« Economy Lab: Why economic forecasting goes awry | Main | Firefighting is not a public good »


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

Is the original reason he was wrong (money demand isn't stable enough) still there, or does this insight into possible dark matter mean that original reason was illusionary? If not, then he's unambiguously more wrong, because we've got the original reason + a new reason (even harder to implement).

Luis: It's more a different reason. Maybe what we thought were shifts in money demand were really shifts in money supply -- just a shift in part of the money supply we didn't see. So the original reason we thought he's wrong is false; but there's a different reason.

So under the lamp post, the complexity of the stock composition of money becomes more obvious.

Should that really provide comfort that the volatility of velocity is any less erratic?

Given the specialization and complexity inherent in the stock, why wouldn’t velocity be even more erratic?

Maybe he was even more wrong than you thought.

If the natural rate hypothesis is orthodoxy, wouldn't papers arguing for hysteresis be heterodox? They seem fairly mainstream to me. And Mark Thoma is hardly alone in doubting whether monetary policy should have prime responsibility for aggregate demand and inflation. Actually the proponents of monetary policy are frantically moving the goalposts, defining monetary policy as anything the Fed or the ECB might conceivably do.

Any policy that depends on using the 'demand for money' as a driving input is even more inherently flawed than one that uses the supply of money. In neither case can one produce an unambiguously useful value, even ignoring the question of definitions.

If I routinely hold an average of $10,000 of money (the medium of exchange), I can be said to have a demand for money of that amount, but it is useless since the total is composed of at least three distinct components with both disparate purposes and economic effects.

First is money held for unpredictable purchases in an uncertain future, where an actual medium of exchange is required for an immediate payment. Think of a sandwich from a vending truck. Even if a credit card payment were sometimes accepted, you couldn't depend on it in advance.

Second is money held for predictable purchases. In this case, think of a rent payment. Even if real money is needed, there is no need to hold it for weeks in advance when some other liquid asset can be converted to money in time to make the payment. Note that a monthly rent payment interacts with the supply of money not only due to the amount, but at least as importantly, the time of possession of actual money in advance of payment. A dollar that I possess cannot be possessed by anyone else at the same time.

Third is a remnant of money for which no preferred alternative is perceived, after any potential transaction costs.

Goods and services are always ultimately exchanged for other goods and services. Money serves as a buffering and time shifting mechanism and is not consumed in exchanges. It is the time of possession of money which is used up.

Regards, Don Lloyd

European Central Bank still pays lip service to Friedman in its monetary policy announcements. This is Trichet in October 2009:
"Turning to the monetary analysis, the latest data confirm that developments in broad money and credit growth remain subdued. In August, the annual growth of M3 and loans to the private sector declined further to historically low rates of 2.5% and 0.1% respectively. This parallel deceleration in money and credit growth confirms our previous assessment of a moderate underlying pace of monetary expansion and low inflationary pressures over the medium term."

Here is Jürgen Stark, Member of the Executive Board of the ECB:
"On the basis of these considerations, over the dozen years since the creation of the ECB, we have continued to employ and develop new tools for monetary analysis. We did this even in the face of vocal criticism from a significant camp of economists. They questioned whether money and financial frictions had a role to play in the inflation targeting frameworks that they had identified as “best practice” for central banking.

Money has been ignored. In the canonical New Keynesian macro-model on which the intellectual foundations of inflation targeting rests, money has been considered a redundant element in the monetary transmission mechanism. At best, money is seen as a useless appendix to the model, serving only to confuse and distract any policy-maker misguided enough to consider it.

Now that the financial crisis has exposed the fault lines underlying this model;

now that it has been recognised that the inflation targeting approach focused unduly on short-term cyclical developments in the real economy;

now that economists agree that insufficient attention has been paid to financial imbalances and vulnerabilities, critics of the ECB have acknowledged the benefits of monetary analysis.

Yet, they tell us that we have pursued such analysis for the wrong purpose!

The right purpose, they tell us, is to support a broader view of our mandate, paying more attention to risks to financial stability in formulating our policy decisions. It is a little bit as if they are telling us to forget about monetary policy in principle."

Jürgen Stark again:
"Not all historical episodes of private sector money and credit balances going off track have been followed by threats to financial stability. But, every major economic crisis in the 20th century was preceded by the emergence of monetary imbalances. <..>

The Great Depression is another case in point. It is certainly true that the banking crisis brought about by the failure of central banks to understand their role as a lender of last resort under the gold standard was detrimental. But equally detrimental was the sheer unavailability of monetary data that could have signalled the need, given the collapse in money and credit, for monetary policy to be accommodative much earlier on, to a higher degree and for much longer."

JKH: Maybe, or maybe not. It's ultimately an empirical question. But the answer to that empirical question is now a lot less obvious than I thought it was. I thought the answer was that Ms didn't fall but PY did. But if Ms fell too...

Kevin: there's hysterisis and hysterisis. To say that it takes time for the short-run natural rate to return to the long-run natural rate isn't very heterodox. But even that is not in the canonical New Keynesian models. To say it never returns, so there are multiple natural rates, or a continuum of natural rates, is more heterodox. (But still not the same as a stable exploitable long-run trade-off between inflation and unemployment.)

Pre-Friedman, the mainstream view was that fiscal policy should control AD, and target unemployment. Monetary policy should target the composition of output between consumption/investment, and domestic absorption/net exports, by targeting interest rates and/or exchange rates. And inflation was to be dealt with by.....price/wage controls, industrial policy, weakening unions, whatever. I'm old enough to remember how it used to be done in the 60's and early 70's (more UK). There was a total re-assignment of which policy instrument was assigned to which target. And it was Milton Friedman's assignment that became the new orthodoxy. Inflation was assigned to monetary policy, the composition of demand to fiscal policy, and unemployment to.....the rest of the grab bag of policy instruments, whatever, or whatever was left. Yes, there's been a bit of a switch back towards fiscal policy to control AD this last year or two, but most people think of this as a temporary measure.

"Milton Friedman said that if the money supply kept growing at k% per year, where k was some small positive constant like 4%, nothing much could go very wrong with the macroeconomy, so that's what central banks should do. That never became part of conventional wisdom. Some people believed it in the 1970's. Almost nobody believes it now. I stopped believing it 30 years ago. So did the Bank of Canada, when it stopped targeting money growth. "We didn't abandon M1; M1 abandoned us" as Governor Bouey is reputed to have said. Money demand just isn't stable enough, and fluctuations in money demand can be just as damaging as fluctuations in money supply. And so central banks need to offset fluctuations in money demand while maintaining some sort of nominal anchor like an inflation target."

Question I don't know the answer to: How much of the drop of support for the k% rule (1960) in the 1970s is due to the Lucas Critique (1976)? Now the Lucas Critique is more about large scale macro problems, but it does seem to throw cold water on the general idea that we can make rules based on past observed macro relationships.

Don: I don't think there is any (useful) relationship in Macro that can be shown to be stable on purely theoretical grounds. Ultimately, it's always an empirical question. (Damn! I'm channelling David Laidler again! It was your Laidlerian "money as buffer stock" comment that did it).

TMDB: Welcome! I'm very glad you showed up here, since you understand this better than I do. I have little to add to your comments, but I can't resist this quote from Jurgen Stark:

"Money has been ignored. In the canonical New Keynesian macro-model on which the intellectual foundations of inflation targeting rests, money has been considered a redundant element in the monetary transmission mechanism. At best, money is seen as a useless appendix to the model, serving only to confuse and distract any policy-maker misguided enough to consider it."

Yep. My metaphor is that money is an epiphenomenon in NK (strictly, Neo-Wicksellian) models. Caused by, but not causing. Laidler says it's the ghost in Hamlet.

Mike: "How much of the drop of support for the k% rule (1960) in the 1970s is due to the Lucas Critique (1976)?"

The version of the Lucas Critique applied to money targeting was....damn! Mental blank. British guy. But it was more empirical than theoretical.

I'm with Luis. How does the fact that the money supply is harder to measure than we thought, imply that money demand is stable?

OK, since nobody else has, I'll bite.

You have been led astray by Gorton's "bank-run" analogy. It is not true that collateral-money is a "full-blown medium of exchange." You cannot buy anything with it, without first selling it for money, or following up later with money.

1. The only thing you can "buy" with collateral is another financial asset.
2. It is true that this "purchase" can be financed indefinitely with collateral, but so long as this is done, the purchase is not final. It can be reversed at the whim of the seller. That is what makes the "bank run" possible.
3. As a practical matter, practically all collateral-money is used to buy itself. A treasury bond is used to finance a position in the self-same treasury bond. The same was true of ABS during the credit boom; an SIV is a mechanisms for financing a super-senior tranche with itself.

So it is clear that collateral money does not have the features of a general medium of exchange. How, then, can it validate Friedman's opinions about money?

Nick, you might find this interesting.


Dark matter money is easily annihilated. Seems to me that it's mostly created as leverage and when it goes bad it turns future promised/expected/imagined returns into present demands for cash.

What is "dark matter money" ?

Re-hypothecated collateral creates a fragile chain of loans that can, potentially, break at any link, just as it did in 2007. The amount of cash remains roughly constant within the chain regardless of the number of links -- there is no mysterious "dark matter".

JW: We saw what looked like an excess demand for money, which meant that either money supply fell or demand increased. Seeing no evidence that supply fell we assumed that demand must have increased. But if supply really fell, maybe demand stayed the same.

Phil: Gary Gorton says it's a medium of exchange. You're saying he's wrong, right? One of the reasons I wrote this post was to see if we could get an answer to that question. What about when there's re-hypothecating? So the same Tbill, (or colateralised security) in effect gets used multiple times. Isn't that like fractional reserve banking?

(BTW, Milton Friedman wasn't such a medium of exchange nut himself, but I am; to me, money matters because it's a medium of exchange.)

Lee: Thanks. Good luck with the blog. It can take a lot of investment, especially at first,

Patrick: Regular bank deposits were also relatively easily annihilated, before deposit insurance and lender of last resort.

"So the same Tbill, (or colateralised security) in effect gets used multiple times. Isn't that like fractional reserve banking?"

I do not see how it is equivalent to creating credit by commercial bank.

If we have 9 people holding $1 cash each and one person holding a $1 T-bond, the bond will travel along the possibly closed chain possibly infinite number of times (not in the US where r.h. < 140%, but perhaps in the UK where r.h. is "unlimited"), but there will still be one the $1 T-bond and $9 cash in the chain.

The whole problem may have more to do with base money demand, rather than just money demand. All else being equal, an increase in currency demand is also an increase in base money demand and creates an excess demand for money. (See my post here: http://philosophyandeconomicsblog.blogspot.com/2010/10/monetary-e.html). Banks also increased their reserve demand considerably, holding reserve ratios well above normal, which can also create an excess demand for money. On top of all that, money demand probably did increase.

Isn't rehypothecation like fractional reserve banking? In some respects, yes. But the analogy is only partial; it might be better say that it is like the listed equity market. B wants to short a stock, so he borrows it from A and sells to C. C can turn around and sell to D, and so on. This creates a chain of matched assets and liabilities in virtual short and long positions in the stock, supported by only one actual share holding. There has been a corresponding expansion of credit, but only in the market for the underlying stock.

The traditional repo market, in which practically all collateral was treasuries or agencies, worked like the equity market. The modern market, with privately created underlying collateral, seems to occupy an intermediate position. To the extent that the raw material for the collateral is supplied by extending new mortgages (or other loans), the process creates credit in the real economy. To the extent that these mortgages are home equity loans, it creates generic retail credit.

However, it is still true that the only debt that collateral money can be used to settle is a debt in the collateral security itself. I don't think that can be considered a true medium of exchange. It is more like privately created scrip than private money. A demand deposit, by contrast, can be used to buy generic real goods. Perhaps some confusion arises because private collateral money has been created by the purchase of real goods. But that is not the same thing.

"Isn't rehypothecation like fractional reserve banking? In some respects, yes. But the analogy is only partial"

On the second thought, I'd agree with the analogy as bank credit extension can be seen as "a chain of matched assets and liabilities" as well.

Very much similar to commercial banking credit extension but without access to the interbank market "free" cash.

Fair to say that given the chance to start over from scratch, this isn't likely the monetary system we'd create. What a hack.

Phil, vjk: In regular banking, with fractional reserves, if BMO customers pay cheques to TD customers, and TD customers pay cheques to BMO customers, the offsetting liabilities are cancelled in the central clearing. Only the difference is paid by BMO to TD, or vice versa. That, to my mind, is a crucial part of what we mean when we say that demand deposits at TD and BMO are media of exchange. Is there anything analogous to those cancelled offsetting liabilities here?

Phil Koop is absolutely right in both comments.

Huge confusion here about how the repo market works versus how the medium of exchange works.

The repo market is all about taking advantage of credit spreads and other risk spreads in financial intermediation. It's equivalent to a mini hedge fund transaction - establishing a short position, usually in a low risk security, using the credit of the security to secure low cost funding.

Not about buying refridgerators, for example.

i.e. "A demand deposit, by contrast, can be used to buy generic real goods."

I want to buy a fridge for $100, and sell it back in 2 years for $70. Instead, I deliver a $100 Tbill to the fridge store, and the fridge store delivers a fridge to me. 2 years later I return the fridge, and pay $30 cash and get my Tbill back. Less cash changes hands. Does that example work?


Sure - you’ve repo’d your t bill as a way of paying for the lease cost of your fridge.

There’re just not a whole lot of fridge buyers doing that, unless they've just been fired from a hedge fund and are suffering from post traumatic stress.

"I want to buy a fridge"

You want to get cash first because the store does not deal in Tbills.

So, hypothetically, you repo your $100 Tbill at minus say 2% haircut minus 1%/360 interest to be paid. You buy the fridge with the borrowed money plus some to cover the haircut. The next day, or thereabout, you will have to repay the loan with the interest and will get back you Tbill.

Assuming some dealer takes a fridge collateral, you order the fridge that gets delivered, you repo the fridge at some haircut and interest and pay the store with the loan and some of your money. The fridge, unfortunately, is the dealer property now, legally, so you cannot use it !

"What a hack. "

That's called "financial engineering"

See Greenspan for more.

Also, see Volcker's comments re. ATM

Nick, from the horse's mouth:


"The Federal Reserve began reporting monthly data on the level of currency in circulation, demand deposits, and time deposits in the 1940s, and it introduced the aggregates M1, M2, and M3 in 1971. ...Over time, however, new bank laws and financial innovations blurred the distinctions between commercial banks and thrift institutions, and the classification scheme for the money supply measures shifted to be based on liquidity and on a distinction between the accounts of retail and wholesale depositors.


The Full Employment and Balanced Growth Act of 1978, known as the Humphrey-Hawkins Act, required the Fed to set one-year target ranges for money supply growth twice a year and to report the targets to Congress. During the heyday of the monetary aggregates, in the early 1980s, analysts paid a great deal of attention to the Fed's weekly money supply reports, and especially to the reports on M1

Following the introduction of NOW accounts nationally in 1981, however, the relationship between M1 growth and measures of economic activity, such as Gross Domestic Product, broke down. Depositors moved funds from savings accounts—which are included in M2 but not in M1—into NOW accounts, which are part of M1. As a result, M1 growth exceeded the Fed's target range in 1982, even though the economy experienced its worst recession in decades. The Fed de-emphasized M1 as a guide for monetary policy in late 1982, and it stopped announcing growth ranges for M1 in 1987.

By the early 1990s, the relationship between M2 growth and the performance of the economy also had weakened. Interest rates were at the lowest levels in more than three decades, prompting some savers to move funds out of the savings and time deposits that are part of M2 into *stock* and *bond* mutual funds, which are not included in any of the money supply measures.
Thus, in July 1993, […] Fed Chairman Alan Greenspan remarked […] "The historical relationships between money and income, and between money and the price level have largely broken down, depriving the aggregates of much of their usefulness as guides to policy."
[emphasis added]

iThe repo market is all about taking advantage of credit spreads...

I'd say about leveraging primarily since security purchase is "self-financing":

Buy a security

Repo the security

Pay for the security with the repo'ed cash (plus the haircut).

If delta T were required to be zero (immediate settlement) instead of the usual T+3, the trick would not work.

I've seen it argued that barrels-of-oil can be thought of as money, or as underlying money. The return on barrels-of-oil invested in new oil projects or in various renewable energy projects is lower than it was in historical oil projects. Thus, there is a contraction in money supply in terms of barrels-of-oil.

Nick, I am asserting that your fridge example doesn't work, for two reasons.

The first is the one JKH alluded to: you can't actually buy a fridge this way. If it were possible for you to do this, and for your fridge retailer to pay his supplier with your t-bill, and for the supplier to pay the manufacturer with the t-bill etc, then of course the t-bill would be money. But all you can buy with a t-bill is another t-bill.

Well then; in the repo securities market, isn't collateral functioning as money? Not entirely, because of the second point: the transaction is not settled. The fridge store can put your t-bill back to you at any time within the 2 year period without warning, whereupon you have to come up with $100 cash. (Term repo exists, but for periods of days rather than months or years.)

A final remark: there is also an important difference between repo and historical parallels such as chains of discounted commercial bills, because in the latter case, each step in the chain finances real economic activity.

"Term repo exists, but for periods of days rather than months"

Actually, three month repo are not unusual.

Technically, since repo is nothing but a lend-borrow contract with specific terms, nothing prevents the parties, except common sense, to create a hundred year term repo.

"Is there anything analogous to those cancelled offsetting liabilities here?"

Nick, I don't want to ignore an explicit question, but a proper answer to this question would be rather long and detailed. A shorter answer is that there can be some netting in how the exposure to be collateralized is measured, and in the transfer of collateral, but this does not shrink the balance sheet. A "general collateral" repo allows a number of broadly similar securities to be posted as collateral, but once posted, you have to return the exact security. This does not promote balance sheet consolidation.

I would like to add that I am not persuaded that netting is a particularly monetary characteristic. Nothing nets better than a future, but futures are not much like money at all.


WRT deposit insurance and lender-of-last-resort, Canada's banking history and practice is quite dissimilar to the American experience.

The Bank of Canada wasn't created until 1937; before then the Bank of Montreal was the fiscal agent for the Government of Canada and led the Canadian Banker's Association in overseeing financial stability.

We didn't have deposit insurance at all until 1967. We went the entire Great Depression without it. We didn't have a bank run either.

Before 1967 the last bank failure was the Home Bank in 1923. The Government bailed out the depositors on a special basis. We did the same thing when the Northland Bank and the Commercial Credit Bank failed in 1985, the next failures of note, on top of what CDIC covered.

The usual practice in Canada to this day is for a merger to be arranged among the banking community to deal with weak banks. This is how the Bank of Hamilton was rolled into the Canadian Bank of Commerce (it had similar issues to the Home Bank) in 1923 and the Bank of British Columbia was turned over to HSBC in 1985.

Nick, regarding collateral netting, I should have mentioned that in the US treasury market there is a central clearer, the Fixed Income Clearing Corporation (FICC), which allows for multilateral netting. But I know of nothing analogous for privately created collateral such as AAA ABS, which is the more interesting case. And it is still the case that treasuries don't net nearly as well as cash (because different issues are not fungible for netting purposes.)

"If delta T were required to be zero (immediate settlement) instead of the usual T+3, the trick would not work."

I do not understand why this should be true, since the settlement lag does not affect the relative timing of the cashflows. I mean, you don't have to deliver the security for 3 days, but you don't get the money until then either.

Sorry. I shouldn't have used "fridge" as an example. That was supposed to be a metaphor. Suppose it's not a fridge, but some financial asset, like an IBM share. Instead of paying cash for the IBM share, I deliver a Tbill as security. Then when I want to, I sell my IBM share back to the dealer, and only pay cash on any negative balance, if the share has lost value in the meantime.

Let me also ask you guys this: when Gary Gorton says that Tbills are functioning as money, as a medium of exchange, is he wrong? (That is not intended rhetorically.)

The trouble is that the metaphor doesn't work very well. It is easier to work with real cases. Your IBM/T-bill transaction does not fit the repo model, which is a matched purchase/repurchase agreement of an asset for cash rather than an exchange of assets.

In a typical repo you do not actually own the collateral a the start of affairs and the execution of the transaction does require a lot of cash temporarily. It would work like this: B wants to buy a T-bill from A, so he borrows the money from C by promising to post the bond as collateral. B needs only enough cash to cover the haircut on the loan, but the full purchase price of cash flows from C to A.

What is money-like about this? Well, C has temporarily lost the use of his cash, but the bond he is holding is almost as good. He can post the bond as collateral and use the borrowed money in any way he likes - to buy some IBM, for instance, or even just a lot of good champagne (bearing in mind that he will have to pay the money back!)

Now, can he repo out his IBM shares? Yes. But shares are not nearly as good collateral as T-bills, and they command much higher haircuts. So if he uses his borrowed money to buy anything except another T-bill, the chain of money "creation" is going to peter out pretty quickly. But a chain of T-bill transactions is not very interesting. So when Gorton says that T-bills are functioning as money, I would say "yes and no."

Phil (and all): thanks. I'm still trying to get my head around all this.

Determinant: Gorton kind of covers that in his book showing how NY banks used to band together during crises, similar to your canada example.

"I do not understand why this should be true, since the settlement lag does not affect the relative timing of the cashflows"

If there were no settlement leeway, no matter how short, leverage would be impossible. What make it possible is re-arranging transactions:

buy a security -> borrow cash using the just "bought" security as a collateral-> pay cash for the security


buy a security -> pay cash for the security -> borrow cash using the security

at the end of the settlement day thanks to netting.

Theoretically, you can transact with zero capital, practically there is Regulation T limiting the leverage (except hedge funds I believe).


Don't forget credit cards. Since 1950, economists have claimed they aren't money, since they are ultimately paid off with a check, note, or coin. In 1840 they said that checking accounts weren't money, since they were ultimately paid off with notes or coins. In 1710, they said that paper notes weren't money, since they are ultimately paid off with coins.

I predict that by the year 2050, economists will recognize credit cards as part of the money supply; but by then, some new kind of money will come along, and economists will deny that it is money, since it has to be paid by credit card, check, note, or coin.

Mike: That's how my mind was sort of working. That's what I fear. That's why I brought up the netting out of settlement balances, to try to argue that there was a distinction between demand deposits and credit cards. But I'm not sure if the netting out really works, since it depends on what should be trivial features, like whether there is immediate clearing or they wait till the end of the trading day.

Does it really come down to this: any asset you can *quickly* and *easily* borrow *a large percentage of the value* of, is almost as good as cash. Like my cows, for example, if I get them valued I can go to any bank and borrow against them.

Phil was talking about fungibility above, and the lack of fungibility of real assets makes them unsuitable as money. But if you can get them valued reasonably well, you can borrow against them quickly and easily with just a small haircut, so it's very close to monetising those assets.

I don't agree with all of Gorton's arguments about collateral, but
I do agree that at least some overnight repurchase agreements serve
as media of exchange.

While I think overnight loans have served as media of exchange for
decades, starting with the eurodollar market in the sixties, the
development of sweep accounts have taken away all doubt.

When a bank sweeps funds from checking accounts to money market saving accounts
at the bank at the close of business each day, and then reports to the Fed
the balance in the money market saving account and not the checking account, and
all the while the depositors have complete access to the fund in the checking account,
then the balances in the money market savings account serve as media of exchange.

A web search of sweep accounts shows that money market savings accounts at the
same bank is just one option. Repurchase agreements are another. While
balances in money market savings accounts are included in the M2 and MZM measures
of the money supply, overnight loans by "depositors" secured by securities don't
count in any measure. Still, the depositors have complete access to the funds
at all times.

That this scheme to avoid reserve requirements and restrictions on the payment
of interest on demand deposits has become computerized and perfected just means
that the long standing practice of using overnight lending as part of cash management
didn't involve the creation of money.

Concern that you can't spend repurchase agreements are beside the point. Neither
are the deposits that are checkable directly spent. It is the checks that actually
change hands. So?

Of course, we can focus on the quantity of base money and demand to hold it, and consider
checkable deposits as something that impacts the demand for base money. Or, we can included
checkable deposits as money too, and see money market mutual funds as something that impacts
the demand to hold currency and checkable deposits.

If someone has a sweep account invested in overnight repurchase agreements and they have more
than they want to hold, they just write checks to spend the excess balances. Sure, as the checks
clear, the bank has to reduce its overnight lending. But that is true with conventional deposits
as well. And those receiving the checks deposit them, and their banks get excess reserves to lend.

Similarly, if someone has a sweep account invested in overnight repurchase agreements and they have less than they want to hold, the can simply continue making deposits and write fewer checks. Sure, their
bank accumulates reserves and makes more overnight loans. But that is always the case. Other banks have
less reserves and must contract credit somehow. While the amount of credit extended by the banking system
is unchanged, the person demanding more money spent less.

Think about a banking system where everyone writes checks against overnight repurchase agreement accounts.
It isn't that hard. As long as the banks clear the checks and then roll over the remaining funds, you can
keep the illusion that it is overnight, and everything works more or less the same with a conventional banking system. Rather than making loans, of course, the system is funding whatever securities are used
for the repurchase agreements.

Of course, today we have all sorts of deposits. But it seems to me that for at least some people, overnight
repurchase agreements serve as media of exchange.

And, if people don't trust those borrowing overnight, or rather, the securities that had been lending to them against (the collateral) and so shift of conventional (FDIC insured) deposits, then this contracts the quantity of money if the quantity of conventional deposits doesn't expand.

It really is like a shift from checkable deposits to currency.

Overnight commercial paper? I didn't even know it existed. But clearly, the shadow banks were creating money, even if payments had to pass throug

“Does it really come down to this: any asset you can *quickly* and *easily* borrow *a large percentage of the value* of, is almost as good as cash.”

That’s fine, Nick. Then the question becomes who is doing that kind of borrowing and how much of the money borrowed is going into the real economy versus the financial economy. Almost entirely the latter in the case of the repo market I would argue.

BTW, your conclusion here is the basis on which the MMT’ers say there’s no difference between the government borrowing with bonds versus monetizing its expenditures without bonds – on the basis that bonds can always be repoed for cash. Hence they argue for no bonds at all (some of them). I think that the repo liquidity availability is an overly simplistic argument, but it meshes with your theme here.

Nick, here's a quote I picked up from nowhere in particular - it's definitely an oversimplification , but makes a parallel point to above:

"I think one big thing going on in all this, is that we are seeing Cantillon effects to the Nth degree. This is probably an oversimplification, but think about it: The economy is in a shambles, and Bernanke gives a trillion dollars in new money to investment bankers. What’s going to happen? They’re not going to rush to the store to buy milk and eggs. No, they’re going to buy financial assets, and indeed we saw the stock market go up 40% after Obama got in, which makes absolutely no sense. We have also seen gold setting records, which makes perfect sense too. But we haven’t seen CPI go up. In retrospect, is that really surprising? The people who buy milk and eggs are still broke; they didn’t get the trillion dollars."

This is turning into another of those posts where the commenters do all the hard work, and all I've done is raised the question. Which is fine. Just wanted to let you all know that I realise that's the case.

Does it really come down to this: any asset you can *quickly* and *easily* borrow *a large percentage of the value* of, is almost as good as cash. Like my cows, for example, if I get them valued I can go to any bank and borrow against them.

Nick, I have long been under the impression that you repudiated this idea:

It may be that in certain historic environments the possession of land has been characterised by a high liquidity-premium in the minds of owners of wealth; and since land resembles money in that its elasticities of production and substitution may be very low [1], it is conceivable that there have been occasions in history in which the desire to hold land has played the same role in keeping up the rate of interest at too high a level which money has played in recent times. It is difficult to trace this influence quantitatively owing to the absence of a forward price for land in terms of itself which is strictly comparable with the rate of interest on a money debt. We have, however, something which has, at times, been closely analogous, in the shape of high rates of interest on mortgages.
[1] The attribute of "liquidity" is by no means independent of the presence of these two characteristics. For it is unlikely that an asset, of which the supply can be easily increased or the desire for which can be easily diverted by a change in relative price, will possess the attribute of "liquidity" in the minds of owners of wealth. Money itself rapidly loses the attribute of "liquidity" if its future supply is expected to undergo sharp changes.

Are you deserting Clower?

Nick, I thought Bill Woolsey made some good points. His post was a long one, so I am going to offer a condensed version of what I liked. I deposit money in a bank, and from my point of view, I still have my money. The bank will invest this money in a way that is opaque to me. It may loan the money, creating another deposit in another bank - i.e., money. In the old days, if the bank had no lending opportunities, it would park the money in fed funds. These funds would work their way to the central money market banks, where it would fund leveraged positions, creating money. Nowadays, it might be parked in a money market mutual fund. Such funds invest in commercial paper; ultimately, this results in a loan, so they too create money. Or the bank might lend the money by executing a reverse in the repo market. This will fund a leveraged position not too different from what the fed funds would have done; it could be viewed as a substitution.

The repo market is an enormous money market, and has been for decades. Nobody can dispute that it has the effect of expanding balance sheets and therefore credit. But is repo a "medium of exchange"? That comes down to definitions, and it seems that I do not think that phrase means what you think it means. In my lexicon, something that is not a medium and cannot be exchanged is not a medium of exchange. In particular, in my opinion there was never a time when AAA ABS tranches were a close substitute for cash. Even T-bills are not a perfect substitute. So I do not view AAA ABS as "privately created money."

I don't understand how shadow banking is creating money.

The only way it could create money is if it lends some fraction of its deposits while also making the full amount available to the depositor on demand, right?

How does this work with repurchase agreements?

Kevin: yes, I was going a bit soft in that comment you quoted. A bit "wobbly" - I think that's the right term in this context! If it's like money for the individual, is it still like money in aggregate? Is it part of supply, or something that reduces demand? Sometimes I like to draw back from my opinions (go soft), and see if I end up re-establishing them.

Phil: I thought Bill's comment was a very good one too. I am trying to reach a synthesis between you and Bill, and it's great to see you do your own synthesis.

Doesn't defining the medium of exchange mean understanding Gorton's point that there is a difference between information insensitive securities (e.g. a dollar, where there is no asymmetry of knowledge - you can count on $1 being worth $1 tomorrow) and information sensitive securities (e.g. AAPL stock, where there is an asymmetry of knowledge - someone could know something that others don't which could effect its value and potentially leave a lender holding the bag), and that securities can oscillate between the two?

For example, in 2005 AAA MBS and other AAA structured products indeed functioned like money because there had never been a historic default in these products, etc. Through repurchase agreements, one could borrow on such collateral at 2-5% haircuts (almost a substitute for money). But of course, in 2008, those who demanded collateral realized that some of these structures contained a ton of subprime and that defaults could eat up to the AAA tranche. Suddenly, the repo market for AAA MBS collapsed (completely unmoney-like) and when it eventually came back, haircuts were sometimes as onerous as 30-50% (still not money-like). To me this means, that the money supply depends on what securities the market considers money-like at any given moment and its never a fixed basket.

"any asset you can *quickly* and *easily* borrow *a large percentage of the value* of, is almost as good as cash."


You cannot take $1 repo, go to a store and buy a pair of shoes(Consume) or buy a newly built house(Invest).

You can do C or I with $1 cash.

Therefore, cash != repo.

Likewise, with any security.

Whoever holds a security postpones C or I by ceding those I/C rights to someone else.

A chain of repos can explode the balance sheet infinitely, with only two participants exchanging $1 Tbil for $1 cash in a loop, with no I or C occurring.

It is quite possible that I may be missing something ;)

"lack of fungibility of real assets makes them unsuitable as money. But if you can get them valued reasonably well, you can borrow against them quickly and easily with just a small haircut, so it's very close to monetising those assets."

What it comes down to is that anything of value can be either used directly as money, or else taken to a reputable individual (e.g., a bank, a pawn shop, a local Mafioso, etc...) and swapped for that individual's IOU's, which can then be used as money. So a quantity theorist trying to pin down the quantity of money is stuck in Neverland.

Now compare that to what the backing theory says: The value of any given money, however defined, does not depend on "how much money is chasing how many goods". The value of any given piece of money depends on the value of the assets backing it. Quantity theorists might argue forever over whether a gift card to the local grocery store is money or not, but there's nothing Neverlandish about the fact that the gift card is backed by the assets of the grocer.

Mike: sure. But the "backing theory" would also explain the price of assets in a barter economy. It misses the point that money is used as a medium of exchange, to overcome real trading frictions, and when everyone wants to hang onto money, and nobody wants to part with it, things can go badly wrong with the volume of trade, and whether all mutually advantageous trades can be consummated. And the demand for money, as a stock, and hence the equilibrium price of money, is not unaffected by the fact that people actually find the stuff useful, precisely because it helps us overcome those trading frictions.

vjk: You are not missing anything, at least, not as far as I know. In fact, what you are saying sounds very much like what I would have said, before I tried to get my head around the Gary Gorton stuff.

But, I know you can't go to the local store and buy shoes, or buy a new house, with repo. But can you buy financial assets (or anything) with repo?

You see, I wondered if *I* was missing something!

MrRearden: It's a long-standing tenet of monetary theory that the good that serves as money has to be something where information about its value is symmetric. Used cars (Akerlov's market for lemons) won't work very well as money because the seller knows more about the value than the buyer. When Gary Gorton talks about "information insensitive" assets he's harking back to an old literature in the microfoundations of money (knowingly or not). Alchian had an old paper, very much in the Mengerian tradition, where he says that the good whose quality can be cheaply measured by all people becomes the medium of exchange, and other goods are bought and sold by specialist traders.

So it's a necessary condition for a good being chosen to use as money (medium of exchange). But it doesn't mean it necessarily gets used as money. It would make it a lot easier to buy and sell, and a lot easier to borrow close to its full value (small haircut). And it would be *easier* to use it as money if we chose to. But does anybody actually choose to?

Bill: I'm working through your meaty comment. Slowly.

First thought. If people use money during the day, but the stock of money gets measured at night, and if there are incentives to switch wealth out of monetary accounts into non-monetary accounts every night, then yes, money will be mismeasured.

"It's not a demand deposit, but I will convert it into a demand deposit immediately if you ever need me to, on demand".

"But can you buy financial assets (or anything) with repo? "

The repo seller (borrower) can buy anything, most likely securities rather than shoes, since he gets cash. The repo buyer(lender) has to wait for his cash plus interest to come back (or "re-repo" the collateral) before buying anything including other securities.

Repo is just a loan, like HELOC, with some specific features regarding bankruptcy provisions, legal collateral ownership and such.

Security dealers settle in cash through their custodian banks. I am not aware of settling in securities (other than netting), but I am not a trader so may be off base here. Perhaps it's done as a kind of barter but I'd imagine it would be rather unusual.

vjk: "netting" is presumably (at it's simplest) where I owe you 10 shares, and you owe me 6, so we cancel out the 6, and say I owe you 4, and only deliver (or pay cash for) 4?

If so, then netting is like barter, because it's the like peculiar case of a double coincidence of wants, where barter works fine.

And when you say settle in "cash", you presumably mean something like my chequing account at BMO?

It's all sounding just the same as pawnbrokers.

"10 shares, and you owe me 6" of The same stuff, yes.

"And when you say settle in "cash", you presumably mean something like my chequing account at BMO?"

more or less, possibly with some overdraft privileges.


"the demand for money, as a stock, and hence the equilibrium price of money, is not unaffected by the fact that people actually find the stuff useful"

We also find air useful, but so far that hasn't driven up its price. The supply of air is horizontal at zero, so its price is zero. If we traded with 1 oz. silver coins, and if people could also trade with slips of paper that reliably promised 1 oz., then the supply of those slips would be horizontal at 1.00 oz., so that would be their price, no matter how much consumer surplus we got from them. Even if physical convertibility of those slips was suspended, their issuers could still maintain their value at 1.00 oz. just by standing ready to buy them back for various assets that were worth 1.00 oz.

I'm still contending that there is no historical example of any bank, private or governmental, whose paper money traded at a premium over its backing value. Still waiting for a contradicting example.

Mike: Zimbabwe. Negative backing value. Obvious to everyone that Mugabe would just keep printing the stuff, to buy goods for the army. Obvious to everyone that it was a Ponzi scheme, only with massively negative real rates of return. And yet the stuff kept a positive value, for so long. There's one hell of a lot of ruin in even a really crappy medium of exchange. All other countries, with their inflation targets and promised negative real rate of return on money, are just micro Zimbabwes. Negative backing.


Zimbabwe, for all its faults, was still a recognized government, still owned valuable stuff, still had the ability to collect taxes, and still received foreign aid, some of which people might have expected to be eventually used to redeem the currency. It's not at all clear that backing value was negative. In fact, it's doubtful.

We could try looking at defunct governments, where people had no hope of the government redeeming the money, but when we do, we find that zero backing leads to zero currency value. The Iraqi Swiss dinar presents one example of a valuable currency issued by a defunct government, but the dinar was eventually redeemed by the new government, meaning that its positive value reflected the expectation of positive backing.

Now look at ordinary moneys: dollars, pounds, pesos, euros, etc. None of them trade at a premium over their backing value, but somehow mainstream monetary theory says that backing isn't even needed.

You've expressed some very salient doubts about mainstream monetary theory, and there's a good reason for your doubts: The theory is wrong.

Whenever you write a check to purchase a good, in the background someone somewhere in the payment system is either selling some asset for cash and then using the cash to buy another asset. It just isn't you.

Why does this distinction matter?

Wouldn't it be easier to just assume that anything that can be quickly and relatively costlessly sold for cash serves as a store of value?

I have an integrated brokerage/banking/checking/credit card account. Bonds, stocks, money market mutual funds -- to me, it is all money. When I want to spend, I shift money from one account to another -- it takes about 30 seconds. To me, 100% of my portfolio is money. I set up bill pay from it and charge purchases on it. The fact that, in the background, someone is buying and selling assets in order to allow me to do that is immaterial to any of my decision making. It is wealth and I spend my wealth to buy things, or refrain from buying to increase it.

In the background others may need to execute a chain of sales to make it all happen within a certain settlement period, but so what? All traded financial assets are money.

Its an old story. Broader money definitions--ones that include the shadow banking sector enclose financial innovation better:


The basic issue is that M3 is too narrow. Too much credit is created outside of the banking system.

"Suppose you were a monetary economist in the US in the 1930's. And suppose you knew that currency was a medium of exchange, but you didn't know that demand deposits at fractional reserve banks were also media of exchange. You would have noticed the runs on banks, and bank closures, but you wouldn't think that these had anything to do with the supply of money. As far as you knew, the money supply, which you thought of as currency, would have seemed OK. You would completely miss the fall in the money supply."

Do you see one of the differences between price inflating with currency denominated debt and price inflating with currency?

Jon said: "The basic issue is that M3 is too narrow. Too much credit is created outside of the banking system."

Bingo and good chart!

The Tiff Maklem speech has him talking about trying to find a definition for M and the BOC taking into account credit flow.


You go to the store to buy milk. You want to pay for this by reducing your holdings of Microsoft. How do you do this exactly?

Or, does someone else decide that you have made a purchase, and the best way to fund this is to sell your microsoft stock?

The specifics are important if we are to determine whether or not all wealth is money for you. Worse, to make these determination, we have to make judgement's about the sellers too.

P.S. some of the comments in this thread confuse money and credit. Money is an asset people hold. Credit is lend money for people to spend. The quantity of money isn't the amount of lending.

When you make receive deposits, say a direct deposit, does your brokerage firm purchase particular stocks for you?

My guess is that your brokerage provides an account that is just like a checkable deposit and provides you a line of credit secured by all of the assets they hold for you. If you deposit money and do nothing, then the brokerage fund has borrowed from you and owes you the money back. You can tell them to buy securities with it. If you run out of money, then they will lend you money against the security of your asset holdings.

Anyway, the balance in your account is money. Generally, if you instruct them to put it in a money market mutual fund, it is money. But your other holdings of securities are not money.

Can the price of a dollar balance in your brokerage account change from one dollar?

But, if there is a situation where you don't have to first order a sale of stock, and then that money is credit to your account, and then you can write checks, or you have to choose to sell a stock, and then the funds are used to pay down the money you owe to the brokerage, but rather, you just write a check, and the brokerage sells assets to cover your expense, then we are moving into territory where those assets are monetized.

However, it is a bit grey if the typical buyer of the security isn't in your same situation. Even if your brokerage is selling stock for you to cover your payments without your decision, but they are selling to someone who is choosing to buy for investment purposes, then the security is like money to you, but not to the buyer.

But suppose the buyer of the security is a brokerage that is purchasing the stock because a client deposited a check, and that is what they are doing with the funds?

Suppose such a system is universal. This is the sort of system described by Fama. He said there is no money. I say that this system monetizes a whole set of heterogeneous financial assets. If you assume (or believe) financial asset prices are determined by a mathematical formula, then there is no monetary disequilibrium. If you don't assume that, then I believe these assets used by the payment system will be subject to the problems of monetary disequilibrium. I don't think Fama could see that for some reason.

Edeast: I just read Tiff's speech. Yes, the interesting part is near the end. In normal times, data on M doesn't seem to help the Bank target inflation. But in abnormal times it does. (Similar to what David Laidler has been saying recently). Here's the link to Tiff's speech (yours doesn't work for me): http://www.bankofcanada.ca/en/speeches/2010/sp051010.html

Bill: Suppose I have a margin account that I can write cheques on. I have $100 in stocks, zero "cash" balance (I'm fully invested), but the margin rules allow me to borrow up to $30 on those stocks. How should we measure M in this case? $0 or $30?

Bill: Fama is too much focused on stocks. He thinks that his 3 risk factor model fits data well without any need to consider liquidity risk. Here is one random example of asset pricing models with liquidity risk premium:


Your margin account is in effect part of the brokerage account at the brokerage's custodian bank that grants a line of credit or collateralized overdraft allowance to the brokerage.

As such, your margin account workings are no different from that of an ordinary line of credit which in its turn is not different from any credit extension i.e. "loan" except being an off the balance sheet bank exposure, but that's an unimportant detail.

Re. "$0 or $30?"

Until you actually borrow and money flows from your credit allowance into someone's deposit, it's zero (with a potential of becoming $30). When it's realized as a $30 deposit, it becomes a stock and can be measured as such although the utility of such measurement may be disputed as the deposit may be destroyed pretty soon as a result of some loan repayment.

Bill: I feel like the credit - money distinction is like the chicken and the egg. If the repo market for AAA MBS collapses like in did in 2008, are we observing an outright contraction in the money supply or a credit contraction which creates excess demand for money?

Bill: The distinction between money and money substitutes is both important and irrelevant depending on the context.

The simple truth is this, unless you take the radical position that only M0 matters--That's your position is it? --then you cannot draw a distinction between time-deposits and float created by the credit card system.

There does not to be any base money for the visa card system to work. Its a money supply that expands elastically and is always in equilibrium with demand. Just like money the yield is zero.

vjk: there's a certain logic to your answer. Trouble is, suppose I get a loan of $30 against my stocks, and deposit in my chequing account. Then M=$30, even if I don't spend it. But what is the fundamental difference between the two cases?

Jon: This to me is key: when I spend currency it doesn't go out of existence. It leaves my pocket, but goes into someone else's pocket. Same with demand deposits, which just goes from my account to someone else's. A "money" that disappears, in aggregate, as soon as I spend it, isn't "money" in the same sense. It might be the same for the individual, but doesn't have the same macroeconomic consequences.


"even if I don't spend it. But what is the fundamental difference between the two cases?"

You did spend by the very act of depositing lent money because now you owe the bank the interest. When you repay the bank, with an interest, the deposit is destroyed.

The very reason to be for the bank is to profit from lending.

The mission was accomplished by your making a deposit.

John is holding a lot of cash (but in what form?). He wants a low risk way to invest -- something like a chequing account. But the FDIC only insures chequing accounts up to $100k, and John has a lot more than $100k to deposit. So he turns to shadow banking. John approaches a bank, hands over his cash and gets an IOU back. The IOU stipulates that John is the temporary owner of some asset; this is collateral that John may keep if the banks fails to repay its debt -- a substitute for FDIC cover. Technically, the IOU is not redeemable on demand, but since it is extremely short-term (usually overnight) the difference is negligible. The next day the bank is ready to repay John's IOU, but it gives him the option to "roll over" his IOU until the next day, and the day after that, and so on. Some days John withdraws a portion of the IOU and other days he deposits more. The arrangement works very similarly to an ordinary demand deposits, including the fact that banks do not need to match all outstanding debts with readily available cash, because most "depositors" just roll over their investments each day. For all intents and purposes, these quasi-chequing accounts are money, and subject to almost all the same principles as ordinary banking.

During the bust, the value of the collateral started falling, and banks balance sheets started wavering. The quasi-depositors stopped rolling over their IOUs and all tried to withdraw at once. Of course, with no Federal Quasi-deposit Insurance Company to step in, banks started a "fire sale" of assets to meet demands and the panic truly set in.

Is that right? I really don't understand the financial world very well. I mostly just like to think about pure theory, so many of the terms and concepts people are throwing around here are unfamiliar to me.

What form were these original deposits made in? That seems important. Was this a shift out of currency or ordinary chequing accounts? Currency is also base money, so that distinction seems important to me, though I haven't quite figured out why.

Surely that must be a faulty definition; it eliminates currency as a form of money, which as you know elastically gyrates up and down on a daily basis in order to meet the varying settlement demand--just as about the amount of credit outstanding varies to match demand.

This is precisely why banks cartelized to establish clearinghouses. Most of the transactions net-out, so credit can replace most of the demand for medium exchange. Much as paper notes circulated and were claims on real money--gold. Gold demand was then lessened because it was only needed to settle foreign obligations.

Under the gold regimes of the past two hundred years, it was understood that notes were a contingent claim on money--that they were money substitutes and literally credit. You cannot seriously dispute that.

Just because notes became irredeemable then does not mean that the entire class of credit became not money a substitute--and keep in mind that through most of history, notes were not universally accepted at your corner store just as in many countries where multiple currencies are in routine use, not all merchants will accept settlement in all currencies.

In the financial markets, government debt was and still is used for settlement purposes. Commercial paper was and still is accepted for settlement.

Even if you cannot use those things to buy bread, they contribute as substitutes for true money for certain things and as such they lessen the demand for true money. When those those things cease to be accepted as money, you get an increase in the demand for money.

Ergo, it does matter what those things are and how they add up. What does not matter is an index which carries over components no longer accepted as money.

"In the financial markets, government debt was and still is used for settlement purposes. Commercial paper was and still is accepted for settlement."

neither is -- the custodian's cash account is used for settlements there, just as anyone's checking account is used for the same purpose. Rocket science it ain't.


"You go to the store to buy milk. You want to pay for this by reducing your holdings of Microsoft. How do you do this exactly?"

Not necessarily MS for Milk. If I have a checking account, I write a check to the grocery store. The grocery store deposits the check with their bank. My bank, as part of its cash management operations, may sell some commercial paper and send the funds to the grocer's bank. It does sell *some* asset: the liabilities decreased, so the assets decrease as well. The grocer's bank must buy some assets. It increases the funds available to the grocer, and say uses the cash to buy commercial paper.

At the end of the day, all that happened was a swap of who holds the commercial paper -- my bank or the grocer's bank. You can pretend the cash did not exist. Of course this does require assuming the short term funding markets are working properly. Let's say the CB makes sure that this happens, that markets are liquid so all these assets are close substitutes and that there are no settlement failures.

Now, does it make a difference whether I sell the commercial paper myself, or someone sells it for me? Does it make a difference if I sell a share in a money market fund? What about a stock mutual fund?

At least in a simple, first order approximation, you would assume that all these assets are equivalent. Later on, you can talk about duration matching so that I would be more likely to sell my MS holdings for a car and my commercial paper holdings for milk. And you can also talk about heterogeneity of pricing, so I would be more likely to sell my MS holdings when I thought they were overvalued as opposed to my bond holdings, and vice versa.

But you should be able to make the "excess demand for money" argument work even if there is only one financial asset -- say risk free bonds and cash, and only a single good, and you should be able to make it work even with homogeneous expectations.

Or, is the argument that there can be an excess demand for money only when there is short term funding crisis, in which case the recession only lasted a couple of months, and things were back to full employment when LIBOR returned to earth?

Actually, the whole excess demand for money is itself a category error. Demand and supply apply only to flows, not stocks. You cannot take an indifference curve between a stock and flow, or talk about the MRS between a stock and flow. The "medium of exchange" is a stock. There is no desire or demand for a stock. The size of the monetary base does not enter into a walrassian equation, or into any utility function.

I think if those adherents to the excess demand for money phenomenon were to write down what they really mean, then they would realize that it is income, or money received per period, that is demanded, and not the total stock of money. Then your whole argument relies on the money multiplier mechanism in order to show that an increase in the monetary base will by necessity generate an increase in incomes. But we can refute this empirically, and yet the excess demand for money model refuses to die.


The way to combine stocks and flows and to look at flows and desired changes in stocks.

This can all be done with only one asset--money, and production and consumption of a single consumer good over time.

With multiple assets, then of course the extra consumer goods that can be obtained because of the yield on the assets also impacts the desired change in the the stock of money held. No one denies that.

The desired change in the stock of money isn't the same thing as the desire for income. You must be kidding.

The money multiplier process is not necessary for there to an excess demand for money. What if there is no money multiplier at all because currency is the sole type of money? What if there is no money multiplier because there is no monetary base?

Finally, your notion that all assets are money because a payment involves one bank increases assets and another decreasing assets...
you must be joking. That is supposed to be news?

If you sell an asset to fund the purchase of a consumer good, then this puts downward pressure on the price and upward pressure on the yield of the asset. You provide an incentive for someone else to hold the asset rather than spending on consumer goods or capital goods. This frees up resources to produce the consumer good you buy.

If you write a check to buy a consumer good and your bank ends up selling an asset and the seller's bank buys an asset, no such signal is generated. There is no reason to believe that the seller who receives the added balance in his checking account wants to hold that money rather than spend it. No signal or incentive is created to keep the demand for current output in line with productive capacity.

Similarly, if you earn money income and refrain from making expenditures, you accumulate money balances. Your bank will be accumulating assets to match that. Great. However, the banks of those whose payments formed your income will be selling off assets. Assets held by the banking system are unchanged.

Suppose that instead you received money income and spent it on some asset. This puts upward pressure on the price of the asset and downward pressure on the yield. This creates a signal and incentive for others to hold fewer such assets and instead purchase consumer or capital goods. While your bank would sell assets, the bank of the person selling the you the assets would buy assets. The assets of the banking system would be unchanged.

An excess demand for money is possible if there is no base money. It is the relationship between the demand to hold a certain type of bank liability and the quantity of that liability created by banks. If you want to compare this to a flow, (and how it impacts the flow of money expenditures is the point,) then it is the change in the desired stock of those bank liabilities over time and the change in the desired stock of those bank liabilities that banks choose to offer. The differences will be the same whether you look at the stocks or the changes in the stocks.

And so, nobody gets all worked up about this stock flow issue.

Patinkin covered this in the fifties, I think.

Anyway, I think it is conceivable for there to be a monetary system where many, even all, ordinary securities are effectively monetized. Nobody but "banks" ever buys or sells securities, they just write checks or receive them. And those managing the payments system always make offsetting purchases or sales. (Fama) Similarly, we can imagine economies with no financial asset markets other than bank deposits and lines of credit. People fund expenditures on output by writing checks and they can only hold balances in checking accounts. (Black.)

Black and Fama imagine that this makes monetary disequilibrium impossible. Wrong--it just spreads it out over all financial markets. Instead of being a special problem with those assets generally accepted in exchange, it becomes a problem with all assets.

If you assume asset prices and yields are always at equilibrium (like with a Walrasian auctioneer,) then there would be no disequilibrium. But what market process would cause those yields or prices to change? No process.

"My bank, as part of its cash management operations, may sell some commercial paper and send the funds to the grocer's bank"

A sane bank's treasury would maintain a cash buffer as part of its daily cash management operations. The cash buffer would be funded primarily by the bank's liability side of its balance sheet: deposits, interbank borrowing, etc. Daily cash flows due to deposit/withdrawal activity is relatively stable and pretty predictable. So, in the above scenario, the two banks will most likely settle directly in cash without selling any assets.

An insane bank, such as Northern Rock plc, may think that all "money" (CP, CDOs, ABS, etc) is cash and engage in commercial paper trickery to manage daily cash flows, with predictable consequences.

"Let's say the CB makes sure that this happens, that markets are liquid so all these assets are close substitutes and that there are no settlement failures. "

After 2007 lessons, that's quite an assumption ! Fortunately, empirically, some banks learned (at least temporarily) that the interbank market is not quite frictionless, not everything is "money" and that their balance sheets are not as goods as they looked at the quarter end, and chose to hoard about 1Tril extra cash on their balance sheets' asset side rather than buy commercial paper or some other "money" equivalent.

"The version of the Lucas Critique applied to money targeting was....damn! Mental blank. British guy. But it was more empirical than theoretical."

Nick, sorry for coming into his thread late, but the name you're looking for is Charles Goodhart (of LSE):

"Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes" (1975)


Balance sheet problems of banks are real, but this is not due to a shortage of media of exchange, neither does this constitute a failure in the payment system. The latter applies only when there is no confidence that trades can be successfully executed at any price -- because you do not believe that the trade will settle or that your counterparty will be able to honor its obligation. We had some of that for, say, a month, but this cannot explain the current output gap. Moreover, the temporary liquidity crisis was a symptom of the persistent balance sheet crisis, rather than as the sole explanation for why there is persistent mass unemployment.

The fact that your assets are re-priced downward does not constitute a settlement failure. It means that you have a solvency, not a liquidity crisis. There is no liquidity crisis for banks when they can borrow at zero rates. The balance sheet problems facing banks are because they overpaid for assets, not because the assets are illiquid.

Neither are the banks "hoarding" cash. The CB sets the marginal cost of reserves, which is currently zero, or near zero. It is illogical to hoard something that you can acquire costlessly and in unlimited amounts. Actually, "the medium of exchange" is the only thing that is *never* hoarded when the CB operates in a rate targeting regime -- the supply of cash is perfectly horizontal at overnight terms. If everyone were to withdraw cash and not redeposit it, but were to continue to be current on their loans, then banks would not be starved for cash as the CB would create more as needed and the balance sheets of banks would remain sound.

But this does not happen in an era of deposit insurance, so that technical factors determine the amount of cash that banks need as vault cash or required reserves to meet their cash-management needs, and any additional cash is unnecessary. Banks then proceed to try to lend this cash out in the overnight market, driving down the overnight rate. In order to keep this rate at the target, the CB drains the unwanted cash by selling treasuries into the market -- this is the only mechanism by which the excess cash can be removed from the banking system.

If for some reason, the CB does not sell enough treasuries or otherwise remove the excess cash, then overnight rates will be zero and the excess cash will sit on bank balance sheets as unwanted reserves. That is a far cry from hoarding. It is the opposite of hoarding. Something has a marginal price of zero when no one wants it -- not when it is hoarded. What is in short supply are profits -- income net of expenditure -- not the medium of exchange.

The only reason why we have excess reserves now is that the CB is not draining. Initially during the crisis, the CB sterilized its lending operations and drew down its stock of treasuries. Since then, it has stopped sterilizing and rebuilt its stock of treasuries, but has continued asset purchases. As a byproduct, you see excess reserves in the banking system. Banks would much rather have those assets be interest bearing government bonds than cash sitting on deposit with the CB. But they are stuck, and this situation unwanted reserves is supposed to stimulate the economy.

hisham: Goodhart's Law. That's the one!


"The fact that your assets are re-priced downward does not constitute a settlement failure. It means that you have a solvency, not a liquidity crisis. "

A settlement failure occurs when there is no cash on your account to settle your obligations with immediacy and no one is willing to lend or buy your CDOs. Solvency and liquidity can, to a degree, be separable during "normal" times which exist in theory but rarely in practice. More often than not one leads to the other for obvious reasons.

Neither are the banks "hoarding" cash. ...It is illogical to hoard something that you can acquire costlessly and in unlimited amounts.

That the cash is "costless" and "unlimited" is a dangerous misconception to promote. Some economists, especially of MMT persuasion, assume frictionless interbank market that distributes cash across the "system" efficiently and a Central bank that creates required cash in infinite amounts to replenish the "system". None of those assumptions are even remotely true in practice. Basel III finally admits that cash management is as much, if not more, important as capital requirements and makes specific numerous recommendations and requirements to provide reliable cash management, being the first Basel to do so. Should cash be a free resource MMT claims it to be, why would the BIS bother ?

The *fact* that banks hoard cash was discussed a lot during the August 2008 Jackson Hole meeting. There, McCulley (who was allegedly the first to use the "shadow banking" name) admitted : "I confess I was a large liquidity hoarder even though I was a net lender to the System last fall".

"The only reason why we have excess reserves now is that the CB is not draining."

Not at all. The banks could have easily bought government securities on the open market without any CB participation thus exchanging their reserve cash for bonds or any other security of their choice. Some of the reasons they *choose* no do so is weakness of their balance sheet, real or perceived counter-party risk, lack of trust in the interbank market efficiency in distributing cash.


An individual bank can purchase a treasury on the open market, but this just shifts the cash to the bank of the treasury seller. Unless the treasury seller is the central bank, in which case the cash is destroyed. In the same way, a member bank can borrow in the overnight or short term funding markets, but this just takes cash from another member. Neither of these operations increases or decreases the total stock of cash in the system. Only the CB can do that.

So if we have an economy with 10 banks that each need $50 billion for their vault cash and reserve requirement needs, then in aggregate they need $500 billion. The economy as a whole only needs 500 billion to support the given cash-flows. Assume each bank has enough. Now the central bank buys $100 billion of treasuries, the seller deposits the money into his bank, and that bank, as it already has enough cash, will try to buy some short term asset, driving down the price. The seller of the short term asset deposits the proceeds into his bank, which also has enough, and it buys some other short term asset, driving the short term rate down even lower. The process repeats until the short term rate is zero, or very close to it, at which point whichever bank(s) have excess cash will be indifferent between holding onto to the zero yield cash and purchasing a zero yield short term asset. All throughout, some bank(s) will always have this money in their reserve accounts, in excess of their required reserves. Therefore the CB has just created $100 billion in excess reserves and has simultaneously driven the short term rate to zero.

In the reverse process, if there is an overall system shortage of cash, the short term rates will spike up, some banks fail and you have a payment crisis. But the CB will not let them remain elevated -- it will supply all the cash that banks need for their vaults and required reserves by buying enough assets to drive the rate back down to the policy level. By definition, when the overnight rate is stable at the policy level, then there cannot be an overall shortage of cash. And when the overnight rate is zero, then there cannot be any excess desire for cash. That does not mean that specific banks with bad balance sheets don't go bankrupt or that asset prices don't fall. This says nothing about the long term risky cost of capital for non-financial investment. But it means that the banking system as a whole is not liquidity constrained and the payments system is working, even though investors may not like the prints.

But the main point here is that you only need enough cash (as a stock) to meet the existing transaction flows -- due to netting, a tiny amount. Cash is just a technology that allows this complex set of flows to work. Doubling the stock of cash so that there is a trillion dollar excess over what is needed is not going to cause the spending flows to double, anymore that VISA, by doubling the reliability of its payment system is going to cause people to buy materially more with their credit cards.

You should be able to formulate a model of recessions that does not rely on the payments system malfunctioning as your source of demand failures, and a simple model of the economy should just assume that the payments system is working and that the economy is cashless.


"Neither of these operations increases or decreases the total stock of cash in the system. Only the CB can do that."

Not at all -- the CB is not the only player here. The banks are free to participate in Treasury auctions if they decide to do so, and that's what they do to get rid of extra cash:

"banks bought a total of $40 billion from the Treasury in March, according to analysts at Deutsche Bank." (Banks May Not Be Lending, But They Are Buying Treasurys).

Of course, a private party buying government paper through its bank would have the same drain effect on reserves.

Likewise, Treasury spending "injects" cash in the system and taxation drains it.

One has also to keep in mind that cash was not forced on banks by the evil feds -- the excess reservee were the result of banks *willing* to increase their cash holdings for whatever reason.

Technically, no, the treasury department has an account with the CB and is just another user of the CB's currency. Under current arrangements, the treasury does not have the power to create or destroy cash, and so sales of newly created treasuries by the treasury dept. do not remove excess reserves. This is because the treasury sells bonds in order to spend, so whatever cash is drained by the treasury sale is then spent by government and ends up back on some bank's balance sheet in the form of payroll, interest payments to households (and banks), vendor payments, etc. This does not increase or decrease the quantity of cash in the system.

Now, if the government ran a budget surplus, then in principle the treasury would have a surplus of cash on account at the federal reserve, but this would not constitute excess reserves in the private sector banking system. So in that case, you are right, by running budget surpluses, the treasury would be draining the banking system of cash, and if the system was in a normal state (e.g. no large excess reserves), then the CB would need to offset the budget surpluses by purchasing government debt, otherwise it would through the banking system into a crisis. Alternately, the treasury could establish accounts with member banks and shift its cash holdings to them.

And yes, banks are absolutely helpless to get rid of the excess reserves and have, in aggregate, no control over the level of excess reserves. This is a simple arithmetic fact.


" This is because the treasury sells bonds in order to spend, so whatever cash is drained by the treasury sale is then spent by government and ends up back on some bank's balance sheet "

That's a good point. Assuming the government spends everything it gets through taxation and security paper issuance, the aggregate reserve flows will net to zero, and that's a very realistic assumption.

However, the bank sector still can change its balance sheet composition by buying paper at Treasury's auctions and thus moving its cash assets, via the eventual government spending channel, to the non-bank sector so that part of bank cash reserves would be balanced by non-bank deposits(liability).

In any case, the reserve bloat is a result of the banking sector volitional act, at least partially, -- no one was holding a gun to its head as it were. It would be interesting to know what part of the "injected" cash is a non-banking sector asset as percentage of the total bank reserves, perhaps banks do not "hoard" so much after all ...

"However, the bank sector still can change its balance sheet composition by buying paper at Treasury's auctions and thus moving its cash assets, via the eventual government spending channel, to the non-bank sector so that part of bank cash reserves would be balanced by non-bank deposits(liability)."

Not in aggregate, because the private non-bank sector has accounts with the banking sector, so that doesn't get rid of the cash, either. B of A buys a treasury on the secondary market from Joe Investor, who has a brokerage account with Citigroup, and the cash goes from B of A to Citigroup. Or B of A buys a new issue sold by treasury, and the proceeds are mailed to Jane Social Security recipient, who banks with Wells Fargo, so Wells gets the cash, etc. The cash just hops around from one bank to another; there is no transaction by which the banking sector can free itself from a system wide reserve excess, or make up for a system wide shortfall. B of A can buy beer for a company party, and the beer vendor banks with JPM, so the cash is still in the banking system.

Only the CB can alter the total stock of cash in the system. Banks, short of refusing to accept cash deposits, or refusing to accept payments in cash, are forced to hold whatever levels of cash are in the economy.

I'm not saying that the system is airtight. There is some mattress money. Mail Float. I think the biggest leakage is eurodollar lending.

But it's more or less a closed system, and certainly if banks at some point have roughly the cash that they need, and then the CB decides to purchase a trillion in assets from private investors, and those private investors deposit the proceeds of the sale in the banking system, then the banks will be stuck with about a trillion in excess reserves, and they wont be able to rid themselves of those reserves until the CB sterilizes them by selling an equivalent amount of assets.

"Not in aggregate"

I did not say in aggregate. What I meant was the cash distribution amongst the bank sector, the rest of the financial sector and the households.

Even though the cash grand total remains the same, the sectoral distribution might provide some insight as to the motivation of those groups to sell government paper in exchange for cash. The aggregate reserve number, the fact that the cash sits on the bank reserve account, is pretty meaningless by itself.

To rephrase in a less muddled way:

1. Let's ssume there are following sectors other than the government: non-financial companies C, households H, banks B, finance minus banks(brokerages, mutual funds, pension funds, insurance, etc) F.

2. What might be contribution of each group into the 1tril government securities swap for cash that resulted in 1tril excess reserves ?

1tril = C + B + H + F

One would speculate that H and C are pretty much negligible in comparison to B and F. I doubt such information is available, but had it been available, it might have been useful to predict the QE2 outcome, at least to a degree.

VJK, "Finance" is considered part of the financial sector. ABS issuers, funding companies, thrifts, savings and loans, commercial banks, etc.

Actually many of these non-bank finance firms are captives or special purpose entities created by the big banks. A better breakdown would be: Households, foreign sector, CB, financial firms, non-financial firms, and investment firms. Investment firms corresponds to pension funds, insurance firms, etc. You can roll the latter into households if you want. The point is, none of the non-financial firms run their own bank vaults or have reserve accounts. All the other sectors have deposit accounts with the financial sector, and you can assume that any FRNs that are created are either going to be in the form cash in someone's wallet, or or they will be deposited within banks in the financial sector. Moreover, within the financial sector, any cash that is received is either stored in a bank vault for cash management needs, or it is used to acquire reserves at the CB. That is what is meant by, "in aggregate". No pension fund is running a bank vault, storing cash in it, everyone has a deposit account at some bank.


A better breakdown would be: Households, foreign sector, CB, financial firms, non-financial firms, and investment firms

Right, behaviorally, households and investment firms can be merged, and probably non-financial firms firms too, into one sector.

I'd separate depositary institutions (commercial banks and such) from the shadow banking.

In any case, this line of reasoning seems pretty scholastic in the absence of numeric data as to how government paper flows are distributed amongst the players.

No pension fund is running a bank vault, storing cash in it, everyone has a deposit account at some bank.

That much is obvious ;)

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad