Why an excess demand for money matters so much

Suppose there were an excess demand for antique furniture. Antique furniture is not part of GDP. By Walras Law, if there were an excess demand for antique furniture, there must be an equal and offsetting excess supply of something else, like newly-produced goods for example. Could an excess demand for antique furniture cause a general glut of everything else, a drop in aggregate demand, unemployment, and a recession? Why not?

Why is an excess demand for money (which is also not part of GDP) so different from an excess demand for antique furniture? Most people own a lot more old stuff than they own money, so it can't just be a quantitative difference.

1. Money is the unit of account, and antique furniture isn't. If there is an excess demand for antique furniture, the price of antique furniture must rise to eliminate it. But if there is an excess demand for money, since the price of money is the reciprocal of the price of everything else, the price of everything else must fall to eliminate it. It is a lot easier for the price of antique furniture to rise than the price of everything to fall.

2. That wasn't the most important reason. This is. Money is the medium of exchange, and antique furniture isn't.

Suppose there is an excess demand for antique furniture, and the price of antique furniture cannot rise to eliminate that excess demand. (Just bear with me, and use your imagination to think up stupid legal price controls on antiques). Everybody plans to sell more newly-produced goods than they buy, and use the savings to buy more antique furniture. Walras Law says that the value of the excess demand for antique furniture equals the value of the excess supply of newly-produced goods. So according to Walras Law, price controls on antiques, and an excess demand for antiques, could cause a recession. Why not?

Walras Law is wrong. Or rather, Walras Law is only right for notional excess demands and supplies. Notional demands and supplies are the quantities people plan to buy and sell under the assumption they will be able to go to the market and actually buy and sell those quantities. But if there is an excess demand for antiques, they will go to the antique market and find there is a shortage of antiques, and they can't buy as many as they want. Not enough willing sellers, because by assumption there's an excess demand for antiques. Notional demands and supplies are irrelevant in disequilibrium, because they are plans made under an assumption that people will learn is false.

Frustrated in the antique market, people revise their plans in the market for newly-produced goods, taking that constraint on antique purchases into account. Unable to satisfy their excess demand for antiques, they will revise their plans to buy less newly-produced goods than they sell. Unless they plan to hold those savings in money, creating an excess demand for money, they must now plan to buy exactly the same value of newly-produced goods as they sell. The excess supply of newly-produced goods disappears.

An excess demand for antiques logically cannot create an excess supply of newly-produced goods (unless it creates an excess demand for money as a side-effect). It would violate the budget constraint if it did.

But an excess demand for the medium of exchange is different. Suppose everybody wants to add to his stock of money, rather than add to his stock of antiques. And suppose the price of money (the inverse of the price of everything else) cannot adjust. They plan to go to the market for newly-produced goods (and the market for antiques) and sell more than they buy in all those other markets. Every market is the market for the medium of exchange. They fail, of course, because it is impossible for everybody to do the same thing. They are unable to sell as much as they wanted, and so revise their plans of how much they will buy, taking that constraint into account. They decide to buy even less, which means (since everyone is doing the same), they find they can sell even less....

The Keynesian multiplier process continues until sales, output, and income drop enough that people decide to stop trying to add to their stocks of money. The recession gets worse until the excess demand for money disappears.

Problems in the market for antique furniture cannot cause a general excess supply of goods, unless they cause an excess demand for money as a side-effect.

Similarly, problems in the market for anything else, like mortgages, asset-backed commercial paper, credit default swaps, other fancy derivatives, or pet rocks, cannot cause a general excess supply of goods, unless they cause an excess demand for money as a side-effect.

This recession is a monetary problem. The cure must be to increase the supply or reduce the demand for money.

(Nothing here is really new, of course. Clower for example knew this stuff 40 years ago. But people forget, or are too young to remember, or never learned it in the first place.)

Comments

[JKH]

Nick,

If you think a post from me is worthwhile, I’d be pleased if you were to do it by way of a comment hoist.

Options for comment hoist:

a) Raw comment hoist (preferred); can be referenced from your context

b) Re-edited comment hoist, or “cut and paste” (if value added)

I’d rather not construct something from scratch.

No rush. Whatever you feel comfortable with, whenever.

[JKH]

Nick,

(My first response apparently got swallowed up by the machine.)

If you think a post from me is worthwhile, I’d be pleased if you were to do it way of a comment hoist.

Options for comment hoist:

a) Raw comment hoist (preferred); can be referenced from your context

b) Re-edited comment hoist, or “cut and paste” (if value added)

I’d rather not construct something from scratch.

No rush. Whatever you feel comfortable with, whenever.

[RebelEconomist]

"piling up M1": Suppose that easing progresses to the point that the central bank lends to the government, which the government distributes as a gift to all citizens paid into their current accounts. Does QE fail if the banks are not inclined to use these deposits to lend? No, because the increased stock of M1 has some influence via Q. In this case, the banks would be doing no more than transforming base money into a more convenient form.

"the point is moot....": In order to boost M1 directly, the central bank seeks to buy assets from non-banks even when a bank might have offered the central bank a better deal. But this would be futile if the banks simply sold similar assets to the non-banks. The remark about callable liabilities was just an attempt to come up with reason why the banks might do this.

As you can probably tell, I am close to the frontier of my understanding here!

[JKH]

RebelEconomist,

The monetary effect of your first example looks to me like:

a) CB purchases government bonds
b) Proceeds are used to credit (M1) bank accounts of taxpayers
c) Bank reserve deposits increase

These system balance sheet entries are co-dependent, serially dependent and circularly dependent:

M1 “funds” the increase in reserves. The increase in reserves “funds” the purchase of government bonds. And the purchase of government bonds “funds” the increase in M1.

There is no necessary further effect on the composition of the system balance sheet after that. The additional reserves lay fallow, earning interest paid by the Fed. Individual banks may or may not be tempted to redistribute their shares of these reserves by purchasing assets, depending on asset pricing and capital constraints. But there’s no guaranteed effect through that channel, as per my previous comment.

The more interesting question again is what happens to those additional M1 balances? There may be Q and V effects from that. I guess your phrase “piling up M1” puzzled me a bit. That’s really the expected effect of QE via purchases from non-banks. Thinking about it further, maybe I’m saying that a QE M1 transmission effect is more likely to affect GDP than a QE reserve only transmission effect.

Combinations and permutations:

Buyer, Seller, Reserves, M1,

CB, bank, up, neutral,
CB, non-bank, up, up,
Bank, CB, down, neutral,
Bank, non-bank, neutral, up,
Non-bank, CB, down, down,
Non-bank, Bank, neutral, down,

[RebelEconomist]

"a QE M1 transmission effect is more likely to affect GDP than a QE reserve only transmission effect"

Maybe not, because however the assets are purchased, the private sector can trade itself into its preferred position.

Central bank buys assets from banks, banks buy assets from non-banks (ie loans) is how easing normally works, and works because holding lots of reserves is the not the private sector's preferred position. It seems to me that it can work the other way round.

[Nick Rowe]

JKH @12.36: That's exactly how I was thinking too. To say the same thing another way, if the CB increases base money through the normal way, the banking multiplier to M1 is zero. If instead it circumvents the banks, the banking multiplier to M1 is one.

There will probably be additional induced effects on M1 as macroeconomic variables respond to the initial change in M1, and banks respond to those macro variables, but that is another story.

[JKH]

RebelEconomist,

Normal easing (non QE) has little effect on M1. On a cumulative basis, most of the base ends up as currency held outside the banks. On a marginal basis, the change in bank reserves is negligible in the context of the size of the banking system balance sheet.

On the other hand, with CRE/QE the Fed is pushing excess reserves toward $ 1 trillion, which is very substantial. So the choice of the monetization route is consequential for the system balance sheet. And credit easing opens up numerous possibilities for non-bank channels.

But I agree in theory that the private sector can trade itself into its preferred position. That’s the roundabout route I referred to in my first response to you. So re my comment quoted above, it would affect GDP, but with a lag. In today’s environment though, such lending is capital constrained to an unusual degree. That’s why QE M1 easing would work faster than the equivalent roundabout route.

[JKH]

Nick,

You are getting close to reading my mind.

This is a nuanced use of the notion of “multiplier” that I’ve also intended to mention.

I agree that QE via banks direct gives a first order M1 multiplier of zero.

And QE via non-banks gives a first order M1 multiplier of one.

You’re probably now aware that I don’t subscribe to the textbook dynamic of higher orders of multiplier, primarily because bank lending is capital constrained rather than reserve constrained. But this first order effect as we talk about it here is the direct result of the central bank’s intervention; not the presumed result of increased reserves in terms of further bank system lending.

[JKH]

RebelEconomist,

Re my 3:04,

I guess M1 includes currency in circulation. I keep forgetting that, if its the case.

What I really mean by M1 in these discussions is M1 bank deposits.

[RebelEconomist]

I never should have mentioned normal easing; it just confused the issue!

Clearly, the central bank buying assets, say MBS, from non-banks increases deposit money directly, but (1) is it likely that banks will take advantage of the extra demand for MBS to reduce their own holdings of MBS and use the proceeds to cut back their liabilities (eg to improve their capital ratios)? and if so, (2) how quickly would that occur?

[JKH]

RebelEconomist,

First, it’s not necessary for banks to “cut back” on M1 liabilities in order to improve capital ratios, although I don’t know that you necessarily implied it was. They can simply sell to the Fed, replacing risky assets with excess reserves. This improves risk weighted capital ratios, the same as it would if M1 deposits were dropped off. Excess reserves are zero risk weighted.

Along the lines of your question, banks may reduce M1 liabilities if they sell assets to non banks. But you mention the central bank as the source of the demand for the assets, so this might suggest the further brokering of bank sales to the central bank by the non banks. This would reverse your M1 effect.

Further to your point, it’s also possible that banks sell to non banks that retain those assets in portfolio. This is more or less the intention of PPIP. That would indeed have the type of M1 effect you describe.

All that said, the Fed wants to see the banking system resume healthy lending on the basis of a strong capital position. The type of transaction to which you refer frees up bank capital for other purposes. (Substantial capital is allocated to these so-called “toxic” assets.) In addition, the purpose of the stress test exercise, whose results were released today, is to ensure adequate capital in total. To the degree that banks can resume healthy lending, this activity will replenish the type of M1 decline to which you refer.

This entire stew is happening under the shadow of the Fed, which has indicated its intention to do whatever is necessary to ensure a strong banking system and healthy economic recovery, and to use balance sheet expansion and CRE/QE easing as necessary in order to achieve this.

Such Fed balance sheet expansion has an effect on the banking system balance sheet, gross and net, by definition. The entire effect in total obviously depends on the flows happening that are initiated outside of the Fed. This is difficult to predict. But the Fed will respond taking these flows into account.

And if the channel for the related CRE/QE includes M1 expansion, it will increase M1 from what it otherwise would be. And if it doesn't, it won't. But it will still increase reserves from what they otherwise would be.

Whatever the Fed does will make a difference. I think that’s more to the point here.

[Antique Cabinets]

I think if people midn set was to buy for what you have in hand we probably would not in this recession but the society we are iving in forces us not to think that way but rather to to think that something beyong our budget is reachable.