« Where has all the spam gone? | Main | Alberta Juggernaut Continues...For Now »


Feed You can follow this conversation by subscribing to the comment feed for this post.

"Anything whatsoever that changes what one individual does will disrupt that whole pattern of who pays whom when. The optimal order in line will change, and it will take some time for individuals to shuffle their places in the line to re-discover that optimal order."

I've been looking into the idea that a generic reduction in the amount of information moving around (through money) is involved in recessions. This disruption of the ABCDEFG transaction structure could be seen as a specific model/instance of that information loss.

In particular, the order of {A,B,C,D,E,F,G} contains a little more than 12 bits of information that goes away if the order is disrupted.


They all have the same accountant and they all have lines of credit at each others store and then they net out all transactions at some (maybe weekly) interval. No physical "money" needed at all.

Jason: I read your post, or tried to, but I'm afraid you lost me immediately at the start of the second paragraph.

Miami: The physical form of money does not matter, or even if it has any physical form at all. Money is information. Go read Jason, because he's got that bit (pun not intended) right; maybe you can understand the rest of what he says.

What you see and claim is an excess demand for money (not limited to this post) I say is an insufficient supply of money/assets for spending and realizing retirement goals. If people really wanted money why did bond prices and equity prices rise in the US? Because of an excess demand for money?

> If people really wanted money why did bond prices and equity prices rise in the US? Because of an excess demand for money?


If I need to hold money now against a future obligation because of an uncertain or poorly-timed stream of income, I'd like to hold that money instead in an instrument that provides some positive rate of return, which means bonds and equities.

I should have prefaced with a bit more explanation instead of relying on a maze of links. Thanks for giving it a try, Nick :)

A short summary: standard supply and demand curves that intersect at a price P represent "ideal" information transfer where all the information from the demand (ID) reaches the supply (IS), so that ID = IS. In the real world, information is likely to be lost so that ID > IS. This means that the real world price P* falls somewhere in the bottom triangle of the S&D curves which act as an upper bound.

The idea in the post I linked to is that if ε ID = IS with ID > IS and ε < 1, then maybe recessions are periods falling ε (less information than usual getting from demand to supply). The price P* would fall to an even lower value P**. The resulting behavior is similar to Friedman's plucking model, see e.g.


where the straight line is ideal information transfer with ID = IS and the recessions are where ID > IS. This is where I saw your toy model explaining a loss of information -- the order of ABCDEFG has about 12 bits of information in it. But I quickly realized after I posted that the loss of those 12 bits could mean the loss of far more information in the transactions between A and B, etc., themselves.

The model is based on this work:


and is fairly speculative at this point.

John Bryant already modeled somewhat along these lines dont you think ? eg "Coordination, Fragility, High-powered money" among other papers.

If money consists entirely of gold coins then the lower velocity is the more gold coins are needed.

With a good credit system (for example: if members of the chain are prepared to lend one gold coin to whoever needs to make the next purchase at 0% IR) then you can still get by with one gold coin. But if there is only one gold coin and someone loses their nerve and declines to lend the system fails.

If there is bank in the model then in theory that bank could provide all the needed credit with just one coin in reserve. It might take a slice of each transaction as its profit and be one of the participants in the transaction chain when it spends these profits) If the bank got into a panic and cut back on its lending then someone may not be able to complete their transaction and the system fails.

If their was a CB in the model could it address this by buying assets with new money ?

pe: I Googled that John Bryant paper. Could only read the first page from my current location. It looks interesting. I'm not sure whether or not it's along the same lines. My hunch is it's different, but I can't really tell, without reading more.

Miami: because the IS curve slopes up.

Jason: you still lost me. If I trade an apple for a banana, there is supply and demand, but where does information come into it?

TMF: increasing the supply of money, to match the increased demand, would certainly help. If the bank got into trouble, and reduced the supply of money, that would make things worse, of course.

I'm no expert on financial crises, but I find it implausible that crises lead to recessions because they increase the demand for money. If that were true you'd expect central banks to simply accommodate that extra money demand with a greater supply of money, as they do when drug dealers hoard currency. More likely, financial crises lead to recessions because they reduce aggregate supply.

However I do think the recent recession in the US was caused by the Fed's failure to accommodate an increase in the demand for money. This was partly due to the zero bound problem, although not entirely. So your model might well apply to that case.

But unless I'm mistaken it's very unusual for a country to face the zero bound problem during a financial crisis, and hence that can't be the usual reason for the recession. It's either bad monetary policy (falling NGDP), a fall in AS, or both. If someone could find the NGDP data for places like Sweden in the early 1990s, or Thailand and Korea in the late 1990s we might have an answer to the question.

The issue of credit lines that I was talking about on your other post is important here.

In the normal course, agents rely heavily on credit lines to be able to make payments, going into the red if necessary. In a financial crisis, credit lines get cut because people are more concerned about defaults. People have to replace that lost liquidity by holding higher money balances.

I think we can see this in the recent crisis when interbank lending fell dramatically due to credit concerns, leading to a greatly increased requirement for reserve balances.

I very much doubt coordination of the type Nick sets out has anything to do with post crises recessions. Walter Bagehot set out the explanation in the mid-1800s: it’s that the free market left to its own devices creates various forms of quasi-money (bills of exchange in Bagehot’s day). Then crisis hits and suddenly no one trusts bills of exchange: there’s a big collapse in the money supply. Or as Hayek put it:

“There can be no doubt that besides the regular types of the circulating medium, such as coin, notes and bank deposits, which are generally recognised to be money or currency, and the quantity of which is regulated by some central authority or can at least be imagined to be so regulated, there exist still other forms of media of exchange which occasionally or permanently do the service of money. Now while for certain practical purposes we are accustomed to distinguish these forms of media of exchange from money proper as being mere substitutes for money, it is clear that, other things equal, any increase or decrease of these money substitutes will have exactly the same effects as an increase or decrease of the quantity of money proper, and should therefore, for the purposes of theoretical analysis, be counted as money.”

This idea on what is the money supply is made confusing by the idea that banks can supposedly increase the money supply by giving out loans that are deposited in other banks without the money actually changing hands....this then allowing the second bank to loan it our again etc etc.....

This is not real money.....its iou money......and what it contributes to society is allowing the banks to loan out the same money over and over increasing the total interest collected

The interest and principle on the loans of course must ultimately be paid back in real money. ...

Ie money controlled by the goverment (yes the fed et al)....either by printing money oror selling debt or loaning to banks even if the loan is just on paper.....the centralized authority is aware of the money and ultimately controls the supply

They can always print cash to cover the existing debt......if supply is short......but the money supply was expanded when the government incurred the debt... not when thy printed the cash to pay it off...

At least in effect

When investment is delayed either by business personal durable good purchase is that usefully described as a demand for money?
To me it is not, even though of course not investing results in holding more money. As Scott points out, demand for money is easily supplied. It seems to me the driving issue is not wanting to invest, supplying more money is a very indirect way of pushing people to want to invest more.
Your explanation explains the mechanics not the reason. Why not make the spending explicitly investment spending not consumption and ask why does not just a, but abcd all delay investment?

Dan Thorn,

Take two scenarios: a person A spends less and holds more money because they reduce their consumption, and a person B spends less and holds more money because they reduce their investment spending. The reasons for A and B increasing their demanded money balances are different, but that they want to hold money more is common to both.

So I don't think that thinking of the spending change as one of deciding to delay investment is a useful way to think about it.

Scott Sumner,
"But unless I'm mistaken it's very unusual for a country to face the zero bound problem during a financial crisis, and hence that can't be the usual reason for the recession. It's either bad monetary policy (falling NGDP), a fall in AS, or both. If someone could find the NGDP data for places like Sweden in the early 1990s, or Thailand and Korea in the late 1990s we might have an answer to the question."

Central bank policy interest rates were nowhere near zero in all three cases. NGDP fell in Sweden in 1992, and in Thailand and Korea in 1998. Inflation as measured by the GDP implicit price deflator accelerated in Thailand and Korea, so it their case it was also a fall in AS:


As, Nick Edmonds alluded, (short-term) working capital has traditionally solved the coordination problem. The extension of working capital depends on the risk expectations of lenders, which is related to the balance sheet of borrowers. Although Scott Sumner will clearly disagree, and perhaps you as well, this fits fits the simple story http://andolfatto.blogspot.com/2010/07/interpreting-recent-movements-in-money.html ... credit is money, no? ;-}

Scott: yep. The other mechanism is that bank failures (or bank problems) reduce the supply of money, unless the central bank takes offsetting action. And that was important in the 1930's US. But what I'm wondering is whether an increased money demand might be equally important, if the central bank fails to offset it fully.

What surprised me, in working through this toy example, is just how big an increase in money demand we could get, if traders don't coordinate, and how tricky it is for them to coordinate, even in this very simple example.

Nick E: Yep. That makes sense. Trade credit isn't money (unless it is generally accepted) but it may let two traders synchronise their payments and receipts better, and so reduces the demand for money. And as you say, a financial crisis may reduce trade credit through a reduction in trust, and anything that reduces trust, and so reduces trade credit, will probably increase the demand for money too.

It's more than just trade credit. The banks' own liquidity problems cause them to cut back on provision of corporate liquidity facilities, so the corporates need to replace that with bigger cash piles.

It's certainly not the whole story, but I think it's part of it.

The point of bringing in investment is that investment is what fluctuates most. (lets say by crude definition that consumption only declines with declines in income in which case there is no demand for money, while investment spending can change independent of income).
I just think its an identity that cash holding increases. Of course they do, you have income that wasn't spent. How does that situation mean that the people demanded more cash. Nick did offer an explanation, he says well the lower the velocity of money the more money you need. this is well known. if people were paid daily they'd need less cash than if they were paid weekly, than if they were paid monthly, than if they were paid yearly. So sure, Nick's story lengthens the pay period. But why????? why did A suddenly decide to change, and more interestingly why did a bunch of people all decide to do it at once enough to have an aggregate impact. And again to go back to investment spending - the cyclical story is one of lengthening the pay period - spending isn't cancelled, just shifted from today till tomorrow and yes cash bridges the time gap.
I just don't see how demand for cash is a cause - it strikes me as an effect. This is not an assertion, just a question. and even if you want to make a heightened risk aversion the source of the increase for the demand for cash, the interesting question remains: what happened? why the increase in risk aversion?

Mark, Thanks for that info. It seems plausible to me that it was both AD and AS in the Asian crises.

Nick, Yes, it's very plausible to me that there would be a big increase in money demand.

"why the increase in risk aversion?"

This implies a psychological change (not demonstrably rational). The alternative being an expectation of increased risk (rational based on perceived risk). We recently discussed this issue. In practice I expect you see both.

In rational terms you can say:

1) Expectations of the mean future demand decrease (perceived ROI)

2) Expectations of the variance of future demand increase (perceived risk)

3) Availability of credit and expectations of its future availability decrease

This implies a positive feedback effect and changes to the consumption equation to model it. It certainly looks like coordination problems are nonlinear.

"What surprised me, in working through this toy example, is just how big an increase in money demand we could get, if traders don't coordinate, and how tricky it is for them to coordinate, even in this very simple example."

Nick, do you think this might build unexpectedly once in a while like a "rogue wave" on the ocean?

"What surprised me, in working through this toy example, is just how big an increase in money demand we could get, if traders don't coordinate, and how tricky it is for them to coordinate, even in this very simple example."

Coordination requires confidence in the liquidity of counter-parties and their behavior and disclosures. Less confidence means more necessary diligence. Diligence produces delays and increases transaction costs. This directly affects velocity and also indirectly affects it through poorer coordination.

Since the fall in velocity can cause liquidity problems, there can be a feedback effect proportional to the degree of coordination.

Consider the length of chains of counter-parties and the diffusion of risk through securitization of debt and credit default swaps (which creates uncertainty about counter-parties future liquidity).

Lehman and MF Global were both destroyed by doubts about their liquidity, which triggered demands for increased collateral from counter-parties (a cost for continued coordination) and refusal to roll over credit (refusal to coordinate).

An excess demand for money can be described as seeking liquidity, fleeing investment, consumption, and debt.

4) Expectation outstanding debt is no longer self liquidating or sustainable
5) Expectation of asset decline, collateral deficiency, and increased default

Tom: "Nick, do you think this might build unexpectedly once in a while like a "rogue wave" on the ocean?"

Like those studies that traffic engineers do, when the cars are all flowing smoothly, then all of a sudden you get a traffic jam, where they all slow down, even with no obstruction in the road, if the traffic density is high enough.

It sounds plausible to me. The amount of money in the economy is like the average space between cars. If it's big enough, a small delay in payment by one car has little or no effect on the system. If it's too small, a small ripple can build into a big wave, where all the cars stop moving.

I often hear people argue that the amount of money doesn't matter, because each individual can easily and quickly get more money if he wants it. That might be true for the individual, but is it true for the system? "I'm not following to close; I can easily jam on my brakes if I need to!" And then there's a pile-up of 100 cars.

All: thanks for your comments. I don't have much useful to add. I am still playing with this. Not quite sure where to take it.

Leijonhufvud's "corridor" is maybe relevant here. The better coordinated the system, the less money they hold, and the more fragile it is. It can recover from a small shock, but a slightly larger shock will spread.

Suppose initially they are all perfectly coordinated, so V is at a maximum, and M is at a minimum. They shop: A,B,C,D,E,F,G, one a day. If one individual delays his shopping, even by one day, everyone else must delay shopping too. The corridor is very narrow.

If initially they are all shopping in the wrong order, so V is minimised, and M is at a maximum, then one individual can delay his shopping, by up to 6 days, before it makes any difference to anyone else. The corridor is very wide.

And it is precisely these sort of thought-experiments that explain why the distinction between "money" and "credit" is so important. (Not all credit is money). Credit is an individual property (or a relation between two individuals). Money is a system property. What is true for each individual may not be true for the whole.

Nick, if you're looking where to go forward, maybe chat with your engineering colleagues on "queuing theory", esp. if they happen to be designing (internet) packet switches (head-of-line blocking, buffer overflow etc.) There is probably the following analogy: credit=buffer size, coordination=switching/blocking, money=packets.

Recently, in thinking about the value of Bitcoin, assuming there is no hoarding, was guessing Bitcoin$ = (EffectiveFee$)/(VarianceOfBitcoinToCompleteTransaction%) ... in other words if it takes 5 minutes to complete a $<->Bitcoin transaction, and I'm willing to pay $1, and Butcoin's value varies by 1$ in 5 minutes, then the value of Bitcoin = $100. Essentially, its value is it's transactional variance.

In some sense, in the current monetary framework, the Fed control of Fed Funds (pre-2008), was targeting X by controlling the price of credit in the interbank market (buffer size).

"and Bitcoin's value varies by 1% in 5 minutes"

"Credit is an individual property (or a relation between two individuals). Money is a system property."
But credit becomes a system property via the government or banking or other "safe asset" generation. Maybe the financial system matters ;-}.

In practice, using the US as an example, with household net worth at $80T, annual wages of $7T, there is plenty of balance sheet capacity to extend credit ... if they want to. The question then becomes, if the private sector does not want to extend balance sheets, then what should the government do? During the GFC, governments around the world expanded/backed trade credit (e.g. EDC in 2009).

jt: the engineers would probably be good on this. But my math isn't good enough that I would be able to learn from them, because I wouldn't understand them. Instead I'm working on a new post, trying to set up a very simple example, just to illustrate what I'm trying to say. If others with better math find it interesting, they can play with it and get further than I could.

"And it is precisely these sort of thought-experiments that explain why the distinction between "money" and "credit" is so important. (Not all credit is money)"

Only exchangeable credit instruments are sufficiently liquid to be considered money. To be money they have to derive a significant part of their value through provision of liquidity services. Mortgages have a negligible premium and a high cost of exchange (very ineffective as money (but not "bad")). Securitized mortgages have a premium and a much lower exchange cost. GSE guaranteed mortgages have much more moneyness. How much moneyness something needs to be called money depends on term usage context. Context confusion is the source of many disputes about what is or isn't money.

It is possible for something to become money through low information cost, such as a government guarantee or mandate) or cease to be money (through loss of a guarantee, or a fall in the (effective) credit rating of the debtor). This is an example of the cost of diligence (a trade friction).

All money isn't based on credit or guarantee. J.P. Koning has discussed quite a few contrary examples - some quite strange like the Iraqi Swiss Dinar, the Somali shilling, stone money of Yap, cattle (an odd example of commodity money).

If people want to know (much, much) more try


He's good and has links to many other sources.

A financial crisis does not equal increased demand for money, as claimed by Nick. A financial crisis, by definition, is a dramatic reduction in confidence that the so called “money” one has lodged at Lehmans and at shadow banks etc is actually there. That is, a financial crises is a drop in the perceived stock of money.

My preferred solution is to ban private money creation, as suggested by Milton Friedman, Irving Fisher and others: i.e. implement full reserve banking. Since the gyrations in privately created money are largely irrational, i.e. irrational exuberance or lack of, we wouldn’t lose much by such a ban.

@Ralph Musgrave

Full reserve banking won't prevent the private creation of money. People are endlessly inventive when it comes to creating money. It merely needs to be profitable to do it. The liquidity services have to outweigh the opportunity cost. Any ban which raises the value of these services sufficiently will cause the creation of money.

Consider this possibility as the real source of the so called coordination problem, which is a reflection of the problem with the thinking in Chapters 3 and 10 (among others) of the Keynes' General Theory.

Firm A incurs costs to buy inputs (including household X's labor) expecting to sell its outputs for more than the cost of the inputs. When Firm A's costs, contrary to its expectations, exceed its revenue, Firm A's net worth (the real source of its capacity to demand in ensuing cycles) declines as a result. Firm A tends to buy less of household X's labor, and household X's own capacity to demand declines as its income and net worth falls. Neither A nor X buy as much because they no longer can buy as much as other agents in your example had expected to sell. Cost is missing from the example and (revenue below cost) explains the so called coordination failure more satisfactorily. People with rational expectations don't try to make up in volume what they lose on every sale. People with irrational expectations keep trying the same thing until they run out of their own money, lobby for subsidies, and then run out of other people's money too.

The supposed positive relationship between increased spending and increased income does not exist, however, when Seller Costs exceed Seller Revenues. Instead, Seller Net Worth declines, along with Seller’s future purchasing power (including its capacity to demand household X's labor). Under this condition, demand destruction occurs on the producer-seller side, and this demand destruction tends to induce a reduction in aggregate demand to the extent the producer-seller is forced by its balance sheet to reduce spending over one or more ensuing cycles. And, unless Firm A stops producing at a loss (i.e., consuming more than it produces), aggregate demand keeps shrinking. Think Britain spending 2 pounds to mine coal it can only sell for 1). Thank goodness insolvency stops this process in the private sector.

Sales below cost, like building two roads between London and York when one will do, destroy aggregate demand from the supply side. Chapter 3 was new, but it is wrong. The spending to build a road consumes resources, the cost of which are not recovered, unless the economic returns generated by the road over time exceed the cost to build it. An economically productive road only begins its real contribution to economic growth and employment once it is finished and put in use. Once built, the transportation service it provides Farmer (also taxpayer) Jones cuts his travel time and his out-of-pocket travel costs and, by lowering his costs, increases his farming profits even if his gross revenue from farming stays the same. To the extent his transportation cost savings exceed the taxes paid to finance and build the road, the service the road provides in the form of lower transportation costs not only repays Farmer Jones’ investment of tax money but it also leaves extra profits resulting from the lower transportation costs which adds to his disposable income and increases his capacity to demand. Farmer Jones’ standard of living, net income, and net worth all go up, even if his revenue from farming does not. This is the economic mechanism – lowering transportation costs more than construction and financing costs over the time a public asset is in use – that makes public infrastructure spending economically beneficial and contributes to economic growth, greater employment, and greater demand. Spending to build a second bridge does not raise the income of the community; it doubles the cost and halves the yield of public expenditure, lowering growth, employment, demand, and output potential. A road to nowhere strands the resources used to build it in a permanently unproductive use.

Peter N,

You claim that “Full reserve banking won't prevent the private creation of money. People are endlessly inventive when it comes to creating money.”

I agree that whatever laws we make in relation to banks, they’ll try to get round them. But same applies to all laws: e.g. laws relating to speed limits on roads, laws relating to drug dealing, etc. That’s not an argument against laws in relation to speed limits and drugs.

Moreover there is a big difference between private money creation and speed limits, drugs, etc. The difference is that only relatively large organisations can create money so they’re easy to identify and control. Whereas any INDIVIDUAL PERSON can exceed the speed limit or deal in drugs.

To illustrate, IN THEORY I can create money. I can issue IOUs and try to get them accepted. But I’d be wasting my time. In contrast, everyone accepts IOUs issued by a large well known bank.

Ralph Musgrave,

I agree - fix what's easy to fix, and see where you stand.

You have John Cochran on your side, it seems:

" John Cochrane has also argued for a similar system. In his words, “the Federal Reserve should continue to provide abundant reserves to banks, paying market interest. The Treasury could offer reserves to the rest of us—floating-rate, fixed-value, electronically-transferable debt. There is no reason that the Fed and Treasury should artificially starve the economy of completely safe, interest-paying cash”



Nick, you said: "If I trade an apple for a banana, there is supply and demand, but where does information come into it?"

For a single good (bananas), the information lies in the allocations of the of bananas. The initial allocation (endowment) is all of the bananas in the hands of the producers. Different prices result in different final allocations of bananas among consumers, with (hopefully/theoretically) a price that represents an equilibrium allocation of bananas among consumers and producers (either no trading, or "detailed balance" where people still trade bananas, but there is no effect on the overall allocation).

If each 1 represents an economic agent with a banana and each 0 without, the initial allocation is 000000000.... and depending on the price, the final allocation is any sequence 100101111001... (price p), 1111111110010.... (p'p, fewer bananas sold) etc.

The information is the desired allocation (demand) which is transferred to the supply through the market mechanism.

For the two-good apple/banana market, the desired allocation of all the bananas and apples is the "information", thus a trading an apple for a banana changes the distribution of both bananas and apples, and represents a change in the information content.

My prior comment is somewhat of a mess. I decided to take some time to try an explain where the information comes in clearly (with pictures!) here:


The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad