Paul Krugman seems to have gotten into an argument with the MMT guys about commercial banks creating money. I'm just giving my own views on this question. I'm basically following Leland Yeager [update: and David Laidler]. Strangely, while I agree with the MMT guys on many points (banks do create money out of thin air) I end up (I think) in much the same place as Paul.
Start out with an economy with no commercial banks. Just a central bank issuing currency. Suppose the central bank wants to increase the stock of money. It does not have to persuade people to want to hold more money. Money is weird like that. People accept money in exchange for the goods they sell, not because they (necessarily) want to hold more money, but because they plan to exchange that money for something else.
Money is a consumer (and producer) durable, but it's very different from other consumer durables.
The stock of money is supply-determined in a way that the stock of refrigerators is not. A manufacturer of refrigerators needs to persuade people to hold more refrigerators if it wants to increase the stock of refrigerators in public hands. A manufacturer of money doesn't. That's because money is the medium of exchange. If every household only wanted to own one refrigerator, and already owned one, any extra refrigerators would be left lying on the sidewalk. That could never happen with money, if it was still being used as money. People would pick it up and spend it.
An excess supply of money is like a hot potato. Nobody wants to hold it. But everybody is perfectly willing to accept it, because he knows he can pass it on to someone else, who in turn accepts it because he knows he can pass on to another. I might buy a refrigerator even if I knew I could never sell it again. I would never buy money if I thought I could never sell it again.
The hot potato process is a disequilibrium process. The average person expects to receive $100, and plans to spend $110, to get rid of a $10 hot potato. But he is surprised to discover that he received $110 instead of the $100 he expected, because everybody else is doing the same thing. So he tries again, and keeps on trying until eventually expectations adjust, prices adjust, and incomes adjust, and the extra money is willingly held, in the sense that the average person plans to spend an amount equal to what he expects to receive. Supply of money, eventually, creates its own demand.
Now lets bring in commercial banks.
Commercial banks cannot create GICs out of thin air. Individually and collectively, if commercial banks want to increase the stock of GICs in public hands, they must persuade people to hold more GICs. GICs are like refrigerators.
Commercial banks can create money out of thin air. The commercial bank just buys something. It doesn't matter what it buys. It could be a new computer, or an IOU from someone who wants a loan. The bank pays for it by writing a cheque, drawn on itself. When that cheque is deposited, either at the original bank, or at another, the stock of money in public hands has increased. It does not matter whether the seller of the computer, or seller of the IOU, wants to hold more money in his chequing account. Most likely, he wanted the money because he planned to spend it. That's why people accept money in exchange for goods and services and IOUs. They accept it, only because they believe that someone else will in turn accept it from them.
If that was all there was to it, there would be no difference between commercial banks and central banks. Both can create money out of thin air (or paper, or ink marks, or electrons) just by buying something. They can increase the stock of money in public hands without having to increase the stock of money demanded. But there is one important difference between commercial banks and central banks.
Commercial banks promise to redeem their money at a fixed exchange rate (at par) for central bank money. Central banks do not (banking crises aside) promise to redeem their money for commercial bank money. Redeemability is asymmetric. Just like if the Bank of Canada fixed the exchange rate of the Canadian dollar to the US dollar, but the Fed made no commitment the other way. And because each individual commercial bank promises to redeem its money in central bank money, the central bank is at the centre of a fixed exchange rate system in the same way that the US Fed was at the centre of the other central banks under Bretton Woods.
The bank that makes no promises is free to do what it wants. The banks that promise to redeem their money for another bank's money aren't free to do what they want.
An individual commercial bank can create money out of thin air simply by buying something. But the money it creates may not be its own. Its money may subsequently be redeemed for the money of another bank, or the central bank. The individual commercial bank that wants a permanent increase in its stock of money may need to persuade people not to redeem its money. Whether or not it needs to do anything to persuade them all depends on how the other banks, especially the central bank, react.
I can imagine a world in which an individual commercial bank can permanently create money. All it needs is that the central bank, which is free to do what it wants, allows the supply of its own money to increase in proportion to the supply of commercial bank money. This is what would happen if the central bank targeted a rate of interest, for example. One commercial bank creates money, and all the other commercial banks, and the central bank, respond to the hot potato process by increasing their own money supplies in response to the rising demand for loans and deposits and currency.
But it doesn't have to happen that way. It didn't ought to happen that way. The central bank is free to do what it wants, and it didn't ought to let individual commercial banks set monetary policy in this way. Just because the Bank of Montreal wants to expand its balance sheet by 10% doesn't mean that all banks and the Bank of Canada should expand their balance sheets by the same 10%. Instead, the Bank of Canada should be doing something sensible, like targeting inflation (or NGDP). If the Bank of Montreal wants to permanently increase the size of its balance sheet, and if the Bank of Canada doesn't want to increase the size of its balance sheet, the Bank of Montreal will need to persuade people not to redeem its monetary liabilities for the monetary liabilities of the Bank of Canada and all the other commercial banks. It will need to pay higher interest on deposits, cut fees, open more branches, or whatever, to try to change the public's desired composition of the total stock of money.
The Bank of Canada doesn't need to do that. Because it makes no promises to redeem its money. If the Bank of Canada wants to increase the supply of Bank of Canada money, it just does it. It doesn't need to persuade people to want to hold more.
Worthwhile Canadian Initiative 
Good proces explanation.
Too many economists mess this up by focusing solely on the equilibrium condition.
Bill: exactly! Thanks.
Ok but Krugman said that commercial bank money is limited by reserves and the central bank’s limit on the quantity of currency. He also said that an individual bank must get deposits from somewhere. This is truly just nonsense. Of course there’s some kind of macro equilibrium money supply which the CB can determine given the dynamics of the economy. But it’s definitely a result of a complex system, and not some kind of exogenous constraint. The most sane thing to say is that if the CB holds rates constant, and the natural rate rises, then people will borrow and invest/consume more and the quantity of deposits will rise as a result of borrowing. That is the NK model. Then you can layer the CB response on top of that. But to claim that the money supply is exogenous in our economy is pretty absurd.
K:
Over the long-run the money supply is exogenously set by the central bank as it follows something like a Taylor Rule. In the short-run, with an interest rate target, it is endogenously determined. The confusion of endogeneity vs. exogeneity of the money supply often comes down to confusing the horizon one is analyzing.
K:
This is truly just nonsense
This is not nonsense.
The bank lending ability is limited primarily by its capital, not its reserve balance account, true.
However, if the bank needs to settle with another bank, it does need some M0 on its reserve account, even in Canada where the reserve requirement is zero, or the UK. The retail deposit base is the cheapest source of funding to procure settlement money for banks. Those banks that ‘think’ otherwise usually die. Witness the fate of Chicago Continental in the US or Northern Rock in the UK that chose to rely on wholesale market for interbank cash rather than on traditional retail deposits.
The lender of the last resort notwithstanding who will lend up to a point and then ‘resolve’ the bank.
“strangely, while I agree with the MMT guys on many points (banks do create money out of thin air) I end up (I think) in much the same place as Paul.”
It should be mentioned that Krugman got into an argument with Steve Keen, not an MMT person. This post looks somewhat confused to my eyes.
“If the Bank of Montreal wants to permanently increase the size of its balance sheet, and if the Bank of Canada doesn’t want to increase the size of its balance sheet, the Bank of Montreal will need to persuade people not to redeem its monetary liabilities for the monetary liabilities of the Bank of Canada and all the other commercial banks. It will need to pay higher interest on deposits, cut fees, open more branches, or whatever, to try to change the public’s desired composition of the total stock of money.”
how so? I’m assuming that you’re talking about a deposit drain from lending. If increased lending growth is higher for the Bank of Montreal and this results in a deposit drain, the Bank of Montreal will increase it’s demand for reserves. This will be supplied by other banks on the interbank loan market. If withdrawals of cash cause the commercial banks as a whole to be short of reserves, not providing sufficient reserves will cause a liquidity crisis. Not supplying sufficient reserves to the banking system goes against the reason for a Central Bank’s existence!
Click to access wp97-8.pdf
Nick:
The bank pays for it by writing a cheque, drawn on itself</>
If the above implies that the bank can loan to itself thus manufacturing credit money, that is not true at all. It cannot, and if it did it would be fraudulent. The bank can only lend to another entity, not to itself.
In order to buy stuff, the bank will exchange its assets, possibly the bank’s deposit at another bank for the stuff in question.
Closing italics 🙂
Krugman’s post was silly.
The money supply is not “exogenously set” by the central bank. It is influenced by the central bank, and the degree of that influence depends how close substitutes private liabilities are for c.b. liabilities. Krugman implicitly assumes that demand for c.b. liabilities is perfectly inelastic — i.e. a fixed proportion of assets will be held as outside money. It would be a better approximation today to say that the demand for c.b. liabilities is perfectly elastic. — that any change in the price of c.b. liabilities will lead to an arbitrarily large shift between them and other assets. In other words, the condition that Krugman calls “the liquidity trap” is just the normal operation of modern financial systems, and has been for probably 15 or 20 years now.
Your post is not silly, but I still don’t agree with it.
(1) When you say “But it doesn’t have to happen that way. It didn’t ought to happen that way,” is that, as it seems to be, a tacit admission that it actually does happen that way? Because, it does. As a practical matter, the stock of c.b. liabilities is not a constraint on deposit creation by private banks.
(2) You keep talking about asymmetric redeemability. I don’t think there is any such animal. In any real-world situation, it is exactly as easy to convert cash to to a checking account, as it is to convert a checking account to cash. And people do not hold commercial bank liabilities because they can be converted to money. They hold commerical bank liabiliteis because they can be converted to goods and services, i.e. because they are money.
(3) “Start out with an economy with no commercial banks. Just a central bank issuing currency.” It would be interesting, one of these days, if you were to start out from the opposite premise. Imagine a world with no central bank (or commodity money), just private bank liabilities used to settle transactions.
MMTers often point out that banks can lend first and finance later (this is the point made in that BIS paper everybody likes to quote, and that chap Holmes that Keen quotes in his response to Krugman). This might matter, if you thought that banks have to have loanable funds up front before they lend, and hence giving them more reserves could loosen a constraint and boost lending. But otherwise it’s unimportant. In the money multiplier story taught to undergrad, nothing hangs on which order things happen in, that I can see.
Some MMTers also say that banks create deposits when they lend – they can just out of thin air create $1000 in the deposit account of somebody who borrows a $1000. This is also true but unimportant. That’s just an entry in an accounts ledger. As soon as the borrower actually uses the money, the bank has to be able to finance it. Banks cannot finance their lending via these notional deposits created in the act of lending. We know banks need people to deposit funds with them, because they go to great expense to attract savers.
There’s a post by asymptosis on angry bear (sorry no links, in a hurry) where he/she uses an example to argue it doesn’t matter whether people save or not, the money is all there in the banking system somewhere to finance lending. I think that’s wrong. Borrowers still require savers. Somebody somewhere has to keep a positive balance in their bank to finance somebody else having a negative balance. If everybody (individuals, firms) did decide to spend their income as soon as they got it, and keep their bank balances at zero, there wouldn’t be anybody to fund lending (except the central bank?).
What I need to give more thought to is the idea that lending adds to aggregate demand because it creates new money to be spent. Krugman doesn’t like that idea, saying that the spending of the borrower is offset by the saving of somebody else. That sounds suspiciously like “the money has to come from somewhere” argument against fiscal stimulus. If you think of credit expansion as either increasing the money multiplier or endogenously drawing forth an increase in base money, or both, they it increases the money supply and don’t people believe increases in the money supply increase AD? But the idea the new borrowing adds one to one to AD doesn’t sound right either- what then would be the difference between the money supply and AD? they’d just be different ways of saying the same thing.
This idea of Keen’s that credit expansion adds to AD except when it “leaks” into higher asset prices looks like pure horseshit to me.
The bank lending ability is limited primarily by its capital, not its reserve balance account, true.
This is already a big advance over Krugman.
However, if the bank needs to settle with another bank, it does need some M0 on its reserve account
Yes, but it matters how inelastic this demand is. Some of the demand for cars falls on Ford Fiestas. But that does not mean that Ford, as the monopoly supplier of Fiestas, can set the price of generic “cars”.
“They can increase the stock of money in public hands without having to increase the stock of money demanded.”
Nick, I am confused by the way you are wording this, and I am not sure I agree. The banks don’t have to increase the stock of money demanded because they are responding to an increased demand for money. Someone wanted a loan (i.e. they wanted more money), so they bank supplies it at a price. But the way you write it makes it seem like banks can just push more money on people without persuading them (see your second paragraph); that’s missing the whole point that the bank is responding to increased demand for money. Is this different from what you’re saying?
K:
I do agree that these words of Krugman’s are questionable:
First of all, any individual bank does, in fact, have to lend out the money it receives in deposits.</>
But, the MMT point of view that the bank just extends a loan and “worries about reserves later” is quite naive and simplistic too. In the sense of being able to settle interbank obligations, the reserve account position is a real issue although not a one-to-one metric of the bank’s lending capacity. That issue resolution is the fed fund desk manager’s full time job at bigger banks or a treasurer’s at smaller banks who try to predict fed fund inflows/outflows proactively in order to ensure their bank survival.
Anoter italics problem.
not a one-to-one metric of the bank’s lending capacity.
Right, the relationship is complicated. But to do economics, we need to represent it in a simple way. So the question is, which first-order approximation is better: Krugman’s, that there is a one to one relationship between reserves and bank loans? Or the Post Keynesian, that bank loans are not constrained by reserves?
Personally, I think Krugman’s approximation was ok in the postwar decades when he was first studying economics, but the PK approximation has been clearly better since sometime in the 1980s, or 1990s at the latest.
(Note that talking of “banks” is already missing a big part of the story. What we’ve seen in recent decades is that if the supply of reserves does constrain bank lending, then we will just get more credit creation by non-bank financial institutions.)
Luis Enrique- you’re missing the point that when the bank needs reserves to settle withdrawals of some sort (like when a borrower uses their loan), they can get their reserves from the interbank market. And the CB supplies the right amount of reserves to keep the interest rate in the interbank market at their target. So the bank always knows the cost of reserves is X% in the interbank market- and that will serve as the “base” loan rate. But it’s not about quantity of reserves, it’s about the price administered by the CB. Now, you are right, new deposits are also a source of reserves for banks. If banks can acquire new deposits in a way that is lowering their cost of liabilities more so than if they were just borrowing reserves in the interbank market, then they can perhaps make loans at a lower rate. But this doesn’t change the intuition being offered by Post-Keynesians, Keen, MMTers, about how bank lending works. It’s about price, not quantity of money deposited or supplied by the CB etc.
Borrowers still require savers.
Not true. Keen has a confusing and idiosyncratic way of expressing himself, but he is absolutely right to insist he is following Schumpteter and Wicksell as well as Keynes in saying that credit creation by banks does not require an affirmative decision by any economic agent to increase saving.
What’s remarkable to me about Krugman is that not only is he unfamiliar with the heterodox literature (forgivable) but he doesn’t seem to know anything about Wicksell, who is supposed to be one of the founders of the profession and who was theorizing a system of pure credit money over a century ago.
The banking system makes a loan of $1,000 to A. This increases both banking system assets and liabilities by $1,000. Fro the point of view of the non-bank economy, money stock has increased by $1,000. A spends the new money on goods and services, transferring the liabilities to other actors who increase their spending in turn, until prices and/or income have risen in the aggregate by just enough to increase desired money holdings by $1,000. Nobody at any time wished to save more. Again, this is what both Wicksell and Schumpeter believed was the normal process of money creation in the banking system. (Altho unlike Keynes, they believed that it was mainly or entirely the price level, rather than aggregate income, that would adjust.) The central bank money story is a later addition that corresponds to a new set of regulations imposed in the first half of the 20th century on top of this credit money system — reserve requirements etc. Historically, the evolution is the opposite of the one Nick describes.
Ahh, that beautiful reserve constrained vs. capital constrained debate again.
Nick, I think it is time to introduce ‘desired reserves’ in the Macro lexicon, just like ‘desired savings’.
Also, a simple reductio ad absurdum of a bank that met capital requirements and still had excess reserves can be useful, no? I don’t know where to get this data.
Loanable funds is fundamentally a two-period (at least) model of expectations. I have never quite understood how/why MMT tries to disprove it by using a cash flow statement, which only captures what happened, not what was desired, and is an after-the-fact one period model.
“They can increase the stock of money in public hands without having to increase the stock of money demanded.”
Yes, I agree with wh10, that the above is incorrect. Unless someone comes into a bank and asks for a loan, there is nothing the bank can do to extend a credit without a willing counter-party. No loan -> no credit -> no money creation.
You left out one of the most common ways for banks to create money, lower lending standards. They do create money out of thin air. They are not alone in this but do operate with much higher leverage and a lender of last resort which non banks lack.
VJK- maybe it is an oversimplification, but I think JW Mason rightly identifies it as a econ 101 simplification that doesn’t change too much and is intended to drive the point that “loans create deposits” and banks can always require reserves, if needed, at some price. That said, from a high-level, simplified, literal perspective, because of the lagged reserve accounting system, it kind of looks like banks “worry about reserves later.” But if you read the actual literature, it won’t be that simplified. These are obvious complex, contemporaneous matters. Even in the informal setting, MMTers will say that banks are always ‘worrying’ about the cost of acquiring reserves, so it’s clear they don’t mean “worry about reserves later” in all senses of the words. In any case, I don’t think MMT banking experts like Scott Fullwiler would disagree with what you are getting at.
Also, I think it is possible to defend loanable funds from the ‘loans create deposits’ challenge by invoking J W Mason’s great ‘what adjusts’ argument. In ‘loans create deposits’, a cash flow entry is being invoked holding income constant. This obviously does not scale in the aggregate because in the aggregate, income adjusts. That’s what a demand-driven business cycle is all about anyway.
which first-order approximation is better: Krugman’s, that there is a one to one relationship between reserves and bank loans ? Or the Post Keynesian, that bank loans are not constrained by reserves?
Well, JW, in my opinion neither is good enough due to both being too extreme and simplistic. One has to be able to model price of liquidity management in banks’ lending activity. That price, for the bank, may vary from zero to infinity under various market circumstance during relatively short periods of time. And calculating that price is extremely tricky taking into account that fed fund/M0 borrowing is uncollaterized and largely depends on the borrower credibility and/or CB whims which in many cases are unquantifiable. The cost of funding is not just the current price of interbank money set up by the CB.
Lord:
“Thin air” emphasis is rather odd and misleading too.
The bank, by extending a loan, risks its capital. If the loan defaults, the bank capital decreases by the amount of the loss, simplifying somewhat in the spirit of Econ 101 and assuming the loan is uncollarized and the TBTF put is not in place
vjk- right, but like I said, the process is more accurately understood from a cost of funding perspective, not a quantity constraint. that’s the point of the oversimplification, and it’s an econ 101 oversimplification that still drives that main point home.
Ritwik- I don’t understand the points you are trying to make.
“Also, a simple reductio ad absurdum of a bank that met capital requirements and still had excess reserves can be useful, no? I don’t know where to get this data.”
Banks in the US right now have TONS of excess reserves, due to QE and IOR, and are meeting capital requirements. What’s your point?
As far as “desired reserves,” banks desire reserves to the extent that they need them to settle payments or meet reserve requirements. They often hold a small buffer of excess reserves to avoid penalties. Banks have no use for reserves beyond this, so they loan them out in the interbank market to earn something off of them. Or in the US today, they just hold them due to IOR, which becomes the interbank rate.
My god, I understood that! Bless you, Nick Rowe, a thousand times bless you.
This argument has had me confused from jump street, but I’m not an economist… I just watch them on tv (and the Internet).
Oops, sorry, ‘loans create deposits’ is also a two period thing, and income adjusts there as well.
I think I will have to go back to ‘desired reserves’.
wh10
My argument is simple, if lending was never reserve constrained, then you should never see excess reserves.
“Banks have no use for reserves beyond this, so they loan them out in the interbank market to earn something off of them. Or in the US today, they just hold them due to IOR, which becomes the interbank rate.”
So, you’re saying, there are negligible lending opportunities in the US beyond a risk adjusted rate of 0.25%? The world does not change radically if IoR is 0.25% instead of 0%.
Ritwik – before IOR in the US, you didn’t see excess reserves, beyond the small buffer I mentioned, precisely because of the points I was making. See http://research.stlouisfed.org/fred2/series/EXCRESNS
But that buffer is not properly thought of as ‘constraining’ lending, because again, this is a story about cost of funding, not quantity constraints.
There aren’t literally negligible lending opportunities in the US now, but credit appetite and willingness to supply the loans has certainly diminished given the state of the economy and the debt binge we’re recovering from. The reason there are so many excess reserves is because the Fed swapped bonds for reserves like crazy via the QE program. Normally this would make the FFR plummet to 0%, but the Fed is paying IOR to support it at a positive level.
if lending was never reserve constrained, then you should never see excess reserves.
This is backward. It is reserve-constrained lending that is incompatible with excess reserves.
Anyway, in the real world, we never see excess reserves unless the central bank pays interest on them.
(Normally we get reserve-requirement carrying lending up to the limit of reserves. The we get lending that does not carry reserve requirements, at an additional funding cost of epsilon. The cost to a abnk of converting some demand deposits to time deposits is trivially low. As is the cost of substitution from bank to nonbank lenders, once the fixed costs of entry have been paid.)
I just can’t get past how embarrassing the Krugman post is. He really did say that individual banks need to get deposits to lend them out! And I agree with you, Anon1, that the misunderstanding can be interpreted as confusing an exogenous constraint with an equilibrium condition. And he actually said that he thinks the currency supply is controlled by a CB limit. I’m dumbfounded.
But come to think of it, it really explains a lot about his discomfort with the NK model that he himself has used as recently as his 2010 paper with Eggertsson. He always wants to come back to ISLM, and the reason, it now appears is that he truly believes the LM curve to be an essential part of the macro dynamic. The NK model doesn’t have a quantity of money, not because the money supply doesn’t exist, but because it is extraneous to the problem, a mere vestigial appendage. As JW Mason points out, perhaps some, like Krugman, whose understanding was developed in an earlier reserve banking era can’t shake their essentially monetarist intuitions.
Krugman’s reserve-constrained lending model (which Nick endorses) makes a simple prediction: there is a stable money multiplier. If we don’t see that — and we don’t — his model needs to be replaced.
neither is good enough due to both being too extreme and simplistic.
Sorry, this won’t do. Economics — communication in general — is impossible without simplification The only question is which simplification to choose. Map is not territory, etc.
Krugman has a model in which, normally, the binding constraint on the volume of credit, and the level of real activity in the economy, is the stock of outside money set by the central bank. (Nick R. has this model too.) If there is not a stable relationship between the supply of outside money and the supply of bank money, this model needs to be replaced. You can say, well, yes there’s a relationship but it’s constantly bouncing around. but in that case, it’s not changes in the stock of money that’s driving fluctuations in real activity. It’s whatever’s causing the relationship to bounce around. If you you know what that is, put it in your model. If you don’t know, then you need to treat bank lending as exogenous.
Any valid theory can be represented in a simple model. If you can’t summarize your theory, you don’t have a theory. The statement, “the volume of lending is driven by a vast complex of variables whose interactions are too complex to be summarized” is operationally equivalent to, “I don’t know what drives lending.”
JW Mason: “the volume of lending is driven by a vast complex of variables whose interactions are too complex to be summarized…”
Hilarious! And bang on.
JW Mason,
How is that backward? If ‘finding reserves’ after the fact of ‘making a loan’ is not a challenge (lending is not reserve constrained), then a bank will never hold excess reserves. Hence, if lending were never reserve constrained, then you would never see excess reserves.
wh10
I agree with your point of why you see excess reserves in the system due to QE and the constancy of the FFR – Buiter has made a similar point when he argued that having an IoR/FFR (he views the two as same conceptually) is incompatible with quantity reserve requirements (you can’t control the price and the quantity of anything at the same time). I will go further to argue that this also explains the near zero reserves in other ‘liquidity demand’ recessions – had the Fed done QE then, my hypothesis is, you would have seen excess reserves shoot up even at an IoR of 0%.
Notice how you say ‘credit appetite’ and ‘willingness to supply the loans’ in the same sentence – this is exactly my view of the world. In all states of the world, investment is both demand constrained and supply constrained. ‘Loans create deposits’ captures the demand bit and the interaction between borrowers and financial intermediaries. Loanable funds captures the interaction between savers and financial intermediaries. You can only bring the two together by trying to model default risk and liquidity demand (and not the commercial banking system per se, which anyway is not the most important financial intermediary anymore).
“loans create deposits” isn’t intended to describe the entire process, obviously. And obviously there will be other premia attached to loans beyond the cost of acquiring reserves (default, term, etc). I am well aware I said those words, but they don’t change anything regarding my point. I don’t know what your model of loanable funds is, but if it tells you Person A needs to save and deposit in a bank for the bank to lend to Person B, then it’s wrong empirically and theoretically. It is missing the Fed and its role in making sure reserves are always available at whatever its target is. Banks always have that.
If ‘finding reserves’ after the fact of ‘making a loan’ is not a challenge (lending is not reserve constrained), then a bank will never hold excess reserves. Hence, if lending were never reserve constrained, then you would never see excess reserves.
To say “lending is reserve constrained” is to say “the volume of lending is determined by the stock of reserves.” In a world where this was true, any change in the quantity of reserves would be reflected in a proportionate change in the volume of lending. In other words, to say that lending is reserve-constrained, is to say that banks are lending as much as they can given the supply of reserves. So there should never be excess reserves.
By contrast, to say that lending is NOT reserve-constrained, is to say that volume of loans and the volume of reserves vary independently. An increase in lending simply involves the financial system substituting away from liabilities that carry reserve requirements. If the liabilities involved are close substitutes — as they are — then this process will be effectively costless. In this case, we would expect to see a large fall in lending and/or large increase in reserves associated with the appearance of excess reserves.
Ritwik:
not the commercial banking system per se, which anyway is not the most important financial intermediary anymore
The commercial bank is the only institution that can manufacture credit money. Other intermediaries operate with loanable funds that are pre-created by banks and vitally depend on the former in their operations and the very existence. Other intermediaries do not create new deposits but merely provide ways to swap existing ones (ownership transfer). Thus, in the light of the above your quoted statement is incorrect.
apols if this is a duplicate, earlier comment got eaten
wh10 – sure, the savings can arrive via the wholesale (interbank) or retail markets, it doesn’t matter.
JWMason,
as I said, when a bank creates both an asset and a liability by lending £1000, that doesn’t create any funds to finance loans – that requires a saver. In your story, when A spend the money the money has to come from somewhere. If banks really could create money out of thin air, they’d be printing it like the central bank, and would have no need to pay savers interest on their deposits to attract them. When you write that borrowing does not require an affirmative decision to save by somebody else, I half agree. When A spends the money buying roller skates, the money ends up in the skate vendors deposit account. If the vendor then spends it, the vendor’s bank can’t use it to fund loans. If the vendor keeps it on deposit for a while, they are ‘saving’ it and it can be used to fund loans.
JW Mason
I think this is one of those A->B vs not A -> not B type divergences. To say lending is reserve constrained is not to say that an increase in the quantity of reserves would lead to increased lending. It is to say that a decrease in the quantity of reserves would lead to decreased lending. The two are not symmetric.
By corollary, if lending were exogenously determined and not reserve constrained, then a ‘reverse-QE’ type action by the Fed now would lead to a decrease in excess reserves. The ‘desired reserves’ hypothesis would say that lending would fall.
We are unlikely to see this scenario (reverse QE, decrease in gross quantity of reserves) play out anytime soon, and I anyway believe this is only one half of the story. But I do believe that the mechanical accounting process of making a loan does not negate this half of the story.
In my ideal theory, the ‘desired reserves’ would be subsumed by ‘liquidity demand’ and reserves would not be treated differently from other short term interest bearing government liabilities. Currency and reserves would be modelled separately, with the current redeemability being an artifact of history and not a model necessity. The transactions demand of money would be preceded by (and indeed arise from) default risk. Money would be both endogenous and exogenous, investment would be both demand and supply constrained and ‘liquidity demand’ and ‘default risk’ of the financial sector (not just commercial banks) would explain disequilibria and recessions. This is close to the Goodhart/ Buiter way of thinking.
“It would be interesting, one of these days, if you were to start out from the opposite premise. Imagine a world with no central bank (or commodity money), just private bank liabilities used to settle transactions.”
That’s where Keen starts from (and why he gets grief from his MMT friends).
Nick,
What if the Bank of Canada wants to decrease the supply of Bank of Canada money. Does it still not need to convince anyone to hold less?
Sorry, chuck the previous comment. It does not make sense.
Nick,
Suppose you expect to receive $100,000 of monetary income in the forthcoming year. You have a detailed plan for spending and saving that income. In accordance with that plan, you expect that exactly one year from today, you will have spent $90,000 and saved $10,000. Your planned saving/income ratio is therefore 1/10 and your planned spending /saving ratio is 9/1.
Call this the “expected outcome” (EO) – the outcome in which your income and spending/saving ratio are exactly as described above.
Now, your income in the coming year might be different than what you expect. It might be higher. And you might save either a higher or lower percentage of your income as a result. There are many ways in which these outcomes might diverge from expectations. But let’s just consider a few such as the following in which you still spend exactly the expected amount of $90,000:
(O1) You receive $110,000 in income. Your saving to income ratio is 2/11 and your spending to saving ratio is 9/2.
(O2) You receive $120,000 in income. Your saving to income ratio is 3/12 and your spending to saving ratio is 9/3.
(O3) You receive $130,000 in income. Your saving to income ratio is 4/13 and your spending to saving ratio is 9/3.
(O4) You receive $140,000 in income. Your saving to income ratio is 5/14 and your spending to saving ratio is 9/4.
Etc …
Now my claim would be that a rational person obviously ranks each of these alternative outcomes as preferable to the expected outcome EO. And yet in each case, the alternative saving to income ratio is higher than the saving to income ratio in EO. In each of the alternatives, you end up holding both a higher percentage of your income, and a higher quantity of money overall.
Now, perhaps I do not understand what the hot potato thesis is. But as I understand it, your claim is that the behaviorally fundamental variable in all this is the expected quantity of income saved. That’s your target. So as your income rises in an unanticipated way, your marginal propensity to spend out of income rises with it.
But to me, that seems backward. Your target quantity of saving is simply an epiphenomenon of your anticipated income and desired spending, and has no independent preferential significance for you. There is no reason at all to think that income earners would regard unanticipated income as a hot potato because it forces their expected quantity of saving higher if not spent. The expected quantity of saving isn’t a target that rational agents desire to maintain; it is a by-product of a constraint that they are very happy to see change if the constrain is lessened.
I believe it would be equally a mistake to believe that the behaviorally fundamental variable in all this is the ratio of spending to saving, or equivalently, the ratio of saving to income. If those ratios were your target, then for every one of the alternative outcomes O listed above, there would be a preferable outcome O’ that we can list this way:
(O1’) You receive $110,000 in income. But your saving to income ratio remains 1/10 and your spending to saving ratio remains 9/1.
(O2’) You receive $120,000 in income. But your saving to income ratio remains 1/10 and your spending to saving ratio remains 9/1.
(O3’) You receive $130,000 in income. But your saving to income ratio remains 1/10 and your spending to saving ratio remains 9/1.
(O4’) You receive $140,000 in income. But your saving to income ratio remains 1/10 and your spending to saving ratio remains 9/1.
Etc …
Some might argue that O1’ is preferable to O1, and also to any other alternative in which the income is $110,000 and the saving rate goes us. They might make the same point about O2’ and O2, etc. But this does not seem to accord with empirical evidence as I understand it, which generally finds that as income goes up, the marginal propensity to consume goes down.
… So since there is no hot potato effect, there is no mechanism preventing the aggregate motives of the public from being the mere sum of the aggregate motives of individuals. There is no ceiling on the willingness of the public to hold more money. As long as someone is giving them money as opposed to asking them for something in exchange for it – either goods or services or promises of more money in return – they will be happy to accept more money.
But this is one place where the asymmetry between commercial banks and central banks plays a role. Commercial banks do give monetary instruments away – they always exchange it. They have to because the money they create and “give” to their loan customers is a real liability for them, a debt for something that they don’t control. The central bank’s “liabilities” corresponding to issued currency are not genuine debts.
And the other difference between the two is that the central bank, even if it wants to give money away, is more limited by law as to whom it can give it to and for what purposes. It doesn’t matter whether they can sock as much money into commercial bank reserve accounts or securities accounts as they want. That activity has little influence on what commercial banks will do.
The commercial bank is the only institution that can manufacture credit money. Other intermediaries operate with loanable funds that are pre-created by banks
Wrong. Non commercial banks create credit money all the time. Credit cards are issued by institutions other than commercial banks. Money market mutual funds function as means of payment. Etc.
as I said, when a bank creates both an asset and a liability by lending £1000, that doesn’t create any funds to finance loans – that requires a saver.
You did say, but it isn’t true. In the process I described, money was created and no one made any decision to save. Again, this is not new. Schumpeter describes exactly this process in Business Cycles (and other places.) Wicksell describes it. Keynes describes it.
In your story, when A spend the money the money has to come from somewhere.
Nope. When A “spends the money”, they transfer their liability on the banking system to person B. have you ever written or received a check? Have you ever used a credit card? These are bank liabilities that are transferred from payor to payee. The bank liability itself is the money, there is no other money coming form somewhere else.
If banks really could create money out of thin air, they’d be printing it like the central bank, and would have no need to pay savers interest on their deposits to attract them.
The statement that that banks may face funding costs associated with expanding their balance sheets is not the same as the statement that they need outside money provided by the central bank. In any case, demand deposits typically do not pay interest. The banking system creates deposits at the same time it creates loans. the private sector holds those deposits willingly because the process of credit creation increases nominal income by enough to increase transaction-balance demand by the amount of the newly created deposits.
Martin Wolf: “money is just the liability counterpart to private credit decisions.” That would be Martin Wolf, the FT columnist. Maybe he knows a little about the financial system.
Canada is a complicated example, but it’s worthwhile to understand the plumbing. The plumbing determines the rules of the game and that is intentional.
In Canada we should not talk about “reserves” but about capital. In order to make loans and participate in the banking process (actually make payments) you have to have sufficient capital to participate in the Large Value Transfer System. LTVS has two channels for payment, Tranche I and Trance II. Tranche I is secured by the clearer’s Bank of Canada deposit account, so it is “reserves” in a classic sense. You cannot clear a transaction in Tranche I greater than your deposited amount, if you want to do so you have to deposit more funds.
Tranche II is “defaulter pays” (from the BoC account).
Tranche II is a secured pool of funds provided by clearing participants. It is expected that the pool will cover any default and if it doesn’t (lottery odds) the Bank of Canada will cover the residual.
Banks are expected to arbitrage the cost of clearing in Tranche I and Tranche II, normally Trance II handles 90% of all transactions.
You can’t participate in Trance II if you can’t participate in Trance I, though if you are eliminated from Tranche II you an still clear in Trance I, by design.
So to participate in LVTS you need enough money (capital and deposits) to have a Bank of Canada deposit balance and if you have that, you can also use Tranche II by providing liquid money-market capital.
Monetary policy in Canada consists mostly of setting account requirements for clear and the “spread” between the overnight deposit and loan rates on Bank of Canada accounts. Buy design, interbank loans for Tranche II are less than the BoC spread, otherwise an immediate arbitrage opportunity is created to force the rate back down.
Retail cheques are cleared in another system called ACSS that has interest rates 1.5% above the LVTS rates. It’s LVTS that determines monetary policy.
“Money is a consumer (and producer) durable, but it’s very different from other consumer durables”
I’m not sure I agree. The demand for money derives from specific attributes it possesses. For example: Marketability (it can be exchanged for just about anything) , expected future value, ease of storage etc. Other goods have some of these attributes and may be used as partial substitutes for money. With commodity money the supply is derived by the market as for any other good. With non-commodity money the supply is regulated and controlled by the monetary authorities.
If the monetary authorities do a poor job of managing supply then at one extreme (where supply increases too quickly) people will not hold money because they expect it to be worth less in the future than it is now. In this situation people may well choose a refrigerator (that they do not need to store food in) over money if they expect that the value of the refrigerator will hold up better. At the other extreme people may choose to hold onto money as an alternative to spending it on other things. If the monetary authorities do not adjust the supply to meet this increased demand then the value of money will increase. If prices happen to be sticky then we get a slowdown in economic activity. People may well resort to barter or alternative monies if official money is doing a bad job of optimizing market transactions .
In a modern economy the banking system is one way that the authorities regulate the money supply. When banks “create money out of thin air” then they can only do so safely if they have access to funds that will be acceptable to people who end up selling goods to the takers of the new loans. It will typically be the monetary authorities who control the size of and access to these funds and who ultimately use this to control the size of the money supply.
JW Mason/ wh10
Not sure if you’re still interested in this, but I just wanted to understand what are the implications of ‘lending is exogenous to the monetary base’.
Let’s say reserve requirement is 20%. Let’s say at the beginning of time, there are no excess reserves : the balance sheet of the banking system (B) has $20 reserves, $20 govt bonds, $60 private loans as assets (with various corresponding deposits as liabilities) and the central bank (CB) has $80 govt. bonds as assets and $20 reserves as liabilities (with $60 net worth). Assume no currency.
Assume I am the only active private sector non-bank player in the economy. Everybody else is sleeping. Now, I go to a bank, ask for a loan of 10, it makes me the loan, creates equivalent deposits. It gets its reserves shortfall from some other bank, which gets it from some other bank, etc. but B as a whole is still short of reserves (because it had zero excess reserves at t=0). So B goes to CB and asks for $2 reserves. CB buys $2 govt bonds from B and adds $2 reserves to its account. B’s balance sheet increase by $10, CB’s balance sheet (and monetary base) increases by $2. Mishkin/Mankiw see this as a money multiplier of 5 – we laugh at their ignorance for mistaking cause and effect. All is well with the world.
Tomorrow, I go to the bank to ‘pay back’ my loan. B’s balance sheet reduces by $10, it now has $2 in excess reserves. B asks CB to exchange its $2 in excess reserves with govt. bonds. CB, run by a perversely ‘inflationary’ monetarist, doesn’t want to reduce the monetary base/ balance sheet. It declines. $2 shows up as excess reserves, even though these reserves pay zilch.
1) Is this a fair description of what you would expect to happen?
2) If yes, why is 2008 the first time that excess reserves showed up (beyond the various frictions that wh10 describes)? Are you saying that it has never happened in modern American history that the private sector has wanted to reduce debt and the central bank has not wanted to let the monetary base fall?
3) Tomorrow, if the Fed were to bring the IoR to 0% and also refuse to sell back the government debt that it has, what do you expect would happen to excess reserves?
The various Federal Reserve banks have veritable warehouses full of papers on how bank lending works. I’m amazed that any professional economist could not know this stuff.
Click to access conf39c.pdf
Banks make loans by creating offsetting asset and liability entries. If they don’t want to hold the asset, they can securitize and sell it. This removes both entries from the balance sheet leaving only whatever profit was made from the transactions (and maybe some servicing revenue). Note that what was loaned was money, and the security is (broad) money (assuming it goes through the blessing process to become a AAA CDO). Money has been magically doubled.
There are effectively no constraining reserve requirements. The Fed supplies reserves as needed. Bank lending is constrained by:
Capital ratios and VAR calculations
Demand for loans at rates profitable to lend.
credit standards which swing between lax and paranoid depending on where we are in the business cycle.
other Basel requirements (this stuff is above my pay grade, and I’d make a hash of it.)
Last I looked, There were something like $1 trillion of excess reserves. This is a result of QE and has no effect on bank lending, since lending is still constrained by the factors above, but nobody is constrained by a surplus of reserves.
However, banks are always looking for leverage, since profit is a function of leverage. The more you can lend, the more you can make.
So they have tricks among them:
valuation of tier 3 capital (it’s worth what I say it is). There are limits on the percentage of capital that can be tier 3
off balance sheet entities http://voices.washingtonpost.com/economy-watch/2010/04/lehman_brothers_the_evil_repo.html
securitization of loans
repos which use the much higher velocity of money in the repo market as a multiplier http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1440752
exploiting idiots like AIG (a Goldman specialty) http://en.wikipedia.org/wiki/Joseph_Cassano
Lehman had 40:1 leverage. And eventually they were constrained out of business, but it took years and they took a few people with them. Some of the European banks have 50:1 leverage.
As for whether debt adds demand, have a look at this chart.
http://research.stlouisfed.org/fredgraph.png?g=67L
When you teach fractional reserve banking, what on earth do you teach?