Alpha, beta, and gold

Central banks issue currency. (They issue reserves too, but "reserves" are just the name we give to central bank currency held by commercial banks in electronic form rather than in paper form, and it makes no difference whether money is printed on paper, on plastic, or on computer disks.) Commercial banks issue deposits. (They used to issue notes too, but again it makes no difference whether money is printed on paper, plastic, or on computer disks.)

Central bank currency is alpha; commercial bank deposits are beta. Commercial banks promise to convert their deposits into central bank currency at a fixed exchange rate; the central bank does not promise to convert its currency into commercial bank deposits at a fixed exchange rate. It is the commercial banks' responsibility to keep their exchange rates fixed. The alpha goes where he wants to go; and the betas follow the alpha. The central bank decides monetary policy.

In the olden days, under the gold standard, central banks promised to convert their currency into gold at a fixed exchange rate. Did this make gold the super-alpha?

I was reading David Glasner's good post on the 1920-21 recession. As expected, I learned some monetary history. But David has also caused me to re-think my views on the theory of the gold standard.

I think we need to distinguish two cases:

1. The "small country" assumption. Central bank stocks of gold are small relative to the total stock of gold.

2. The "large country" assumption. Central bank stocks of gold are large relative to the total stock of gold.

Implicitly, I was thinking in terms of the small country assumption. David's post uses the large country assumption. And Wikipedia tells me that even in 2011 central banks held about 17% of the total world stock of gold, which is not really that "small".

Most central banks today have replaced the gold standard with the CPI standard; they target 2% CPI inflation, rather than 0% gold price inflation. But central banks today hold extremely small stocks of those goods that make up the CPI. They own a few computers, and some furniture, but that's trivial compared to the total world stocks of computers and furniture. The "small country" assumption makes perfect sense for inflation targeting central banks today.

Under the gold standard, the "small country" assumption would still make good sense for the central bank of a small country like Canada. The Bank of Canada's stock of gold would be a very small percentage of the total world stock of gold.

But it would not make sense for a large country, and it would not make sense for a global analysis where the total stock of gold held by all central banks would be a large percentage of the total world stock of gold.

For a small country on the gold standard, we can treat gold as the super-alpha. Commercial banks follow the central bank, and the central bank follows gold.

For a large country on the gold standard, or for the world as a whole on the gold standard, we cannot treat gold as the super-alpha. Central banks can change the mix of assets on their balance sheets. They can hold more gold and less bonds; or less bonds and more gold. If central bank demand for gold is large, relative to the total stock of gold, the actions of central banks will affect the equilibrium real price of gold against all other goods.

If large central banks, or a large number of small central banks, buy gold and sell bonds (raising interest rates on bonds in the process and increasing the opportunity cost of holding hold so people sell gold to central banks and buy bonds from the central bank) that raises the equilibrium real price of gold against goods. And that raises the equilibrium price of central bank currency against goods. Which means the equilibrium price of goods falls against currency. It's a deflationary tightening of monetary policy, even though central banks keep exactly the same exchange rate of currency against gold.

Under the gold standard, gold is the super-alpha, but large central banks (or Departments of Finance) can lead it by the nose, by changing the percentage of gold in their portfolio mix. And David shows they were doing that in 1920-21, which is what caused the recession. And it would take a very long time for the gold miners to adjust the stock supply of gold in response to changes in the real price of gold caused by changes in central bank demand for gold.

Is that basically right, David?

[Update: see David's post in response, with a long quote from Dennis Robertson, the source of David's views on this.]

Comments

[George Selgin]

Nick, I know you know it, but as the context of the discussion is the 1920s, it seems worthwhile to make clear that the Bank of Canada wasn't around until 1935. A more important point, I believe, is that the "small country" assumption isn't necessary for any country, whether large or small, that lacks a central bank or other central monetary authority capable of offsetting gold inflows. In other words, in the absence of central banks the only "alpha" (reserve) money is gold itself. Bagehot recognized this point in a contrasting a "natural" or "many reserve" system with England's more crisis-prone "one reserve" system, dominated by the B of E.

The tendency of central banks to undermine the normal workings of the gold standard is one that discussions of the working of that standard too often overlook. Even worse, some authorities, like Eichengreen, imagine that the classical gold standard worked because it was carefully managed by central banks, and by the Bank of England especially. (Gallarotti's Anatomy of an International Monetary System" does an excellent job countering this view.) I have a recent paper, "Law, Legislation, and the Gold Standard," available on SSRN, arguing that the spread of central banking ultimately doomed attempts to restore any sort of gold standard, and, indeed, any durable non-gold fixed exchange rate regime.

One other point I wish to raise concerns your reference to a "CPI standard," which seems to suggest fundamental confusion of a fiat regime with one founded on convertibility. The assets held by modern central banks, excepting those committed to fixed exchange rates, have essentially no bearing on either their ability to or their likelihood of achieving announced inflation rate targets. There have been suggestions for CPI-stabilizing regimes based on convertibility, such as Fisher's compensated dollar and the Greenfield-Yeager BFH scheme. But no modern central bank is committed to such a scheme. Consequently I do not think it correct to suggest that a modern CPI targeting central bank is analogous to a "small country" central bank operating on a gold standard. On the contrary: modern central banks enjoy more rather than less discretionary leeway, if not unlimited leeway, than their "large" gold standard counterparts.

[Scott Sumner]

Yes, that's right. The ratio of gold to the base is an especially useful way of thinking about gold demand under the gold standard. Unlike the domestic money supply and the domestic interest rate, the gold ratio was exogenous, and hence could be viewed as the individual country's contribution to changes in the global stance of monetary policy. When the global gold/base ratio increase sharply, as in 1920-21, and again in October 1929 to October 1930, it represented a sharp tightening of global monetary policy. Then you can look at the gold ratios of individual countries to apportion "blame."

(Now if only my publisher would release my book on the gold standard during the Great Depression, a book complete almost a decade ago.)

[Scott Sumner]

One other quick point. By the interwar period the gold standard had evolved to the point where there was ambiguity as to what was the "medium of account." Legally it was still gold, but when devaluation occurred in 1933 (in America) it was currency that continued on as the medium of account. Arguably there was a sort of dual medium of account in the interwar period. Dollars redeemable for gold, and gold redeemable for dollars.

[Scott Sumner]

And Selgin's paper (mentioned in his comment) provides a nice discussion of how the gold standard evolved from its original structure to the more central bank-dominated form of the interwar years.

[Andy Harless]

George, Have you read the linked post on the CPI standard? Nick moves step by step from the gold standard to inflation targeting, and it would probably be helpful if you could state your criticism in terms of those steps.

I don't think I agree with your criticism, but I have a different one (at least I think it's different): I particularly object to the step of allowing base drift. That, to me, is what makes inflation targeting no longer a "standard." A central bank can deviate from its target and say that it made a mistake, or that it faced an unanticipated random event, or that it had temporary concerns that outweighed the target, or that the deviation made hitting the target more likely in the long run, or whatever; what's more, there is nothing to prevent it from having serially correlated misses that it will explain away, or apologize for, after the fact. This, to me, is what makes inflation targeting essentially discretionary in a sense that a gold standard is not. Legal convertibility is certainly an issue, too, but I think it's less important, in that it can generally be suspended, so it doesn't really avoid discretion.

[Nick Rowe]

George: yep, but the Society of Creative Anachronism said I could have the Bank of Canada around then! It would really have been some sort of mix of the Department of Finance (which still technically owns the Bank of Canada's gold and forex reserves) and the commercial banks.

I'm going to have to think more about the Bagehot point.

"One other point I wish to raise concerns your reference to a "CPI standard," which seems to suggest fundamental confusion of a fiat regime with one founded on convertibility."

That's what I used to think, until I changed my mind. A central bank under the CPI standard has *indirect* convertibility into the CPI basket of goods. Inflation targeting is just a crawling peg, with slightly fuzzy "gold points", and where they adjust the peg for past mistakes.

" The assets held by modern central banks, excepting those committed to fixed exchange rates, have essentially no bearing on either their ability to or their likelihood of achieving announced inflation rate targets."

True. The most important asset by far is the present value of seigniorage profits due to the central bank having a de facto monopoly power due either to law or to network externalities. The central bank could borrow against its future profits, if it needed to buy back currency to prevent inflation rising above target, but had lost all its other assets in a fire.

" On the contrary: modern central banks enjoy more rather than less discretionary leeway, if not unlimited leeway, than their "large" gold standard counterparts."

I disagree. Sure, an inflation targeting central bank can abandon the CPI inflation target. Just like a central bank on the gold standard can abandon the 0% inflation gold price target. But given that they stick to their targets, the CPI inflation targeting central bank has less discretionary leeway (except for the "flexibility" bit in 'flexible inflation targeting") than a gold price targeting central bank. Because the latter can be a large holder of gold.

[Nick Rowe]

Thanks Scott. "Gold/base" ratio is like a (desired) reserve/deposit ratio, for a commercial beta bank.

[Nick Rowe]

Andy: " I particularly object to the step of allowing base drift."

Yep. As you know, I think you have a point there. I think Scott does too. My recent post on this. CPI price level path targeting is much closer to the gold standard, because it doesn't allow excuses for base drift.

[Mike Sproul]

" The assets held by modern central banks, excepting those committed to fixed exchange rates, have essentially no bearing on either their ability to or their likelihood of achieving announced inflation rate targets."

1. If a central bank dumped all its assets in the ocean, would that have only a small effect on the value of its money, as opposed to reducing the value of that money to zero?

2. If the demand for a central bank's (inconvertible) money fell by 30%, then if that central bank wanted to keep the value of its money from falling (by about 30%), then the central bank would have to buy back about 30% of its currency. If the central bank had dumped its assets in the ocean, how would it buy back 30% of its currency?

3. Of all the central banks in the world, why do they all have assets?

"True. The most important asset by far is the present value of seigniorage profits due to the central bank having a de facto monopoly power due either to law or to network externalities. The central bank could borrow against its future profits, if it needed to buy back currency to prevent inflation rising above target, but had lost all its other assets in a fire."

4. What happens if it's a small country, with no monopoly power or network externalities? Do the central bank's assets then matter?

[Nick Rowe]

Mike: your question was perhaps mostly asked of George. But here are my answers:

1. No effect, provided the expected present value of the central bank's seigniorage profits was large enough to exceed any likely fall in demand for central bank currency. (Unless the central bank chose to increase its inflation target to increase seigniorage profits to replace the lost assets.)

2. The central bank borrows against its future seigniorage profits. One way to do this would be by paying sufficiently high interest on reserves to increase desired reserves enough to increase the fall in demand for currency.

3. Good question.
a) to stop accountants and bankers (and real bills theorists!) freaking out.
b) it's probably less hassle to sell government bonds to reduce the supply of currency than to issue its own bonds backed by future seigniorage profits.
c) because it really doesn't matter whether the central bank keeps its assets or gives them to the government. So a or b above would be sufficient reason, since there's no downside to the central bank owning assets.

4. If currency competition is large enough, or if the inflation target is low enough, the present value of seigniorage profits might be too low to be able to buy back currency in the event of a big fall in demand for currency. Assets would matter then.

[George Selgin]

Nick (and Andy), I fear I must continue to disagree, and that my disagreement is such that no amount of "steps" is likely to affect it. It is simply a matter of insisting that the notion of a currency's being "convertible" not be be so stretched as to allow it to encompass any currency that happens to be administered with a particular nominal target in mind!

Surely the common understanding of "convertibility" of a paper currency is that the issuing authority stands ready to redeem it for some definite quantity of something, and not merely for another unit or other units of the same paper. By this (really very loose) definition, no modern central bank notes, save those "pegged" to other currencies, are convertible. You can exchange a worn note for another, or a $5 note for five one-dollar notes, but that's all, as you all know perfectly well. Nor are U.S. dollars "indirectly" convertible in any sense that doesn't do violence to standard usage. In the Fisher scheme, for example, paper dollars command a varying quantity of gold coin depending on gold's current purchasing power. That is certainly a big "step" from fixed-rate convertibility. But it still is convertibility of a sort, and it therefore still subjects the currency issuer with a binding reserve constraint, or would do so if the issuer were subject to any genuine sanctions in the event that it defaulted (as any commercial note issuer would normally be). But no such constraint affects a mere CPI targeting central bank which, never actually being obliged to supply a good or goods in exchange for its notes, has no need for "reserves" of any particular sort. Consequently there is no sense in which such a central bank is "constrained" by the "convertibility" of its currency. None. Certainly the mere fact that a paper currency issuer adheres to a CPI target itself doesn't imply that its currency is "really" convertible, directly or otherwise, and more than the fact that my well-behaved but free-roaming dog doesn't run after cars, implies that she is "really" on a leash.

What practical difference does all this semantic nit-picking make? Well, recognizing that CPI targeting really isn't "like" having a currency that's convertible into a CPI basket is important because it helps to clarify the need for some substitute for the threat of default to make a CPI targeting regime credible, like a contract that (unlike the NZRB contract) truly subjects the central bank governor or chair to certain punishment in the event of failure to stay on target.

[Nick Rowe]

George: so, I can't get you to slide down the steps of my slippery slope argument?

Ah well. Maybe I could get you to slide up them? Start with inflation targeting, switch to price level path targeting, then slowly subtract goods from the CPI basket, until only gold remains. (I'm not sure if gold is currently in there, but never mind.) Since the price of gold is observed daily, it would be very easy for a central bank to target the gold price, even without direct convertibility. Does adding direct convertibility really make much difference, as long as the price of gold is publicly observed?

[George Selgin]

No Nick--you can, and I will read the arguments because they interest me. But the fundamental issue remains one of definitions. Here I see you saying that convertibility doesn't "really make much difference," which is rather different from your previous statement that "A central bank under the CPI standard has 'indirect' convertibility into the CPI basket of goods." I grant that the former statement might be correct--"might" because so far we have yet to see a central bank keep to a "target" unaccompanied by sanctions. But I insist that the second is not.

The practical issue, again, is this: when you speak of price level targeting being "like" a convertibility regime, what you mean is really that it would be like such a scheme assuming that the target is indeed adhered to. But understood this (correct) way, the claim begs the question, to wit: why should the central bank be expected to adhere to the target? To acknowledging that question, instead of tacitly setting it aside, is to realize the sense in which a P-targeting regime is not like a convertibility scheme, because the latter scheme involves both an implicit "target" and a mechanism that constrains the issuer to adhere to the target, whereas the former specifies a target, but no enforcement mechanism whatsoever.

Now I really must read your steps!

[Majromax]

@Nick:

> But given that they stick to their targets, the CPI inflation targeting central bank has less discretionary leeway (except for the "flexibility" bit in 'flexible inflation targeting") than a gold price targeting central bank. Because the latter can be a large holder of gold.

To make this argument even more explicit, recognize that a fiat central bank has only one policy instrument, that being the quantity of base money. A gold-converting central bank has two: the quantity of base money and the quantity of gold reserves.

> it's probably less hassle to sell government bonds to reduce the supply of currency than to issue its own bonds backed by future seigniorage profits.

Would bonds issued by seigniorage profits work as the sole tool? It would seem to cause problems with regards to credibility, as then to avoid default the CB would be committed to running (at minimum) a particular level of seigniorage profits in the future. That level could be high enough to require future inflation.

Imagine that someone finds out that the US Fed issued a zero-coupon, $100 quadrillion bond drawn on the reserve. When that comes due, the Fed would either have to print money to satisfy it or it would have to resort to a vacuum cleaner -- tax policy -- to accumulate the money from the existing stock.

It seems that assets represent a Central Bank's inbuilt ability to permanently reduce the monetary base.

@George:

Be careful about the difference between a path and level target. Much of your criticism would apply equally to a 0% "gold inflation target", where the CB tries to keep the future price of gold equal to the current price but does not account for prior divergence.

[George Selgin]

@Majromax: I actually regard my remarks as applying equally to level and rate targets, for they have to do with the questions of enforceability and credibility, which apply to either. I am not merely observing, as others have, that under rate targeting there is a tendency for the level to wander even when the central bank does indeed adhere to the target. I hasten to say, in particular, that I do not think a regime in which a central bank is supposed to "peg" the price of gold--that is, a gold price level targeting scheme--is "like" a real ("convertibility-based") gold standard, for the reasons I give in my previous comment.

[Andy Harless]

George: " ...why should the central bank be expected to adhere to the target?"

But why should the central bank (or the treasury) be expected to adhere to a promise of convertibility?

The only difference I see here is that convertibility, when it fails, fails discontinuously, whereas a target might fail gradually. But then again, a target might also fail discontinuously, if the target were observable in real time: a central bank might initially go to extremes to maintain a target as the market was losing confidence in it, just as it might do with a convertibility regime.

Real-time observability of the target is a critical issue, though. We can't really tell when an inflation target is failing, because the target is in the future. And, for example, in Yellen's press conference today, she pointed out that changes in market inflation compensation can be the result of changes in the risk premium; there's no presumptively unbiased observable indicator of the central bank's target variable.

But I agree with Nick that the issue of literal convertibility is not very important. Does it make a difference if I promise to buy something from you at a given price or if I, instead, promise (credibly) to find someone else who will buy it from you at that price? (Even the first promise might not be entirely credible, because you can't be sure I will have enough liquid wealth available to buy it from you at any given time.)

[George Selgin]

Andy, I think your comments about convertibility are too sweeping, and that they ignore some centuries worth of banking history, during which convertibility of paper currency into precious metal coins was not merely a privilege banks could dispense with with impunity as soon as they felt like it, but a binding commitment with real teeth to it, often consisting of the threat of outright failure, with shareholders on the hook for the losses--rather as is the case still today (so far at least) w.r.t. bank deposits convertible into central bank currency. Your "take" on convertibility has merit instead only for the last century or so, when central banks acquired the character of sovereign entities, capable of thumbing their noses at creditors who made the mistake of assuming that their promises could still be relied upon.

We may fantasize all we like about targeting schemes that mimic credible convertibility regimes despite lacking any clear mechanism for tying central bankers' hands. But the fantasies are just that, with no analogue in history to the very real historical record I refer to above. So, I repeat, convertibility does--or at least did, matter very much! And the lack of any real substitute for it matters very much today, no matter what economists may imagine central banks doing in the absence of such.

On the matter of what convertibility used to mean and why it ceased to mean it, it is now my turn to encourage you (and anyone else who may care) to read my paper!

[George Selgin]

@Mike Sproul: "If a central bank dumped all its assets in the ocean, would that have only a small effect on the value of its money, as opposed to reducing the value of that money to zero?"

Yes. It could have a big effect on the CB's earnings, of course, and on its net worth. But so long as there was a well-defined real demand for exchange media, which there presumably would be, a quantity that remained as limited as before, and a general understanding that the CB's notes were still the economies generally accepted medium of exchange, the assets wouldn't matter. That is also why a CB might swap relatively sound Treasuries for all kinds of doubtful MBS, and even do so while vastly expanding the nominal stock of base money, and yet not see any dramatic decline in its currency's value.

But all this you have heard a thousand times, Mike. I say it for the benefit of anyone else here who may be tempted to think that the value of an inconvertible fiat money rests, like that of a convertible one, on the value of its issuers' assets.

[George Selgin]

Nick, I see that by incorrectly closing italics on my penultimate post I have unintentionally committed myself to "shouting" at everyone. If you can fix that i will be much obliged to you!

[George Selgin]

@Nick (Referring to your "steps" post): "Now take the limit of that system, as the percentage of gold reserves approaches 0%. There is indirect convertibility of paper money into gold. The central bank never actually buys or sells gold, but it buys or sells bonds to keep the equilibrium price of gold pegged, so that people can still swap their notes for gold in the open market at a fixed price." Here is the step that I have been disagreeing with. It is fixing the price of gold. But it isn't convertibility, The two are distinct. Convertibility is not "fixing" the price of something--not originally, nor for most of banking history, nor for commercial bank deposits today. Paper currency consisted of binding IOUs--debt claims, that is, the dishonoring of which was an act of bankruptcy. A bank that offered to convert its paper into gold, or silver, was no more "fixing the price" of one or the other precious metal than a hat check room today fixes the price of client's cats by offering to return them for its paper tickets! The identification of convertibility with price fixing, which I believe is at the bottom of your notion that keeping the price of X constant is tantamount to "indirect convertibility" is anachronistic at best, in that it identifies a rather late and decadent stage of convertibility--convertibility in its death throws, as it were--with the real contractually-based McCoy.

[Andy Harless]

Let's see if I can close the italics here.

[Squeeky Wheel]

@Mike and @Nick

"4. What happens if it's a small country, with no monopoly power or network externalities? Do the central bank's assets then matter?"

From a practical perspective it seems to matter. Brunei runs a currency board and monetary agreement with Singapore with a 1:1 exchange between Brunei dollars and Singapore dollars (and I have used Brunei currency notes in Singapore). It appears this is doable because Brunei has a lot of sticky black stuff in the ground worth more than all of its base money.

Singapore operates a crawling trade-weighted exchange rate policy, no capital controls, and no interest rate policy (well, aside for the implicit interest-rate-parity). Singapore's central bank (MAS) has commented on the risks of a currency attack, and is aware that the size of base money is small enough to be vulnerable. MAS has taken on a few policies to deal with the possibility of attack - 1) it maintains sufficient foreign reserves to bank all paper notes, so what is in your wallet can always theoretically be exchanged for USD or Euros, etc. 2) while the CB assets are public, the total size of government assets is a state secret. 3) MAS has promised not to defend the exchange rate if under sufficient attack. That promise has been fulfilled once.

Back to monetary policy, some interesting facts about small country exchange rate targeting: Uncovered interest rate parity almost holds for USD/SGD (it doesn't for larger pairs like euro) and therefore tightening (raising the value of SGD) causes interest rates to go *down*. It seems that small one economies have a tiller steering instead of wheel steering.

[Nick Rowe]

My massively oversimplified version of George Selgin's paper:

An alpha-beta-gamma monetary system, with only one beta, is unstable. The beta will depose the alpha, and become the new alpha (and the gammas become betas). Gold was alpha, the Bank of England was beta, and the other banks were gamma. Only alpha-beta systems are stable.

(My massively oversimplified version of George leaves out the Hayekian Law-legislation distinction, which is central to George's paper.)

Interesting paper George.

[Nick Rowe]

Majro: I want to add one word to your good argument:

"To make this argument even more explicit, recognize that a fiat central bank has only one policy instrument, that being the quantity of base money. A [large] gold-converting central bank has two: the quantity of base money and the quantity of gold reserves."

Because for a small central bank, that second instrument will have next to no effect.

If there were a risk that the future demand for central bank money would drop permanently to near zero, then the present value of future seigniorage would not be a big enough asset to cover that risk. The value of the asset is smallest just when you need it to be biggest. The only way to cover that risk is for the central bank to have assets apart from future seigniorage. Good point.

[George Selgin]

Nick, I never expected to see my paper paper translated into Greek--and so efficiently!

[Nick Rowe]

George: on indirect convertibility:

Suppose a bank holds reserves of silver, converts money on demand into silver. And it adjusts the exchange rate between money and silver every morning, in proportion to changes in the market price of gold. This would mean that people could take their notes to the bank, exchange them for silver, then go to the market and exchange the silver for gold. And they could exchange them for an almost fixed quantity of gold. The only variance would be due to daily fluctuations in the price of gold.

But even under a pure gold standard, there is always a small variance due to the "gold points" (the buying and selling prices of gold) being slightly different.

The distinction between direct and indirect convertibility doesn't look very important to me. It's rather like a merchant who prices goods in US dollars, but only accepts Canadian dollars, at the current market exchange rate.

[Mike Sproul]

Nick:
"1. No effect, provided the expected present value of the central bank's seigniorage profits was large enough to exceed any likely fall in demand for central bank currency. (Unless the central bank chose to increase its inflation target to increase seigniorage profits to replace the lost assets.)

2. The central bank borrows against its future seigniorage profits. One way to do this would be by paying sufficiently high interest on reserves to increase desired reserves enough to increase the fall in demand for currency."

Some reasons not to have much confidence in that answer:
1. It means central bank money is backed twice: once by assets and again by future seignorage profits. Makes you wonder why the stock of circulating dollars doesn't double in value soon after it is issued.

2. Mexico issues 100 pesos in exchange for US bonds worth $100, and 1 peso=$1. It then spends the bonds on partying, and nothing happens to the value of the peso. The US, seeing the free lunch Mexico just got, encourages dollars to circulate in Mexico, so the free lunch goes to the US instead of Mexico. Makes you wonder why currencies like the peso have survived.

3. As private banks issue rival moneys (checking accounts, credit cards...) the demand for the central bank's money falls, and future seignorage profits are lost.

[George Selgin]

Nick, in both of these arrangements you have what I consider "real" convertibility, so in that sense I consider them similar. But the "indirect convertibility" to which you refer here is not the kind you refer to in your steps post, where there is in fact no longer any question of a note holder having access to the central banks' assets. I see the big difference--one of kind rather than degree--as being the one between the two notions of indirect convertibility.

[edited by NR to fix typo]

[George Selgin]

For my second "access" @ 11:10 please read "assets."

[Nick Rowe]

George: suppose I replaced "silver" with "government bonds", so the bank targets the price of gold by open market (bond market) operations rather than by silver market operations. Does it make any difference?

Or suppose instead the bank adjusts the interest rate it pays on money (easy with deposits, hard with notes) to adjust the demand for money so that the holder of money can always go to the market and buy a fixed quantity of gold.

Mike: I think that Majromax has the strongest argument in defence of your claim that central banks need assets. If there were a large and *permanent* drop in the demand for money, the present value of future seigniorage would drop too, and the bank would be unable to borrow enough against that future seigniorage to buy back the money that people did not want to hold.

"Some reasons not to have much confidence in that answer:
1. It means central bank money is backed twice: once by assets and again by future seignorage profits. Makes you wonder why the stock of circulating dollars doesn't double in value soon after it is issued."

Because the bank will be giving away those seigniorage profits, if it doesn't need them. Those profits will not be going to the note-holders.

[Derivs]

“The distinction between direct and indirect convertibility doesn't look very important to me. “

Because there is none.

If I folow the above example. I have $1,195 convertible dollars. I receive from the bank my 75oz of silver, and then, at a ratio of 75 to 1, I can then convert that silver into 1oz of gold.

Or I can have $1,195 pathetic non-convertible dollars, that I myself can convert directly for 1 oz of gold.

The value of the convertible and non convertible dollar are equal. Actually, they are convertible to each other at 1.

[edited to fix typo NR]

[Derivs]

my apologies.. s/b 75oz of 'silver'... error should be obvious.. point remains the same.

[csissoko]

" It is the commercial banks' responsibility to keep their exchange rates fixed. The alpha goes where he wants to go; and the betas follow the alpha. "

But as a matter of history (or more accurately British banking history), central bank reserves as we know them today were created for the purpose of supporting the commercial bank-based money supply. That is, the central bank's principle goal is to stabilize the money supply. In the 19th this had two components (based on Bagehot): (i) keep the money supply from growing in a way that led to the loss of gold reserves to other countries (today, approximately equivalent to the goal of controlling inflation) and (ii) keep the money supply from collapsing due to "panic" or a general movement to convert bank liabilities into gold, central bank notes, etc.

So the problem with characterizing central bank currency as alpha and claiming that betas must follow the alpha, is that the policies of the central bank are supposed to support the convertibility of commercial bank money into central bank money. And this is arguably (e.g. per Bagehot) one of the most important functions of the central bank.

Basically I object to the leader-follower framing. The whole problem with a bank-based monetary system, is precisely that it doesn't work that simply. The tail wags the dog all the time -- that's why the Fed's hand was forced in 2008.

(BTW, this point is basically from Hawtrey's Currency and Credit.)

[Nick Rowe]

csissoko: Yours is a fair point. When there's a run on a beta bank, the alpha leader will temporarily act like a beta. If there's a run on all the beta banks, the alpha might have to choose a different path, because the beta wolves can't keep up with him. The Bank of Canada would have to allow inflation to exceed its 2% target, to bail out the commercial banks.

If this happened all the time, and if betas were never disciplined for failing to follow the alpha and needing the alpha to bail them out, then my alpha-beta distinction would be meaningless.

[George Selgin]

@Dervis: your statement simply ignores the whole point of my previous comments. No one disputes that the schemes are identical in the sense of implying the same (hypothetical) value for the dollar.

@Nick: somehow I just can't seem to convince you that I'm determined to insist upon the literal meaning of "convertibility."

Convertibility into financial assets (e.g. bonds) meets the literal meaning test. But as a practical measure it leads to potential P indeterminacy if the bonds are denominated in the central banks own currency unit, because the nominal supply of bonds isn't a value independent of the stance of monetary policy.

Again, this is not a debate about whether a CB might in principle make the value of the dollar the same as it might be under a convertible regime despite the absence of any actual convertibility. Every textbook takes for granted a central banks ability to target whatever it wants. What neither the textbooks nor anyone here wants to grapple with is the very real difference between a hypothetical scheme, the enforcement mechanism behind which is unspecified ("Suppose that a central bank targets the price of gold...") and a scheme, otherwise similar, involving a definite enforcement mechanism ("Suppose a central bank is obliged to convert its paper into a fixed amount of gold on demand, or else be declared bankrupt.") The

[Nick Rowe]

George: OK. There is a difference between: specifying a target; specifying a mechanism to follow to hit that target. It's like the difference between "target rule/instrument discretion" (what all inflation targeting central banks do now) and "instrument rule".

[George Selgin]

Yippee!

[JP Koning]

Can we say that modern central banks are alpha? What if a large number of individuals buy current CPI goods and sell non-CPI goods and bonds (future CPI), raising the real price of current CPI goods relative to all other purchasable things. We get an increase in the equilibrium price of central bank currency against non-CPI stuff. And that's a deflationary tightening of monetary policy, even though central banks keep the exact same exchange rate of currency against CPI.

[Mike Sproul]

Nick:

Way to go Majromax!

But Nick, I don't expect it will make you mend your ways and accept the backing theory (except for small countries, as above). So what's your fall-back argument?

[Nick Rowe]

JP: if central banks stopped targeting CPI, and targeted something else instead, would their beta followers still follow them? If yes, then the central banks are alpha.

Mike: the existence of assets reduces the risk that the central bank will be unable to buy back money and allow higher inflation if the demand for money falls. That makes money a less risky asset, and will increase the demand for money. For a given supply of money, that would cause a slight increase in the value of money. But it will not be a 1 for 1 increase (like in your backing theory). Plus, the central bank will simply increase the supply of money to match increased demand.

Think of a central bank's assets as an insurance policy, against worst case scenarios. Again, except in that worst case scenario, the holders of money will never see the returns on those assets. The central bank gives those returns away to its favourite charity.

[Ben Kloester]

Mike, you always bring backing theory to these discussions, which I find very interesting, but I don't see how it can apply when there are no reflux channels, which seems to me to be the case in a regime without convertibility.

As Nick has pointed out, the central bank makes no promise to sell assets in exchange for currency on demand of the banks - I tink of this as being equivalent to Nick's alpha/beta distinction. So the store of assets is there, but not accessible on demand - ie there is no explicit guarantee that you can exchange your USD for CB assets (eg treasuries), so there is no reflux channel, and backing theory is irrelevant.

This was the conclusion that your debate with Scott Sumner came to here: http://www.themoneyillusion.com/?p=26884 (at least Scott decided there was no reflux channel and you seemed happy with that resolution)

[Jussi]

It looks to me that Sproul's arguments/questions are really tempting.

I can figure out a little island where the backing works like a charm. Money demand argument on the other hand is elusive and doesn't work at all if the economy is made small enough. I see how the network effect can work in a big economy but can it be a primary value driver?

Isn't it a coherent story to idealize the first person relying solely on backing when the first money trade was made. When did we switch to rely on networking effect? Looking how the central banks still keep the real assets suggest that we never did. IMO this gives backing a good doze of support.

[Nick Rowe]

Ben: true, but an inflation targeting central bank does promise to buy back its money if the demand for money falls so inflation starts to rise above target.

Jussi: if everybody else you trade with uses Canadian dollars, you will choose to carry a stock of Canadian dollars, even if some other asset yields a higher rate of return. Your stock of Canadian dollars will rise and fall over time, whenever you sell or buy things, but it would be very inconvenient to make that average stock arbitrarily small.

[Jussi]

@Nick

I fully agree, that is essentially network externalities argument right?

But how do get externalities going on without first having backing? Then at what point you can say the backing can be dropped off? Would you say the currency can hold its value without central bank assets or demand created by future tax liabilities?

[Nick Rowe]

Jussi: it sounds plausible to me that you need some sort of fundamental real value to get things started, so that people start using something as money, and it seems to fit the historical facts, AFAIK. Bitcoin seems to be maybe an exception though. JP Koning did a post on this. Maybe a couple of big players started trading Bitcoin, to get the ball rolling.

Language is a network externality. Words have the meaning they do, only because everyone expects them to have that meaning. How did language ever get started? Similar question.

[Jussi]

@Nick

Yes, I almost brought Bitcoin up myself. It is certainly interesting. I think you are right about how they got it rolling. Lets see if it is a just another ponzi scheme (I think it is). But if not it will over time prove that the externalities are enough to drive value. Yet we know many economists like Krugman has been doubtful of externalities being enough.

Yes, I read JP's blog, it is excellent!

I'm not sure I get the language argument. Payoffs of communication and money externalities seem very different from the start.

[Ben Kloester]

JP also wrote about orphaned currency where the backing was lost - value did not go to zero there either. Check out his blog post about the Somali shilling.

There is also the example of the Iraqi "Swiss" Dinar, which continued to be used in the Kurdish-controlled region even after it was no longer official Iragi currency, and due to debasement of the replacement currecy, retained its value far better than the Saddam dinars that succeeded it.

I've often found it useful to think of Bitcoin, and currency more generally, using the analogy of language so it's interesting you bring it up Nick.

[Ben Kloester]

The question about Bitcoin is, is it fated to be Esperanto, or will it be English? The Bitcoin true-believers are convinced it's the latter. I own Bitcoin, but I'm not sure it will ever be any more than Esperanto.

That said, it does allow some seriously cool tech that you simply cannot do with "money" as we know it today.