Would the current financial crisis matter as much in a world without money?
Let me be more specific. Imagine we lived in a world where we still had money as a medium of account, so prices were measured in money. But people did not use any medium of exchange; they used barter instead.
Would your (explicit or implicit) model of the crisis need to be changed in any way? Would monetary policy matter any less?
How does exchange happen in a world without a medium of exchange? Imagine that Ebay and Craiglist merge into some giant centralised barter exchange. Every day, every person who wants to do any trades would submit a list of offers to sell and buy to the central exchange. The centralised exchange would try to match buyers and sellers, not on a bilateral basis, but on a multilateral basis (it is not constrained by needing to find a double-coincidence of wants).
How are prices determined? If you think that prices are perfectly flexible, then you can imagine the central exchange also adjusts prices to equilibrate demands and supplies (and people will need to submit their demand and supply functions to the central exchange, so it can solve for the eqilibrium prices). If you think that prices are set by firms, then you can imagine that firms announce their prices to the central exchange before everyone submits their demands and supplies. Just continue to make the same assumption about price-determination as you make now.
If you assume prices are sticky, then some markets won't be in equilibrium. If so, the central exchange will need to tell people they cannot buy or sell what they wanted to buy and sell, and must revise their demands and supplies accordingly.
The one restriction we must impose is that the central exchange insists that each individual's trades must balance, each day. You cannot accumulate credits or debits at the central exchange. Otherwise, those credits or debits would be a new medium of exchange, and the central exchange would become a central bank. This does not prevent individuals borrowing and lending to each other, of course.
If intrinsically worthless paper money continues to be used as a medium of account, there must be some sort of demand to hold money, even though it's not used as a medium of exchange. So suppose people demand money as "bling": they wear great wads of currency to show off their wealth and status. Naturally, what matters is the real value of the bling, not the face value of the notes. So the demand for money-as-bling is a demand for real money balances. And the demand for real money-as-bling probably depends positively on income, and negatively on the nominal rate of interest, as an opportunity cost of holding money-as-bling.
So we can write the demand function for money-as-bling as Md/P=L(y,i). So far, no different from the normal model. We can even let a central bank determine the nominal quantity of money-as-bling. Or we could let the central bank rent out unlimited quantities of money-as-bling at a nominal interest rate of its choosing. The LM curve looks exactly the same in a world without a medium of exchange.
What about the IS curve? People's intertemporal consumption choices should be unaffected by the absence of a medium of exchange, and its replacement by an equally efficient barter technology. Similarly, firms' investment choices should be unaffected. So if you view the IS curve as reflecting those real intertemporal spending decisions, it too should be unaffected by the elimination of a medium of exchange.
If prices and wages are fixed in terms of money-as-bling, the medium of account, a monetary policy which halved the amount of money-as-bling should shift the LM curve left, and cause mass unemployment.
And if the IS curve went below the axis at full employment, so we have a negative natural rate of interest, that would also cause mass unemployment (unless expected inflation were high enough). Everybody would want to save in the form of money-as-bling, even at zero nominal interest rates.
Convinced? Imagine there's no money……and nothing really changes.
But everything does change. In a world without a medium of exchange, the involuntarily unemployed workers with an excess supply of labour would just barter their labour directly for firms' excess supplies of goods, as long as the real wage were at the correct value. A general glut of labour and goods is impossible. It doesn't matter if there's an excess demand for money-as-bling because the real supply of money-as-bling is too low, or because the natural rate of interest is negative. It won't cause unemployment.
So if your explicit or implicit model tells you nothing is changed, there must something very wrong with your model.
Worthwhile Canadian Initiative 
Nick, In the first response with bling, can I assume you accept that if there is an excess demand for bling, you get unemployment? I.e. if people want to buy $1 billion more in bling than are being offered for sale, then there will be $1 billion excess supply of all other goods, and suppliers will then reduce the supply of all other goods by $1 billion below the optimal level.
Am I correct in assuming that you are also arguing the excess supply would be greater if there were $1 billion in excess supply for cash (the medium of exchange) because it would also interrupt a lot of transactions that otherwise would have gone through with barter if bling was the unit of account? I take this to be your argument, but just want to make sure.
By picking the 1929-30 example I was trying to force you out of your comfort zone, where short run prices are completely rigid and nominal monetary contractions equal real monetary contractions which reduce the real stock of the medium of exchange. Unfortunately I don’t really understand these models well enough to have any confidence in hypothetical outcomes for the thought experiments that you propose. But the following example might give you a better idea of what I think is a typical “monetary shock.”
In 1933 the Fed devalues the dollar sharply. Prices rise sharply for four months while nominal wages are stable. Industrial output rises 57%. Little or no change in the money stock, and a fall in real money balances. Why did output soar? Was it because there was more medium of exchange? Or because the monetary shock increased the expected future money supply—the money supply in out years by which time wages and prices would become flexible? In my view the big drop in real wages (as industrial goods prices soared and nominal wages were sticky) provided almost all of the output boost. If nominal prices had been constant, real wages wouldn’t have changed, and as I already indicated nominal money didn’t change right away, so what would have caused output to rise?
Now I suppose 1933 was an odd year, but in my view it simply showed more clearly what always happens. Major monetary shocks often don’t change the current money supply much, but rather change the expected future path of M. That changes asset prices right now. The current recession is an example. The recession isn’t because cash or M1 or M2 went down, it’s because the expected future money supply 5 years out fell, reducing the expected future price level 5 years out (very clearly evident in TIPS spreads), and the stock market decided quite rationally that if inflation is coming to a screeching stop, and wages are sticky, then profits will plunge and unemployment will skyrocket.
I’m not trying to dodge your question, rather I don’t understand the Barro-Grossman model well enough to have any confidence in my explanation. Instead I am trying to give my sense of what is important. It’s a typical Chicago approach of assuming perfect competition sets prices (goods or assets) and wages are the only sticky “price.” My hunch is that my view is wrong precisely to the extent that monopolistic competition is important, and thus sticky prices are important in the transmission mechanism. With sticky prices you get tight money reducing the real quantity of media of exchange, and that’s probably where you get the mechanism that you focus on and that I overlook. But that’s just a hunch.
Scott:
“Nick, In the first response with bling, can I assume you accept that if there is an excess demand for bling, you get unemployment? I.e. if people want to buy $1 billion more in bling than are being offered for sale, then there will be $1 billion excess supply of all other goods, and suppliers will then reduce the supply of all other goods by $1 billion below the optimal level.”
NO, I certainly do not accept that. (Yes, I am violating Walras Law). To a first order approximation, and the most likely scenario, there will be $0 excess supply of other goods and zero unemployment. This is the really important part of my argument. $1 billion excess demand for bling; $0 excess supply of other goods!
People try to buy $1 billion more bling and $1 billion less other goods. They find they can’t buy any more bling (nobody sells), so they give up trying to buy $1 billion less of other goods. So the $1 billion excess supply of other goods immediately disappears.
“Am I correct in assuming that you are also arguing the excess supply would be greater if there were $1 billion in excess supply for cash (the medium of exchange) because it would also interrupt a lot of transactions that otherwise would have gone through with barter if bling was the unit of account? I take this to be your argument, but just want to make sure.”
YES. The final excess supply of other goods would be a lot greater than $1 billion, given the multiplier process.
That’s the most likely scenario. I would only change my mind if I thought that, unable to buy more bling, they would decide that the medium of exchange is the closest substitute to bling, and so decided to demand an extra $1 billion medium of exchange as a second best.
Let me think about 1933. Yes, you have got me outside of my comfort zone 🙂
In a nutshell: if people try to buy $1 more of good x, and $1 less of good y, and find they can’t buy $1 more of good x, they buy the original quantity of good y instead, as a second best.
This works for any good x, except the medium of exchange.
1933: here’s my interpretation.
We start out with fixed money wages, and an excess supply of labour. But the price level has eventually adjusted to its equilibrium, given fixed wages, so demand=supply for goods.
If we want an increase in output and employment, we need two things in this case:
since W/P = MPL, and MPL is diminishing in L? we need a fall in W/P.
A fall in W/P is necessary, but not sufficient. We also need an increased demand for goods. One way to get this increased demand for goods would be through an increase in the supply of M/P, either through an increase in M or a fall in P. But you say neither happened. The other way to get an increased demand for goods is through a fall in demand for M/P. Maybe this is what happened? The increased inflation raised expected inflation which reduced the demand for M/P?
That’s the only way I can make sense of it. And it’s much the same as your way of making sense of it.
I do like macroeconomics with monopolistic competition. It’s the only way I can make sense of the world. And I believe both wages and prices are sticky.
Nick, A few quick comments. The question of which model is best depends to some extent about what we assume about both monetary shocks and wage/price/asset stickiness. Thus your model may work best if one assume rigid wages and prices and an immediate change in real balances, but might work less well under a situation where the expected future path of money changes, and wages are rigid, and commodity and asset prices change quite a bit in the short run. In the latter case the relative price adjustments might well be more important that the medium of exchange effect.
2. You statement that the Walrasian assumption is wrong seems pretty bold. Is it as bold as it seems? And if so have you tried it out in conferences, journals, etc, to get feedback? If it is not as bold as it seems, is it because of some easily explainable trick? I still don’t quite see why one can’t treat all other goods as a sort of composite, and end up with excess supply of all other goods. But perhaps I should drop this as it isn’t my forte. That’s why I wondered what other people thought of the idea.
BTW, I agree about the paradox of thrift. It is too much demand for money, not saving. The whole GT has serious problems accounting for the price level. It’s a weird sort of real model claiming to be a nominal model.
Scott:
1. Let’s write money supply equals money demand as M=L(Y,P,i). If M falls, and P and i don’t change, we get the direct effect on Y. If P adjusts down, and/or i adjusts up, that means the effect on Y will be lessened. Fixed W (for example), prevent P from adjusting as much as it needs to, because that would violate the MPL=W/P condition for labour demand.
So what you think of a fall in M causing relative price effects, which in turn cause the fall in Y, I think of as partial flexibility of some prices, which ameliorate the effect of M on Y.
My point about Walras Law is as bold as it seems. There’s no trick. But it’s not new. Clower knew it way back. So did a lot of economists in that early 1970’s disequilibrium macro approach. Benassy knew it best. Peter Howitt taught it to me. Walras Law is true for notional excess demands. But in disequilibrium, people are quantity constrained, and reformulate their demands and supplies taking those quantity constraints into account. And Walras Law is not true for constrained excess demands. Because there’s a different utility-maximisation problem in each market, because you consider the quantity constraints in all other markets, but not in the market where you are currently trying to buy or sell stuff. n-1 markets, give n-1 separate utility-maximisation problems.
It all got forgotten in the 1970s, when people forgot disequilibrium. God, am I the only person alive who remembers it?
If you accept that the paradox of thrift is really about excess demand for money rather than saving, you really accept my point about Walras Law!
There is one other way to salvage Walras Law.
Suppose there’s an initial excess demand for bling and excess supply of other goods. But people are unable to satisfy their excess demand for bling, so buy other goods instead.
I would describe that as the excess demand for bling remains, but the excess supply of other goods disappears. So Walras Law is false.
But you could also say (as Dreze said, IIRC) that people give up trying to buy bling, so the excess demand for bling disappears too. (Think of it as the “discouraged bling buyer hypothesis” like the discouraged worker hypothesis which says that some unemployed workers give up looking.
Unfortunately, this interpretation has the unfortunate consequence that all markets will always have zero excess demand, even if prices are fixed. If there is excess demand buyers give up trying to buy; if there is excess supply, sellers give up trying to sell, so the excess demand or supply always disappears without prices needing to change.
Scott:
You are in error. You are just too used to “assuming” that product prices are flexible and wages aren’t. A decrease in aggregate expenditures causes product prices to fall in response to the lower demands of products and the resulting surpluses. However, wages are sticky. So, real wages rise. And there is a decrease in the real demand for labor.
Fine. But how hard is it to imagine that product prices don’t drop. Perhaps there are binding legal price floors on everything. It doesn’t matter.
Then falling nominal expenditures results in falling demands for products. Firms sell less and produce less. The hire less labor beause they need less.
Now, if wages fell and prices didn’t fall, then it is true that maybe firms would substitute labor for capital a bit, but they wouldn’t produce more output because they couldn’t sell it. Because by assumption, then didn’t lower their prices.
Patinkin covers all of this pretty well.
Nick:
The money as bling is just confusing people. Of course, I read the comments and it seems like there are people who read this as if you are advocating that we get rid of money.
Forget the quantity theory business. Barter economy. What happens when there is an excess demand for equities matched by an excess supply of currently produced commodities? Or bonds?
I think the answer is that frustrated buyers shift to other goods.
Suppose people must borrow much less and pay down their debts? Suppose there are a lot of bankrupctcies?
Looking at what would happen in a barter economy would help show that none of these things cause “general gluts” except to the
degree they impact the supply and demand for the medium of exchange.
For folks like Scott and I, what happens when there is an excess demand for the medium of account? The price level needs to drop to clear that market. But, if it doesn’t, frustrated medium of account buyers purchase other goods.
Anyway, I think you are on to something about the problem necessarily being the medium of exchange and not the medium of account.
Scott:
You know that your Depression story of an increase in the price level and lower real wage, and a large increase in industrial production involved a large decrease in the demand for money.
It is the quantity of money and the demand to hold the quantity of money that matters.
Sometimes I am wondering if you don’t drift back into the fixed velocity, changes in M2 determine nominal income version of monetarism.
To say that a gold devaluation does not impact M2 doesn’t mean that it has no monetary consqeuences. If it is going to impact nominal expenditure in the economy, it must have monetary consequences. Expected future nominal income impacts current nominal income through changes in the current demand for money. And there are limits to this effect. Imagine the money supply is near zero now. And is expected to be really high next year so that nominal income is high. There is no way that nominal income will be high now because there is next to no money to spend.
Bill: I am so glad you found your way to this post, and comments. I was hoping you would.
I am very relieved to see someone understand what I’m talking about.
But maybe (unfortunately) you understood what I was talking about because you knew it already? You clearly have at least some familiarity with this literature, from your discussion of the vertical labour demand curve when firms are sales-constrained in the output market. (a la Patinkin, then later Barro and Grossman).
“Anyway, I think you are on to something about the problem necessarily being the medium of exchange and not the medium of account.”
I have made several posts trying to get this point across, in various ways. This was one direct attempt: http://worthwhile.typepad.com/worthwhile_canadian_initi/2009/05/why-an-excess-demand-for-money-matters-so-much.html?cid=6a00d83451688169e201156f7bab5e970c
The same point crops up again when I look at the IS curve: http://worthwhile.typepad.com/worthwhile_canadian_initi/2009/05/the-is-curve.html
To my mind, the whole multiplier process should be understood as the consequences of an excess demand for money, not an excess demand for “savings”. Only a desire to save in the form of money causes the multiplier process. And by “money” here I mean the medium of exchange.
I’m really not sure what to do with this argument, and would appreciate your advice. Is it just a minor “dark age”, where people used to understand this stuff, but nearly everyone has now forgotten it? Or is it partly new?
And you think that “bling” doesn’t really work to help people understand it.
I could try a head-on frontal assault on Walras Law. It would go something like this:
In an economy with n goods (including money, if it exists), Walras Law says that the value of the n excess demands sums to 0.
But in a monetary exchange economy there are only n-1 markets.
But in each market there are 2 goods traded. If a market is in disequilibrium, one of the 2 goods is in excess supply, and the other is in excess demand. So there are 2(n-1) excess demands. So how can Walras Law say there are n excess demands, that add to zero?
Etc. What do you think? Is this new to you? Worth exploring?
Nick,
Your Walras law argument doesn’t further your case becaue nobody is denying it. We all agree that the problem is that the excess saving is being held as money instead of flowing into real assets. We all also agree that this is a problem exactly because money is the medium of exchange.
We only disagree on one point. By phrasing it as an excess demand for money you appear to imply that supplying enough money, only for a short time, can stimulate now while still avoiding inflation later. This is simply false.
There is no amount of money that won’t just be held at current rates of return. The only way to stimulate is to lower the expected real rate of return to holding money.
Thanks Adam. That helped me.
I agree those are two separate questions. The policy “increase the money supply” does not follow immediately from “a general glut is an excess demand for medium of exchange”. They might be linked (I think they are), but you can believe one without the other. I need to keep them separate, and try perhaps to make a separate argument for the link.
Still, I’m not sure what to do with my Walras Law/Medium of Exchange argument. Leave it be (I’ve already made the point)? Or try to develop it, perhaps just by making a post summarising the whole thing, since some of the points only came out in response to comments?
Well Nick, that’s the thing. My comments of your Walras Law/medium of exchange argument have been of the form “Nick you’re missing the point”, not “this is logically flawed”.
Basically, the primitive demand is for real savings and money is the chosen medium only because it’s risk-return profile looks better than the available real investment options.
Thus, the reason I prefer the phrase “excess real savings demand” is because money is being held because it has a risk-return profile that is, right now, prefered to real investment as a medim of savings. Thus, the solution is to make it’s risk-return profile less attractive. Promising inflation lowers the return, taking large amounts of systematic risk on the central banks balance sheet makes money riskier (in the sense that if things go really bad the central bank must inflate to stay solvent, it’s a state contingent promise to inflate).
This has everything to do with the supply of money expected in the future, perhaps contingent on the prevailing state of the world. It has nothing at all to do with the money supply today.
That said, I never said that if money was not the medium of exchange then any of this would be a problem. Like I said your post ‘could the natural real rate really be negative?’, nobody says a negative real rate would be a problem in a world without money. However, in a world that has money full employment will always require that the real return on money remain below the natrual real interest rate (or money is given some aggregate risk so it acquires a real risk premium).
I occurs to me that the whole crux of my argument is stated in the last sentance of the above post and so I should emphasize it:
IN A WORLD WITH MONEY FULL EMPLOYMENT REQUIRES THAT THE EXPECTED REAL RETURN ON MONEY STAY BELOW THE NATURAL REAL INTEREST RATE (OR MONEY BE GIVEN SOME AGGREGATE RISK SO IT ACQUIRES A REAL RISK PREMIUM).
The point is that it’s all about risk and return of money vs real investment. Nothing at all to do with the quantity of money, except in so far as the time path of the money supply determines its real return.
Nick,
My claim that monetary policy can’t break a liquidity trap without accommodating a certain amount of inflation is explained here:
http://canucksanonymous.blogspot.com/2009/06/phillips-curve-in-liquidity-trap_01.html