The Bank of England has published a lovely clear article (by Michael McLeay, Amar Radia and Ryland Thomas) on "Money Creation in the Modern Economy". Thanks to JKH for the tip-off. (Here is JKH's blog post). But I disagree with it.
Thinking about monetary policy in terms of interest rate policy just doesn't work. It doesn't work in theory, and it doesn't work in practice. That's why the Bank of England is having to do QE, and is having to re-introduce the old general theory of monetary policy as a special theory for QE.
This is the part of the article that most caught my attention:
"Like reductions in Bank Rate, asset purchases are a way in which the MPC can loosen the stance of monetary policy in order to stimulate economic activity and meet its inflation target. But the role of money in the two policies is not the same.
QE involves a shift in the focus of monetary policy to the quantity of money: the central bank purchases a quantity of assets, financed by the creation of broad money and a corresponding increase in the amount of central bank reserves.
The sellers of the assets will be left holding the newly created deposits in place of government bonds. They will be likely to be holding more money than they would like, relative to other assets that they wish to hold. They will therefore want to
rebalance their portfolios, for example by using the new deposits to buy higher-yielding assets such as bonds and shares issued by companies — leading to the ‘hot potato’ effect discussed earlier. This will raise the value of those assets and lower the cost to companies of raising funds in these markets. That, in turn, should lead to higher spending in the economy."
There is nothing wrong, in my eyes, with that second paragraph. Good monetarist hot potato stuff! What is wrong is the sentence that immediately precedes it: "But the role of money in the two policies is not the same."
They have two theories of how monetary policy works. There is one theory for when the Bank of England sets a rate of interest: "And reserves are, in normal times, supplied ‘on demand’ by the Bank of England to commercial banks in
exchange for other assets on their balance sheets." And a second, quite different theory, for when the Bank of England does QE.
I'm sure they are not alone in having two theories: one for "normal times"; and one for QE, which is seen as needing a special theory only applicable in "abnormal times". But it is rather peculiar having two different theories of the same thing.
One theory is better than two.
What is even more peculiar is that their special theory for QE is the same as the general theory taught in first-year textbooks. The central bank buys a bond and the money supply expands, because the seller of the bond now owns the money that the central bank gave him in exchange for the bond.
"Quantitative Easing" is just a silly new name for the "Open Market Operations" that first-year textbooks have always said was the way that central banks normally increase the money supply.
If you spend your life teaching first year economics, like I do nowadays (when I'm not blogging), this here modern world looks very peculiar. My general theory has become their special theory, and they have gone and invented some weird new theory of money creation in what they call "normal times", that they admit doesn't work as a general theory, and they still need my old theory to handle the cases where their weird new theory doesn't work.
Here is my general theory: when the central bank buys something, with central bank money, the money supply expands, because whoever sold them that something now holds extra money. Done. It does not matter whether the central bank buys a bond, or a computer, or whatever. Hell, it could just give the money away to its favourite charity (helicopter money), and the result would be the same.
Why can't my general theory work equally well in "normal times"? Let me repeat that above quote, this time with bold added: "And reserves are, in normal times, supplied ‘on demand’ by the Bank of England to commercial banks in exchange for other assets on their balance sheets." See that bit about "in exchange for other assets"? That means the Bank of England buys something. Just like I said in my general theory. The central bank increases the money supply by buying something. See, it's easy!
Now if the central bank is buying something, and someone else is selling something, there must be some sort of market in which that something is bought and sold. But it really doesn't matter, for money creation, what that "something" is. What matters is that the central bank is selling central bank money. It is supplying central bank money. So we want to know something about the central bank's supply function. And that supply function will depend on what it is the central bank is targeting.
We could assume that the central bank is targeting the stock of its own money, so the supply function is perfectly inelastic with respect to anything. That is a very simple assumption, suitable for a first year textbook. But not very realistic, for most times and places.
We could assume that the central bank is targeting the price of gold, so the supply function is perfectly negatively elastic with respect to the price of gold. Realistic in the past, but not nowadays.
We could assume that the central bank is targeting the stock of M1, so the supply function is perfectly negatively elastic with respect to the stock of M1. Realistic briefly in the past, for Canada, but not nowadays.
We could assume that the central bank is targeting expected CPI inflation, so the supply function is perfectly negatively elastic with respect to the expected rate of change of the price of the CPI basket of goods. That is realistic for many central banks nowadays.
Interest rates? Did I hear you say that modern central banks target interest rates? Well, maybe, but only in the very short run, like maybe the next 8 weeks. Monetary policy for the next 8 weeks isn't very interesting, unless it gives us a clue about what the central bank will be targeting in the years after that. And they say they are targeting things like inflation, not interest rates. But if you insist, the central bank's supply function would be perfectly elastic with respect to whatever interest rate you say it is targeting.
See, it's easy. One general theory to rule them all, that can be modified for whatever it is you want to assume the central bank is targeting, just by changing the central bank's supply function.
But what about commercial banks? They create money too. Commercial banks are much easier. We know what commercial banks are targeting. They target maximum profits. And that means commercial banks, like ordinary profit-maximising firms, and like ordinary utility-maximising people, and unlike central banks, only care about real variables. It is the central bank's job to ensure that nominal variables are determinate, by not doing something daft like trying to target a rate of interest for too long.
Just like central banks, commercial banks create money by buying something, and paying for it with the money they create by buying it. They mostly buy non-monetary IOUs, but it doesn't matter what they buy, or even if they just give their money away to their favourite charity.
How much money commercial banks create to maximise their profits will depend on a lot of things. But what I want to focus on, because this is the key policy question, is how it depends on what the central bank is doing. Let me quote again from the Bank of England article:
"The supply of both reserves and currency (which together make up base money) is determined by banks’ demand for reserves both for the settlement of payments and to meet demand for currency from their customers — demand that the central bank typically accommodates.
This demand for base money is therefore more likely to be a consequence rather than a cause of banks making loans and creating broad money."
First, they don't mean "supply"; they mean "quantity supplied". And as I pedantically tell my first year students, the difference really does matter (sometimes, like here). Yes, the quantity supplied (which is equal to quantity demanded in equilibrium) depends on the demand function, but it depends on the supply function too. Both blades of the Marshallian scissors matter in determining quantity, even if one blade is assumed to be horizontal for the next 8 weeks.
But what matters is their acknowledgement that the demand for base money (central bank money) is a consequence of the amount of broad money commercial banks have created. May I make a small simplifying assumption? May I assume that the demand for base money is proportional to the stock of broad money, other (real) things equal? Because that's the only way we can assume that commercial banks maximise profits and so only care about real things and don't suffer from money illusion. Thanks!
"While the money multiplier theory can be a useful way of introducing money and banking in economic textbooks, it is not an accurate description of how money is created in reality. Rather than controlling the quantity of reserves, central banks today typically implement monetary policy by setting the price of reserves — that is, interest rates."
But hang on! You have just agreed (sort of) that the demand for base money is some proportion r of the stock of broad money. So in equilibrium, when the actual stock of base money is equal to the quantity of base money demanded, the stock of broad money must be a multiple 1/r of the stock of base money. And if the central bank shifts the supply function of base money $1 to the right, that must increase the equilibrium stock of broad money by $(1/r). Just like the first-year textbook says it will!
Now you might object that modern central banks don't care about the stock of base money (except when they are doing QE), and target things like inflation instead (except for 8 week periods when they target an interest rate). OK. But if the central bank wanted a temporary increase in the inflation rate, and so a permanent rise in the price level, it would need to shift the supply function of base money, to create a permanent rise in the monetary base, and a permanent rise in broad money, and the textbook money multiplier would tell us that broad money would increase by 1/r times the increase in base money.
One simple (first-year textbook) general theory to rule them all!
What is the underlying problem here? Why do monetary economists resist this very simple and very general theory of monetary policy? The underlying problem is revealed in this quote:
"Like reductions in Bank Rate, asset purchases are a way in which the MPC can loosen the stance of monetary policy in order to stimulate economic activity and meet its inflation target."
It assumes that interest rates are a measure of the "stance of monetary policy". If interest rates were an adequate measure of the "stance of monetary policy", the Bank of England would not need QE. And you cannot define the stance of monetary policy by taking some sort of average of interest rates and QE. A permanent increase in the target price level would mean a permanent increase in the money base but would have no obvious implications for interest rates. A permanent increase in the inflation target would mean a permanent increase in the growth rate of the money base but would mean higher nominal interest rates. There is no monotonic mapping from loose monetary policy into low interest rates. Thinking about monetary policy in terms of interest rate policy just doesn't work. It doesn't work in theory, and it doesn't work in practice. That's why the Bank of England is doing QE, and having to re-introduce the old general theory as a special theory of how monetary policy works.
Worthwhile Canadian Initiative 
Nick, BTW, when visualizing both supply and demand going to zero, I’m seeing a supply curve and a demand curve crossing and going to zero (moving downwards simultaneously). The problem I’m having is that “price” is normally on the x-axis right? But the price of MOA is fixed at unity, so isn’t it a vertical curve? How can we tell if the supply curve ever “goes to zero?”
Tom: economists normally put price on the Y axis, and quantity demanded and supplied on the X axis. (Yes, I know we are weird about that, and it violates all standard conventions, but there are historical reasons why we do it the “wrong” way, and now we are stuck with it.)
The “price” of the MOA is 1/P, where P is the price of goods in terms of the MOA. The vertical axis has (say) 1/CPI. Do the two curves cross at one point, and do they “cross” at 1/P = infinity (P=0)? Scott is assuming you meant the supply curve is vertical at Qs=0, and the demand curve is a rectangular hyperbola, so P=0 is the only solution.
But this is off-topic, and I have other fish to fry.
Odie: “Is it not important to distinguish here between income and loans?”
The distinction between income and loans is obvious.
What is important is to distinguish between the demand for money and the demand for loans. Google “the demand for money”.
ATR: I left a longer comment on your post.
Yep, the Bank of Canada has 3 rates of interest for the 6.5 week period. And when we get into arguments about the BoC setting the rate of interest, and whether the stock of money is demand-determined at the rate of interest, and the role of the central bank’s supply function, we often ignore the two spreads between those 3 rates of interest. Because the BoC nearly always keeps those spreads constant at 25 basis points each side of the overnight rate target.
I see those two spreads as affecting the demand function for reserves. Shift either of those spreads (unless the risks of having positive or negative reserves are zero) and the central bank can shift the demand curve for reserves. But even if we held those two spreads constant, I think I would still be having the exact same argument with the “the supply of money is determined by demand” people.
Nick, thanks… you blew my mind a bit there… I’ll have to think about that more, but back to my quasi-on-topic question above:
So no change in the steady state price level then? (re: my “epsilon” example)
BTW, I really appreciate your help on this.
Thank you for the lengthy response, Nick; it answers a lot but leaves me with much to consider.
Quick clarification – you said: “it is the supply curve of the thing that the bank is buying that interacts with the bank’s supply curve of money”
Is that a typo, or are we really talking about 2 supply curves? I’ve never considered this type of thing.
Nick,
“But hang on! You have just agreed (sort of) that the demand for base money is some proportion r of the stock of broad money. So in equilibrium, when the actual stock of base money is equal to the quantity of base money demanded, the stock of broad money must be a multiple 1/r of the stock of base money. And if the central bank shifts the supply function of base money $1 to the right, that must increase the equilibrium stock of broad money by $(1/r). Just like the first-year textbook says it will!”
Almost all of that base money consists (“normally”) of currency held outside the banks. And the central bank is short the option to provide that currency. The public is long the option to purchase it at par, with bank deposits valued at par (and the commercial banks purchasing it with reserves at par, acting mostly as pass-through agents in this context).
In that sense, the stock of base money (almost all of it normally) is always equal to the quantity of base money demanded – because of that option.
So it seems to me the central bank in effect has no supply curve – other than the demand curve that it must replicate because it is short the option.
How does the central bank shift a supply curve for currency that it doesn’t determine?
I expect you will disagree with this in some way, but could you tell me why – and how would you conceive of the central bank’s supply curve for currency (and therefore for most of the monetary base in normal circumstances)?
thanks
Dustin: that was not a typo. See my new post, for an example of what I mean. (In that new post, the central bank is buying non-monetary IOUs in exchange for money, which means that people are supplying non-monetary IOUs to the central bank, and their supply of non-monetary IOUs, in normal language, is their demand for loans. So it is the central bank’s supply of money, plus the demand for loans, that determines the stock of money. If the bank were buying apples, it would be the supply of apples, plus the supply of money, that determine the stock of money.)
JKH: Here is a way of thinking about it that will be most compatible with your way of thinking about it (I think). (I don’t agree with everything I will say here, and it’s very crude, but that doesn’t really matter for this particular point.)
Put the rate of interest on the vertical axis, and quantity of reserves on the horizontal axis. The central bank has a horizontal supply curve of reserves (it sets a rate of interest). There is a downward-sloping demand curve for reserves. The quantity of reserves is determined where the supply and demand curves cross.
Start in equilibrium. Then the bank decides to increase the price level. It shifts the reserve supply curve vertically down (cuts the target rate of interest), so the quantity of reserves expands, the price level starts to rise, then when the price level is where the central bank wants it to be, it shifts the supply curve back up again. Because the price level is now higher, the demand curve for reserves (which depends on the price level) will also have shifted up/right.
In my own way of thinking, the idea that the supply curve is horizontal doesn’t really work, because if the central bank cares about the price level, which depends upon M, and if the quantity of reserves demanded depends upon M, the supply curve will need to slope up to keep the price level determinate. But that doesn’t really matter for my point here.
Nick,
I asked about currency demanded by the public. Would you treat currency the same way as reserves?
Sorry JKH. I misread you there. But it would be the same. Add the quantity of currency demanded plus the quantity of reserves demanded to get the base money demanded. Then the central bank’s supply curve is the supply curve for base money.
Nick,
Let me set aside the reserve piece for now – for separate consideration.
Suppose we strip out reserves from the total base.
So we have supply and demand for currency to deal with.
That downward sloping demand curve for currency makes intuitive sense (even) to me – liquidity preference and all that stuff (I think).
Suppose we interpret that horizontal curve simply as the current interest rate setting of the CB – and not as a supply curve.
Then I think that amounts to my option interpretation – where the quantity of currency both demanded and supplied is uniquely determined by the intersection of the interest rate and the demand curve, and where the supply curve could be interpreted as essentially lying on top of the demand curve – i.e. matching it because it is essentially a short option position in whatever ends up being the quantity of currency demanded along the demand curve.
Does that make any sense?
In the long run we are all dead.
Nick, you write:
“Roughly speaking, it is not the demand curve for money that interacts with the supply curve to determine the stock of money; it is the supply curve of the thing that the bank is buying that interacts with the bank’s supply curve of money.”
and
“…the central bank is buying non-monetary IOUs in exchange for money, which means that people are supplying non-monetary IOUs to the central bank, and their supply of non-monetary IOUs, in normal language, is their demand for loans.”
As you know, I love the idea of a supply curve for loans (partly) determined by the borrowers. However, my problem with the above is when you bring up the “normal language” replacement for “supply [curve?] of non-monetary IOUs,” namely “demand [curve?] for loans.”
But in the next post when I asked you who determines the “demand [curve] for loans” you said “the borrowers.”
But the borrowers are NOT demanding loans, they are demanding money! I think it’s fair to describe them as either determining the “supply curve for loans” or the “demand curve for money” but NOT the “demand curve for loans.”
What’s wrong with that logic? It’s the bank that should determine the “supply curve for money” or the “demand curve for loans.”
If I’m right though (which I’m 99% sure that not), then we now have TWO demand curves for money: The one that moves to match supply, and the one that helps determine the “stock of money.”
Where have I gone wrong?
Tom: if i want to borrow an extra $100,000 to buy a house, that is a demand for a loan. That does not mean I want to hold an extra $100,000 in cash or in my chequing account from now on. I want to spend it.
Google “The Demand for Money”.
The last line should read:
“The one that eventually moves to match the stock of money supplied (so that stock of money supplied the stock of stock of money demanded) and the one that helps determine the stock of money supplied in the first place.”
I Googled that yesterday, and read the wiki article. Is that the one you recommend? I understand your hot potato logic, I just don’t get the “normal” language: the bank is the one demanding the loans, not the borrowers. All the borrowers have to trade for loans is more loans, so why would they demand them? Say firm X is trading A to firm Y in exchange for B, can’t we say the following four statements apply:
1. X has a supply curve for A
2. X has a demand curve for B
3. Y has a supply curve for B
4. Y has a demand curve for A
X = borrowers
A = loans
Y = banks
B = money
No?
You’re either saying 1. is equivalent to 4., (but 1. is determined by X and 4. by Y., so I don’t think that’s it)
Or you’re adding a 5th line which doesn’t fit the pattern above:
5. X has a demand curve for A
I realize B (money) is special, but it actually changes the symmetry of the statements above?
If X is demanding A, then what are they willing to trade (supply) for it? All they have is A.
The language says they want to trade A for A, but why not keep the A they already had?
… or think of it this way: replace “loans” with “bonds.”
How can
“the bond issuer’s supply curve for bonds”
be equivalent to
“the bond issuer’s demand curve for bonds”
or
“the bank’s (bond buyer’s) demand curve for bonds”
I think I see what you’re saying: there’s not two demand curves. There’s really just one curve we’re talking about here, and it’s a supply curve. It’s just that the so-called “normal” language to describe it is totally screwed up. I’m going to use “bond” for “loan” and “bond issuer” for “borrower” to high-light the silliness of the normal language convention:
So the logical way to say it what you presented 1st:
“The bond issuer has a supply curve for bonds”
The “normal” way to say that makes no sense, but by convention it refers to the same curve above:
“The bond issuer has a demand curve for bonds”
Is that right?
“Here is my general theory: when the central bank buys something, with central bank money, the money supply expands, because whoever sold them that something now holds extra money. Done. It does not matter whether the central bank buys a bond, or a computer, or whatever. Hell, it could just give the money away to its favourite charity (helicopter money), and the result would be the same.”
If the CB does an open market operations (“as buying with central bank money”) it changes it’s balance sheet and thus the amount of reserves. That changes the price of the reserves and the price of the assets purchased. The price (yield) change might stimulates the economy.
Now the helicopter drop is, as I see it, a different beast. It is also a kind of tax relief and thus should be accounted as a fiscal policy tool? The cash will again change the CB’s balance sheet and add the reserves but at the same time it will give more purchasing power (being charity) for the households?
I guess it doesn’t matter so much which sector, banks or household, the government is dealing with but how it affects the private balance sheets. Giving out money, dropping cash, will be a tax break and thus higher private wealth, more aggregate demand. That has a different effect than buying stuff at the market price.
Jussi,
“That changes the price of the reserves and the price of the assets purchased.”
The price of 1 dollar of reserves is fixed at $1 by definition. By what measure are you saying the price changed? By 1/(the general price level)? Measured in 1/$? In other words, do you really mean the value of reserves changes as measured against all other products?
I just meant the interest rate, which I guess can be said to be the price of the reserves, is changed by the higher supply.
“Suppliers of money can “force” more money into existence than people want to hold, because money, unlike other assets, is the medium of exchange.”
I think this is an important idea as this as said would create the hot potato effect.
But I’m not sure I totally get the idea. I can see how currency can be forced to be hold (e.g. tax break / helicopter drop). But that is a small part of the money (even though I admit I’m not sure how it is defined here). So I guess that is not what Nick means here? I think most of the money is deposits but the amount of deposit is dictated by the amount of credit/loans (“loans create deposits”), isn’t it? And loans comes with a price and thus cannot be forced into existence? Can you (anyone) please elaborate? Are the banks part of the suppliers?
Jussi, imagine the central bank buys $10 of assets. Then $10 will be forced into existence that didn’t previously exist. It doesn’t matter who sells the assets. If a bank sells them, the new money will exist as reserves. If a non-bank sells the assets, then both $10 of net new reserves and $10 of net new deposits will be created. But the deposit is both an asset (to the seller) and a liability (to the bank).
In terms of typical aggregates, MB goes up by $10 regardless of who sells the assets, but M1 only changes (increases by $10) if a non-bank sells.
This could happen even if the banks had $0 in loans on their books and $0 in deposits prior to the asset sale.
However, nobody’s equity changed. That’s a difference between a helicopter drop (giveaway from the central bank) and an asset sale. In a helicopter drop, the CB’s equity goes down and the private sector’s equity goes up by the same amount.
Jussi, yes banks are part of the suppliers of broad money (M1). So they can force money into existence in a manner similar to the central bank: buy buying something. It doesn’t matter what they buy. Typically they buy loans from borrowers, because that’s how banks make their money. Think of a borrower as selling an agreement to pay back the principal with interest (almost like the borrower is selling a bond to the bank). But money is created when a bank buys donuts too… or pays it’s electric bill.
The difference between a bank forcing money into existence and a central bank is that a commercial bank’s objective is to maximize profits. It may not be profitable for it to force money into existence. Because the bank doesn’t control everything: the people selling the donuts and the people selling the loans determine the supply curve for those items.
The central bank doesn’t have to worry (too much) about solvency and not at all about profits, so it can buy what it likes w/o regard to whether or not it’s a smart “business move.” But practically they do have to worry about politics, and accumulating too much risk or negative equity could cause congress to do something.
Thanks Tom,
I can see now how the CB can buy things/t-bills from the public and force money into existence. And then sellers will try to get rid of extra cash buy something until the new money is used to pay back a loan and thus destroyed? A hot potato!
What about the case where the CB buys assets from the banks? Banks cannot get rid off the reserves as a sector and thus the forced new base money mainly just sits on the reserve accounts?
Jussi,
“I can see now how the CB can buy things/t-bills from the public and force money into existence. And then sellers will try to get rid of extra cash buy something until the new money is used to pay back a loan and thus destroyed? A hot potato!”
You are correct that it’s possible that the new deposits are used to pay off old loans, however, that’s two things happening. Consider the case where there are no existing deposits or loans. Now if the CB buys $X of reserves, there’ll be $X more money. If the asset sellers were non-banks, then there’ll be $X more deposits too. Since there are not loans to repay, those deposits aren’t going anywhere. And even if there were loans to repay, the $X in reserves aren’t going anywhere… until the CB reverses itself, and sells $X of assets.
“What about the case where the CB buys assets from the banks? Banks cannot get rid off the reserves as a sector and thus the forced new base money mainly just sits on the reserve accounts?”
Whatever the CB buys and whomever it buys it from, this will create reserves at the banks. And those reserves will not go away unless the CB sells the assets again OR deposit holders in the bank remove their deposits in the form of cash.
Take a look at this blog post of mine. There’s an interactive spreadsheet in the middle that you can play with, and I think this should answer all you questions:
http://brown-blog-5.blogspot.com/2013/08/banking-example-11-all-possible-balance.html
So yes, banks can get rid of the reserves: either through selling to the CB (should they be willing to buy), or through cash withdrawals, should depositors be willing to withdraw their deposits in cash. Also there’s the issue of other Fed deposit holders and intergovernmental agencies and foreign trade which I’m ignoring for simplicity (and because I don’t know how to all the accounting for that… I take stab at part of that in the next two blog posts):
Jussi, sorry, but I really screwed this sentence up:
“Now if the CB buys $X of reserves, there’ll be $X more money. If the asset sellers were non-banks, then there’ll be $X more deposits too.”
I should have written:
“Now if the CB buys $X of assets, there’ll be $X more reserves. If the asset sellers were non-banks, then there’ll be $X more deposits too.”
Jussi, my comment above is in reference to a comment that didn’t make it through the spam filter. Perhaps because I included a link to my blog. But you can find the link by Googling the following:
banking example #11
It should be the first on the list. There’s an interactive spreadsheet in the middle that should answer all your questions. Plus I draw out a set of simplified balance sheets with simple formulas in them which should also help (the interactive spreadsheet just implements the formulas)
Nick,
I have expanded the comment I wrote in this thread and turned it into a guest blog post at Insecurity Analyst blog:
http://insecurityanalyst.blogspot.co.uk/2014/03/draghi-is-guiding-interest-rates-yellen_18.html
Has anyone read about the Quantity Theory of Credit by Richard Werner (1992, 1997), see also his book New Paradigm in Macroeconomics, Palgrave Macmillan, 2005, or the article in 2012 International Review of Financial Analysis?