed an increasing disconnect between narrow and broad money, and between the latter and credit. As Moritz Schularick and Alan Taylor write in their brilliant NBER paper on “Credit Booms Gone Bust”:

                Our new dataset allows us to establish a number of important stylised facts about what we shall refer to as “two eras of finance capitalism” […]. The first era runs from 1870 to 1939. Our gold-standard ancestors lived in an age where aggregate credit was closely tied to aggregate money. In this era, money and credit were volatile but, over the long run, they maintained a roughly stable relationship with each other and with the size of the economy measured by GDP […]. This stable relationship between money and credit broke down after the Great Depression and WW2, as a new secular trend took hold that carried on until today’s crisis. But prior to 1930, broad money and loans had been stable at about 50%–60% of GDP for decades, and bank assets stood at about 80%–90% of GDP. In this second era, money and credit began a long postwar recovery, trending up rapidly and eventually surpassing their pre-1940 levels compared to GDP by the 1970s […]. In addition, credit itself then started to decouple from broad money and grew rapidly, via a combination of increased leverage and augmented funding via nonmonetary liabilities of banks. In recent decades, we have been living in a different world, where financial innovation and regulatory ease have permitted the credit system to increasingly delink from monetary aggregates, resulting in an unprecedented expansion in the role of credit in the macroeconomy. By 2007, the typical level of broad money had risen to about 70% of GDP, but bank loans exceeded 100% and bank assets were over 200% […]. [This] would beg the question as to which was the more important aggregate in driving macroeconomic outcomes. At least with respect to crises, the results of our analysis are clear: credit matters, not money.

                I’d like to hear your opinion about this. And about another, related issue: in neither the present article nor the aforementioned one do you touch upon the question of securitization, which surely is hugely relevant to the issue. After all, if a bank can issue a loan and then sell it on to a third party – thus instantly monetizing an otherwise illiquid asset – the “you had better be prepared to hand over $100 in some form whenever I demand it” argument you make in “Hyper-Endogeneity” begins to fall apart, wouldn’t you agree? Isn’t it indeed the case that “securitization implies that there is no limit to bank initiative in creating credits”, as Minsky argued?

                Thanks for you attention.

                • financial matters

                  And it gets worse when you add in derivatives…

                  In his book ‘Extreme Money: Masters of the Universe and the Cult of Risk’ (2011) Satyajit Das talks about the ‘Liquidity Factory’.

                  On an inverse pyramid at the bottom little pinnacle are the central banks, 2%, then there are bank loans, 19%, then securitized debt, 38% and then derivatives 41%.

                  He considers this ‘cotton candy’ which is spun sugar composed mostly of air.

                • “how can it afford to only keep “a fraction of the bank deposits it just created” in reserves?”

                  Because other banks are also making loans, and some of that new money ends up as deposits in this bank. And customers make loan payments every day, too. Each night (or every 2 weeks?), when all payments are settled and the dust clears, individual banks can lend and borrow reserves among themselves, and the Fed can buy or sell repos so that the total reserves in the system is appropriate. (Since QE, there is no longer any such Fed activity required.)

  7. Pingback: Bank Lending and Bank Reserves | Heterodox econ...

  8. While I agree that bank lending is never constrained (or encouraged) by the level of reserves I disagree that IOR has no effect on lending. When a bank lends out it knows that if as a result of that loan it needs additional reserves to settle with other banks that it can borrow that money at the interbank rate, and this become part of the costing of a loan. The lower the interbank rate is then the more lending opportunities the banks is likely to be able to find. If IOR puts a floor on the interbank rate then I fail to see how it can not be a constraint on lending. If there was no IOR there would be loans that would be made , that are not currently not being made.

    • I never really though about it quite this way, and now that I’ve read your nicely stated comment, I definitely think I agree.
      The impact would certainly be small (relatively) and at the very edge of the margin, but I feel comfortable with the tautology that:

      # of bank loans with IOR < # of bank loans at 0%

      And as long its at least one, the tautology would hold.
      Yeah, that sounds about right.
      Thanks RR

    • I agree with that. But that’s just another way of saying that paying IOR is a way of sustaining a target Fed Funds rate, and that the target Fed funds rate is the Fed’s chief monetary policy tool: it is the major factor in determining the cost of additional funds for a bank, and banks make their profits on the spread between the cost of funds and the expected return from loans. Given that the current target rate of IOR is a measly 1/4 of one percent, that is not much of a factor, even if the banks were not carrying excess reserves.

      But, in fact, they are carrying excess reserves. And so the cost of funds question is moot for the foreseeable future, because banks are carrying such a vast quantity of excess reserves that they could expand their loan books from now until Doomsday before running into a need for additional liquidity.

  9. “Banks exist to serve the public interest..”

    And there’s the rub. Perhaps that is what a great many of us wish to be true, but it is more than clear whose interests the banks serve.. and it ain’t us!

  10. Alex Seferian

    Dan, as usual, a very interesting piece. Can you please help me with a related set of questions: How does the Fed accommodate demand for reserves when there are not enough in the banking system, and there are no assets to exchange for reserves? This question relates also to how an individual bank may go about obtaining reserves. Assume a bank has no excess reserves, has no assets to sell in exchange for reserves, and cannot raise enough equity to fund the newly needed reserves. Can such a bank simply borrow reserves from the Fed using the new loan it wishes to create as collateral? Actually, that begs the question, do banks even have to provide collateral when borrowing Fed reserves? In other words, and to simplify, can a bank just borrow reserves “out of thin air”?

    • Thanks Alex. It’s a good question but one I never really thought about since it is also a moot question. Right now, the commercial banks are carrying almost 2.4 trillion in reserve balances, of which only about $79 billion are required reserves. At the same time, they own almost $1.8 trillion in Treasury and agency securities, more than 22 times the amount of their required reserves. So even if bank loans and required reserves doubled in a very short amount of time, the banks have so many treasuries to sell that it is just not an issue.

      But your question might be about an individual bank, not the banks in general. It strikes me that if a bank were in the situation you describe, where it couldn’t raise money by selling stock, then that would have to mean the bank was in a lot of trouble to begin with. But banks can borrow money at the Fed ‘s discount window by pledging loans as collateral:

      http://www.frbdiscountwindow.org/frcollguidelines.pdf

      There are also capital requirements to bear in mind. One of those requirements is the total capital requirement, which now must be at least 8% of risk-weighted assets (RWA). Simplifying, total capital is simply the difference between the bank’s assets and liabilities:

      C = A – L

      So the total capital requirement is

      (A-L)/RWA > .08

      The risk weight for ordinary business loans is 1, and the risk weight for mortgage loans is .5. So suppose a bank’s assets consisted entirely in mortgage loans. Then RWA would be equal to (.5)A. And suppose its total liabilities are 95% of its total assets – i.e. suppose L = (.95)A. The capital ratio (A-L)/RWA would then be equal to:

      (A – (.95)A)/(.5)A

      or

      (.05)A/(.5)A = 10%

      That’s cutting it close. If the bank’s assets consisted of all business loans, then the ratio would be 5%, which is not high enough. So in practice, banks must hold a portion of their assets in the form of cash and near-cash equivalents like government securities in order to meet their capital requirements.

  11. Reserves never cross the line, and leave the banking system for entry into the real economy. This is obvious, by virtue of the fact that nobody regards currency in the hands of the non-bank public as being banking reserves. But rather, such currency forms part of the money supply aggregate M1.

    Commercial banks DO lend out their reserves, but only to other players within the financial system, or to a government agency. They never lend out their reserves to their retail customers.

  12. Pingback: Bank Lending and Bank Reserves | New Economic Perspectives » ##Political Reality Blog

  13. CBs as a system, pay for what they already own. To impose an interest rate differential in favor of these money creating depository insitutions induces dis-intermediation within the non-banks (where the size of the non-banks shrink, but the size of the CB system remains unaffected). This is exactly the same phenomonon that occurred during the 1966 S&L credit crisis. And it occurred for exactly the same reasons. It decreases the supply of loan-funds, impounds savings within the CBs, & destroys financial intermediation (the matching of savings with investment).

    And from the standpoint of an individual CB, they do loan out their excess reserves. They are increasingly re-deposited within any BHC, or are re-distributed & re-concentrated within the TBTF banks in the system.

    Remunerated excess reserve balances do not restrict lending, but they do inhibit investing. And they do “crowd out” investment.

  14. Isn’t that plagiarism Dan Kirvick?

  15. Thanks for the clear and jargon-free explanation of the role of the Federal reserve. In terms of increasing the money supply and creating money out of “thin-air”, at what stage does this occur? In the end, there must be a hard cash or hard reserve account deposit to back up a loan. If the government “spends” it goes through the motion of “borrowing” by selling bonds – which is just reclaiming money already existing in the private sector. But to honestly use the term “create”, can the Fed just increase the Treasury’s account by order of congress?