This archived content originally appeared at, the predecessor site to, and does not carry the sponsorship of the Cato Institute.

Backing, Shmacking

Somali Shilling

I really ought to know better by now than to complain about the "backing" theory of money–the view that the value of, not just some sorts of money, but money of any sort, depends on the assets commanded by the money's suppliers. I've never been able to convince the theory's more determined proponents of the error of their ways, and I risk provoking them to mount their hobby horse every time I bring it up.

But the backing theory keeps rearing its ugly head whether I mention it or not, as it did in the pages of Monday's Wall Street Journal, in an article asking "Do Cryptocurrencies Such as Bitcoin Have a Future?" And I'll be damned if I'm going to just sit idly by while it threatens to amble its way into some as-yet uninfected brains.

The specific perpetrator in this instance was one Eric Tymoigne, an assistant professor of economics at Lewis & Clark College, who invoked the ghastly fallacy in defense of his assertion that "There is no financial logic behind bitcoins’ face value":

Monetary instruments are securities. As such, they have a term to maturity (instantaneous) and an issuer—often a central bank or private banks—that promises to pay the bearer the full face value. Gold coins are a collateralized form of such security. Paper, cheap metal, and electronic entries are the forms such securities take today. The characteristics of these securities allow them to circulate at a stable nominal value (par) in the right financial infrastructure and as long as the creditworthiness of the issuer is strong. This provides a reliable means to complete transactions and, more important, service debts.

Bitcoins, meanwhile, violate all of the rules of finance. There is no central issuer guaranteeing payment at face value to the bearer; in fact, there is no underlying face value, and subsequently no imputed value at maturity, which means they are completely impractical for use in servicing of debt. The fair price of bitcoins as measured by the discounted value of future cash flows is zero.

Although Mr. Tymoigne never actually uses the word "backing," his argument is standard "backing" fare, distinguished only in being more explicit than some other statements in referring to all forms of money as "securities." This explicitness is welcome, for it is precisely in treating all forms of money this way, whether expressly or only tacitly, that the "backing" theory goes awry.

Many types of money are, to be sure, properly regarded as securities, that is, as IOUs or promises to pay. Bank deposits fit this description. So do competitively issued, non-legal tender banknotes, like those issued by Scottish banks today, and by almost all commercial banks a century or more ago. Central-bank-issued notes qualified as well back when they were themselves convertible claims to some underlying form of money, such as gold or silver coins. Finally, a paper currency which, though temporarily irredeemable, is expected eventually to be redeemed in some other kind of money–what von Mises terms "credit money"–also qualifies. Examples of the last include Bank of England pounds during the Bank Restriction period, and U.S. Treasury notes ("Greenbacks") before 1879.

Of course "backing" matters in the case of exchange media that either are convertible into some more basic form of money, or are expected to become convertible into such at some future date. The likelihood that a bank will be able to honor its promises to pay obviously depends on the quality of the assets "backing" its IOUs. When the expected present value of those assets ceases to equal or exceed the value of its liabilities, the bank is insolvent, and its creditors face the prospect of not having their debts paid in full. Likewise, the prospects for resumption of payments on temporarily suspended monetary IOUs vary with the value of their issuers' non-monetary assets. The announcement of the Peace of Amiens substantially boosted the value of the paper pound, while that of renewed hostilities had the opposite effect. All of this is indeed perfectly in accord with the logic concerning how securities of all sorts are valued.

But in imagining that other sorts of money are also securities, proponents of the backing theory commit what philosophers like to call a "category mistake." Free-floating "fiat" moneys are not securities. Precious metal objects are not securities. Finally, bitcoins and altcoins are not securities. These actual or potential exchange media aren't claims to some underling medium into which they are, or are expected to become, redeemable. Consequently their values don't depend on the likelihood that they can be converted into something else, or on their providers' overall wealth as it informs that likelihood. Instead, their value depends on their own supply, and on the demand for them as exchange media (and for other uses, if they have such), with no role for calculations or speculation concerning the prospect of their being converted into something else.

The assets possessed by a fiat-money-issuing central bank–that is, by a bank that issues inconvertible paper, with no promise to ever redeem it in a specific amount of some other money–play no part in determining its currency's purchasing power. Such a central bank might, therefore, swap all its better assets for junk, without influencing that purchasing power in the slightest. And guess what? The Fed and the ECB have done just that in recent years, jettisoning their good assets to make room for doubtful mortgage-backed securities and Greek bonds; yet far from seeing their currencies depreciate substantially, as the backing theory predicts would happen, both, and the ECB in particular, struggled to keep them from appreciating excessively relative to stated inflation targets. It's hard to imagine a cleaner refutation of the "backing" theory than this. Oh, wait! I almost forgot about the "disowned" Iraqi Swiss dinars that remained in use in northern Iraq years after being repudiated by Iraq's Central Bank, and the "orphaned" Somali shillings that continued to be valued and used long after looters gutted the Central Bank of Somalia. Alas, no amount of empirical evidence appears sufficient to refute the backing theory in the eyes of its more obstinate champions.

To be fair to those champions, they are merely giving their intellectual stamp of approval to a view that has always had a substantial popular following. Consider, for example, the definition of a fiat money system as one "based on a government’s mandate that the paper currency it prints is legal tender for making financial transactions. Legal tender means that the money is backed by the full faith and credit of the government that issues it. In other words, the government promises to be good for it."

"Promises to be good for it?" And just what is this "it"? It is…well, nothing, actually. Because modern Federal Reserve Notes are, in fact (in former New York Fed President John Exter's famous formulation), so many "IOU nothings."

It's a bit more disappointing to find the backing theory endorsed by government officials, as it is in an otherwise competent CRS brief on the history of Gold Standard:

The U.S. monetary system is based on paper money backed by the full faith and credit of the federal government. The currency is neither valued in, backed by, nor officially convertible into gold or silver. Through much of its history, however, the United States was on a metallic standard of one sort or another

and in boilerplate included in the U.S. Comptroller General's annual Financial Report on the United States Government:

FRBs issue Federal Reserve Notes, which are the circulating currency of the United States. These notes are collateralized by specific assets owned by the FRBs, typically U.S. government securities. Federal Reserve Notes are backed by the full faith and credit of the United States Government.

In fact the only "faith" that matters for a fiat currency is the public's "faith" in the monetary authorities, that is, the public's willingness to believe that those authorities will refrain from issuing oodles of the stuff. "Credit" has nothing to do with it: once again, a piece of "fiat" money is not a credit instrument. It is not a promise to pay anything. Consequently its issuer's credit standing is of no immediate consequence. (It matters indirectly because, were the U.S. government unable to borrow by selling actual Treasury securities, it might be tempted to issue greater sums of fiat money instead, provided it could get the Fed to go along with the plan.)*

More troubling still are endorsements of the backing theory by Federal Reserve experts. Those experts are supposed to know something about how the value of the U.S. dollar is determined. Yet that doesn't prevent some of them from saying things like "Since 1971, U.S. paper money has been backed by the 'full faith and credit' of the U.S. government", or "Compared with commodity money, which has an intrinsic value, such as gold, or official fiat money backed by a sovereign entity, the current market value of bitcoin to any given user hinges entirely on her expectation of others’ willingness to accept it later at a sufficiently greater value". In truth neither bitcoin nor Federal Reserve dollars have "intrinsic" value, understood in the usual sense to mean non-monetary value. And as for governments being able to prop-up the value of fiat money by means of legal tender laws, tell that to all the governments that imagined that doing so would suffice to rule-out hyperinflation.

Most disturbing of all is the fact that some central bankers themselves appear to subscribe to the backing theory. Take, for example, Alan Greenspan's remarks concerning Bitcoin, given in the course of a December 2013 Bloomberg News interview. :

"It’s a bubble,” Greenspan, 87, said today in a Bloomberg Television interview from Washington. “It has to have intrinsic value. You have to really stretch your imagination to infer what the intrinsic value of Bitcoin is. I haven’t been able to do it. Maybe somebody else can.”

“I do not understand where the backing of Bitcoin is coming from,” the former Fed chief said. “There is no fundamental issue of capabilities of repaying it in anything which is universally acceptable, which is either intrinsic value of the currency or the credit or trust of the individual who is issuing the money, whether it’s a government or an individual.”

"Capabilities of repaying it"! Tell us, Mr. Greenspan, just when does the Fed plan to "repay" the holders of all those Federal Reserve "liabilities" created during your watch, let alone the far vaster quantity created since? And what will it "repay" them with?

If you imagine that Mr. Greenspan can answer these questions, you imagine too well.**
*Even when used, less inappropriately, in reference to actual U.S. government securities, the "backed by the full faith and credit" language is misleading: unlike private debtors the U.S. government doesn't have to possess a stock of valuable, productive or interest-earning assets in order to service its debts. Its securities are instead "backed" by a combination of (1) its power to tax and (2) its ability to rely, at least to some extent, on the Fed's willingness to monetize its debt.
**A commentator thinks I am being too harsh here, as Greenspan merely assumes that Federal Reserve dollars, being "universally acceptable," differ from bitcoins in not having to be secured by a promise of repayment. Greenspan's language isn't the easiest to follow, but I still think his statements beg the question. For my further remarks see the comments below.

  • Koen

    While I agree with your overall point, I'm not sure your interpretation of the second paragraph of the Greenspan quote is correct. I don't think he means that e.g. US dollars are backed by a promise to pay the holder with something else, only that bitcoin would have to be backed by such a promise to pay their holder with something that *is* universally accepted, e.g. US dollars or a commodity

    He'd still be wrong (because a medium of exchange doesnt need to be backed by either a commodity or a promise to pay with something else), but it would not be correct to ask him ' just when does the Fed plan to "repay" the holders of all those Federal Reserve "liabilities" created during your watch, let alone the far vaster quantity created since? And what will it "repay" them with?'

    • George Selgin

      Koen, I don't see that we really disagree. If Greenspan does in fact suppose that a medium of exchange has to be backed by a commodity or the promise to pay something else (as you suggest, and as I also believe to be the case), then he must believe this to be so w.r.t. Federal Reserve dollars themselves. But if that's the case, we are entitled to ask what those dollars promise to pay.

      The point is that whatever allows fiat dollars to command value may, in principle, allow bitcoins to command value as well, to wit, the willingness of substantial numbers of persons to receive them in payment. Bitcoins' actual foothold is of course very tenuous, and they are to that extent not comparable to dollars. But dollars do not owe their value to any features or conditions apart from those that might also serve to keep bitcoins from becoming worthless. To put the point more pithily: if bitcoin can be regarded as a bubble, so can any paper fiat money.

      • Koen

        I think Greenspan would argue (Krugman and many others do) that the difference between unbacked dollars and unbacked bitcoins is that the government requires people to pay taxes in dollars, thereby causing a more or less guaranteed demand for dollars.

        So even though neither bitcoins nor dollars are backed by commodities or by promises to pay in something else, dollars are 'backed' by the more or less guaranteed demand for them that is the result of people having to pay taxes in dollars.

        • George Selgin

          Yes, koen: this is a very frequently heard argument. But it is also wrong. Public receivability of a money certainly contributes to its general acceptance. But it is neither necessary nor sufficient for that purpose, much less for preserving a fiat money's value. On the other hand, if the top twenty retailers of the U.S. all agreed to accept some unofficial currency, their collective willingness to do so would rival the Federal government's willingness to receive Federal Reserve dollars.

          • Koen

            Yep, I fully agree.

          • dufus

            Your argument is very plausible and possibly true. It's also true that fiat money is not backed by a promise of repayment with some commodity.

            Yet there is one detail which is being ignored.

            The backing of fiat money by a government arises through the nation's economic potential and political structures within. A government can back/defend a currency through legislation, fiscal and monetary policies.

            How can such retailers or bitcoin do so, within an organized society?

      • Chris LeRoux

        Money is the most voluntarily saleable good. All these efforts you make to justify using substitutes are rationalizations for aggression, for tyranny. Gold is the most saleable good currently, and thus gold is money. It is only violence which prevents people from circulating it. Remove the violence and your digital-paper counterfeit will collapse. Regardless, bitcon is trivia, a collector item of counterfeit for utopian, central designers/planners.

  • Lawrence Kramer

    George –

    Let me take this opportunity to agree with you for a change. You make two excellent points:

    "the only 'faith' that matters for a fiat currency is the public's 'faith' in the monetary authorities, that is, the public's willingness to believe that those authorities will refrain from issuing oodles of the stuff."

    Where I would part from you, perhaps, is in calling the "backing theory" a theory rather than a metaphor. What people tend to call backing is really just the evidence available that the aforementioned oodles will not be forthcoming. If money can only be issued for gold, or for mortgages (rentenmarks) or for other known valuable consideration, then there cannot be too much of it. We say "the money is backed by" whatever, but we really mean "the scarcity of the money is proved by" whatever. The technology of proof has evolved over the years. Specie is self-proving, a very crude but effective technology for its day. Redeemability from someone with a good reputation – which may not be possible without the printing press to disseminate reputation – is a technology. Eventually, common knowledge that political forces are adequate to suppress oodling – requires governmental technology, a free press, and adequate new dissemination techniques – becomes "backing." It's all about the evidence of reliable scarcity, rendering "backing" a mere metaphor, with some sort of retronym like "real backing" or some such to describe a requirement for creation of money for value other than the sovereign's say-so. (If we regard Treasury debt as tax credits – because you can use money issued against them to pay taxes, maybe dollars are backed by those credits, at least on a "fractional reserve" basis.)

    • Koen

      These are good points, but there may still be value in reserving the term 'backed' for situations in which a money is (legally) guaranteed to be redeemable for a fixed amount of something else (a commodity, other money, some other promised thing or service), which would mean that fiat money is unbacked. (one could say that the dollar is backed by the promise that the government won't send you to jail, although in that case there is no fixed amount of dollars that entitle you to that 'service')

      What really matters in terms of potential value is indeed the scarcity of the money, and scarcity can be enforced by the physical nature of the money or what the money is backed by (e.g. commodity money), by restraint on the part of the issuer in the case of fiat money, or in the case of bitcoin, by the software that governs its supply and the network that enforces the security.

      • Lawrence Kramer

        "there may still be value in reserving the term 'backed' for situations in which a money is (legally) guaranteed to be redeemable for a fixed amount of something else"

        I would agree that there is utility in having a term that describes such a situation; I'm just saying that "backed" has been hijacked and simply is no longer available for the job. Hence my reference to "actually backed," not because I like it, but because I share your view that the concept needs a locution; I don't much care what is used, although maybe you put your finger on it with "redeemable."

        I think the train leaves the rails when we start looking for a reason that money has "value" or "purchasing power." Those are reifications. The question is "Why do we accept something AS money?" Common acceptance is clearly the short answer – we accept it because we can spend it? Is that circular? What holds up an arch? The arch holds up the arch. You cannot look at any single stone, even the keystone, and ask what holds it up. The medium of exchange is the keystone in a barter system that uses a common medium of exchange. The keystone has to have certain attributes, but for it to be part of an arch, it has to be surrounded by other stones assembled by some Invisible Hand because there exists a keystone around which they can form an arch. But the central stone is only a stone shaped like a keystone until it is in an arch; only then does it BECOME a keystone. Examining a currency outside the arch of the commerce may tell us why it CAN be money, but not why it IS money.

  • RalphMusgrave

    Not entirely on topic this, but it strikes me that since those issuing Bitcoins do not promise to redeem them for US base money, Bitcoins are a new form of worldwide base money. Plus it seems to me that existing issuers of base money (i.e. the US, UK, Japanese etc governments) won’t want any serious competition of that sort, so they’ll clamp down on Bitcoin if it gets too popular. Its already banned in Russia, I think.

    • George Selgin

      I would say that bitcoins would be base money if they were money of any sort. In fact they have a long way to go before they can qualify as such.

  • laurelcreek

    We might say fiat money is "backed" by the goods and services we expect to get for it. We could say the same for specie.

    • George Selgin

      laurelcreek, your argument is circular: it amounts to saying that fiat money has purchasing power because, and so long as, it…has purchasing power.

      • elriel

        To me that reads like "fiat money has purchasing power as long as people _believe_ it has purchasing power." The critical word in the original sentence being "expect". This is not really a circular argument but rather a spiral argument.

  • Mike Sproul

    "a paper currency which, though temporarily irredeemable, is expected eventually to be redeemed in some other kind of money–what von Mises terms "credit money"–also qualifies. Examples of the last include Bank of England pounds during the Bank Restriction period"

    The BOE suspended gold convertibility from 1797-1821. During that period, let's say that people thought there was an 80% chance that gold convertibility would be restored. Also, the BOE maintained two other forms of convertibility: (1) Bond convertibility: The BOE stood ready to use its bonds to buy back refluxing paper pounds, and (2) Loan convertibility: The BOE accepted paper pounds in repayment of loans. The BOE recognized its paper pounds as its liability, and held assets against those pounds.

    The Fed suspended gold convertibility from 1933 to the present. Let's say that people today think there is a 10% chance that gold convertibility will be restored. In the mean time, the Fed maintains bond convertibility and loan convertibility. The Fed recognizes paper dollars as its liability, and holds assets against those dollars. Some day, if a superior form of private money is invented, people will return all their paper dollars to the Fed. The first wave of refluxing dollars will be redeemed for the Fed's bonds and loans. Once the bonds and loans are gone, the Fed will use the rest of its assets (buildings, gold, etc) to buy back the last of the refluxing dollars. The Fed couldn't do this if it didn't have assets.

    In spite of these similarities between the Fed and the BOE, you say that BOE pounds were financial securities, whose value was determined by BOE assets, while the Fed's dollars are unbacked fiat money, whose value is wholly independent of the Fed's assets. You talk like convertibility is black and white, when it's obvious that there are different kinds of convertibility, and each kind is a matter of degree.

    That's as much as I can write for now, but within a week I'll try to answer the rest of your points, starting with:
    1) The backing theory does not always imply that a loss of bank assets causes inflation. By analogy, a corporation can lose assets and its bonds might hold their value, even as its stock falls.
    2) Somali currency, Dinars, and bitcoins are not a problem for the backing theory, and they do not amount to significant empirical evidence. The few empirical studies that have tested the backing theory (by Cunningham, Sargent, Siklos, Bomberger and Makinen, etc) have supported the backing theory over the quantity theory.
    3) It means something to say that the government is "good for it". For example, the government will redeem its dollars for taxes or other debts owed to it.
    4) The government does not need to refrain from issuing oodles of the stuff. It only needs to acquire oodles of new assets at the same time. (as evidenced by our experience with Quantitative Easing).
    5) " And what will it "repay" them with?"
    Answer: With its assets, of course.

    • George Selgin

      1) The backing theory certainly does imply that currency values follow asset values, ceteris paribus–or else it implies nothing at all. You choose.

      2) What makes the evidence in question "insignificant," as opposed to inconvenient? And the studies you mention: are they studies of irredeemable fiat moneys or of something else–like, say, Civil War greenbacks? It matters which: as I said, the backing theory makes sense for some types of money, but not for others.

      3)That the government lets me pay taxes in Fed dollars signifies no more than that Wal-Mart let's me pay for goods in them. These are mere fluctuating-value exchanges. There's no fixed rate commitment.

      4)I see. So all those hyperinflations, in which every unit of money issued was issued in exchange for assets of some kind, weren't hyperinflations at all, since the currencies on question were fully "backed." Got it.

      (Goodness, Mike: do you really not get it? Do you really not understand that so long as an issuer of paper money isn't promising to return assets at a predetermined price, there's no limit to how many assets it can acquire, or to how much those assets' nominal prices can rise, along with the prices of all other goods, as the issuer acquires them! Do you not understand how dangerous and misguided your theory is? If not, I hope others do!)

      3)So, just when can I cash in my Federal Reserve Notes for some of those Fed assets?

      • kduda

        Yes, MIke is deeply confused.

        Mike: what are you thinking? The Fed has made no commitment to redeem dollars for assets.

        It has made a very important commitment, which is to stabilize the purchasing power of the dollar. To do that, the Fed needs to contract the monetary base from time to time, and one way it does *that* is by selling assets in the open market and destroying the dollars. The markets understand all of that, leading to the self-fulfilling prophecy of stable dollar purchasing power. If all the Fed's assets suddenly turned into dust, I do think that the markets would be a lot more worried about the Fed's ability to preserve the dollar's value, especially given the size of the monetary base; if velocity returned to pre-2008 norms and the Fed didn't shrink the base, M*V = P*Y would leave us in an exciting place…. So in some sense, it is important (for the dollar's value) that the Fed can use its assets to suck dollars back out of the economy. But countercyclical monetary policy is hardly the same thing as redeemability.


        • Mike Sproul

          I used to be confused about money, from about 1976 through 1989. This was the period when I believed the quantity theory of money. Inconsistencies and paradoxes kept cropping up. Why do central banks hold assets? Why do people hold currencies like the Mex. peso? Why don't they all just hold dollars? Why does the Fed's issuance of money seem to have the same effect as if the money were counterfeited? What happens to the value of the dollar when credit cards are invented? Why does fiat money violate the fundamentals of accounting and finance?

          After about 5 years spent learning about the real bills doctrine and the law of reflux, most of my confusion ended. I came to understand that paper money is valued based on its backing, just like stocks, bonds, etc. The paper dollar does not violate the principles of accounting and finance, and there is no need for any "special" theory of money. No need for circular arguments like "I value it because he values it.", or for silly tautologies like MV=Py.

          These days the think that confuses me the most is why economists accept the quantity theory and reject the real bills doctrine. I suppose it's because the real bills doctrine was developed by practical bankers over centuries of experience, while the quantity theory was developed by eggheads. Eggheads tend to favor their own.

          As for your claim that the fed needs assets to suck up dollars. You could just as well say that about shares of stock. You could claim that shares hold their value because the firm stands ready to reduce their quantity, but that's clearly wrong. The truth is that shares hold their value because the firm's assets back its shares. If a firm used 20% of its assets to buy back and retire 20% of its shares, that 20% reduction in the quantity of shares would not raise the price of shares. The share price would stay the same, since the firm's assets fell in step with its shares. The same is true of money.

    • Koen

      Mike wrote:
      "And what will it "repay" them with?"
      Answer: With its assets, of course."

      Not sure I understand. How would those assets be priced? Who sets the exchange rate between the dollars and the Fed's assets? Are they set on the basis of typical prices for similar assets in the market?

      • George Selgin

        Right question. If there's a set, guaranteed nominal price, the offer to "repay" amounts to a meaningful guarantee. If not, not.

        In the case of the Fed, we haven't even an actual offer to stand ready to surrender assets for notes at any price. If we had such an offer, but with assets surrendered only for their current market price, it would be of no significance.

        • JP Koning

          "In the case of the Fed, we haven't even an actual offer to stand ready to surrender assets for notes at any price."

          George, I disagree that the Fed does not offer to surrender assets at any price. See here for specific evidence from the Federal Reserve Act. However, I agree that it doesn't set a guaranteed nominal price. There's no evidence in Fed documents that specifies an actual price at which it will cancel its liabilities.

          • JP Koning
          • George Selgin

            JP, you refer to a technicality that doesn't contradict my claims at all. The "collateral" requirements to which the law refers are vestiges of the original Federal Reserve Act, that is, written when those notes were actually claims to gold. The law has never been modified to reflect the present, irredeemable status of Federal Reserve notes. But it is basically a dead-letter. In any case it doesn't imply either (1) that Federal Reserve dollars aren't capable of depreciating to an arbitrary degree, depending on how many are created relative to the real demand for them; or (2) that the general public can ever expect to have access to the assets in question.

          • George Selgin

            JP, some U.S. Treasury info concerning why the law remains as it does, which explains the one circumstance in which both the "first lien" status and the "collateral backing" of Federal Reserve notes might still have significance:

            The idea was that if the Congress dissolved the Federal Reserve System, the United States would take over the notes (liabilities). This would meet the requirements of Section 411, but the government would also take over the assets, which would be of equal value. Federal Reserve notes represent a first lien on all the assets of the Federal Reserve Banks, and on the collateral specifically held against them. (My emphasis.)

            I think you will admit that this is a very special case–and one that doesn't imply any sense in which FR notes can be said to constitute genuine liabilities from the point of view of the general public. It is a mere matter of allowing Congress to dissolve the Fed and let the Treasury take over responsibility for its (fiat) currency.

          • JP Koning

            George, you say that the wording in the Federal Reserve Act is the vestige of an era when those notes were actually claims to gold. But the Bank of Canada Act also includes similar wording, as my link shows (i.e. notes are "a first charge on the assets of the Bank"), yet the Bank of Canada was formed after Canada had forsaken the gold standard and had begun to issue irredeemable media. I'm not convinced the wording in question is as vestigial as you claim.

          • George Selgin

            JP, I think my other comment recognizes your (valid) point. The language was in the old act, but there was reason for Keeping it beyond the era of gold convertibility.

          • JP Koning

            With regards to your other comment, James Laurence Laughlin, an indirect architect of the Act via Henry Parker Willis, claimed that:

            "Although not so stated literally, the notes are liabilities of the Reserve Banks, since they must redeem them, and since the notes are a first lien on all the assets of such banks."


            …which is a more general statement than the 'special case' you have found in your US Treasury link. But even if you stick to your guns regarding the Fed, you must admit that, given the wording in my earlier link, the Bank of Canada seems to have an actual offer to stand ready to surrender assets (although not at a set price).

          • George Selgin

            JP, Laughlin's work was written when FR notes were convertible into gold, so referring to his work doesn't address the point at all. As for the Bank of Canada, the general view at the time was that gold convertibility would eventually be resumed.

            But all this nit-picking of documents seems to me besides the point. If you want to prove your point, why, just go to the Fed, or the Bank of Canada, or any other central bank you like, directly or by making a request of your banker, and ask to "cash in" any quantity of its notes for something–anything–other than a fresh batch of the same. (I exclude here banks that operate fixed exchange rates with other fiat currencies, as I have consistently in this discussion.) When you succeed in getting hold of some non-central-bank-created assets this way, let me know, and I promise to eat a dozen hats, or a thousand for that matter!

          • George Selgin

            Alas, JP, now it seems I have two people to bicker with! But in any case the possibility that the Fed might be "dissolved" in any sense that would give public FR note holders, rather than the U.S. Treasury, claims against its assets, is one for which there is, so far as the documents we've been discussing are concerned, not the slightest possibility. Moreover, if there were such a possibility it would not supply a scintilla of evidence in favor of Mike's theory, for the "backing" in that case is a nominal backing only, the nominal "quantity" of which is free to adjust with other prices. As far as it is concerned we are in a nominal indeterminacy situation that in no way serves to determine or anchor the dollar's purchasing power. Insisting as you do on the purely hypothetical possibility that the Fed would be liquidated, with its assets distributed among its creditors, merely gives the appearance of merit to a theory (or set of theories, for we are in "real bills" territory also) that is in fact bereft of merit.

        • JP Koning

          George, I think you've misunderstood me. I'm talking about payout on windup, like what a shareholder gets when the company is dissolved. Not instant redemption. Going to a central bank that hasn't yet been dissolved and asking for assets, as you suggest, won't settle the matter.

          I am nitpicking because the fine print is the only way to resolve the eternal bickering between you and Mike Sproul, at least as far as I see it.

          • Koen

            So in conclusion: People can't redeem their dollars at the Fed in exchange for a fixed amount of Fed assets per dollar, and quite possibly the Fed will not be willing to exchange any non-USD assets for dollars should somebody ask them to.

            At the same time, according to the rules, holders of dollars are entitled to Fed assets in case the Fed dissolves, although it is doubtful that should such an event occur holders of dollars will in fact receive any Fed assets.

            The only way in which one could say that dollars are 'backed' by the Fed's assets is through a liberal interpretation of 'backed', namely in the sense that the Fed can to some extent / with some degree of confidence stabilize the price of dollars and provide a floor and ceiling for that price by increasing and decreasing the supply of dollars through the buying and selling of assets. This ability and willingness on the part of the Fed makes causes people to have confidence in the dollar, which in turn helps to maintain its value.

      • Mike Sproul

        Think of the BOE restriction period. Say the BOE's assets add up to 100 oz of gold, while the BOE has issued 400 paper pounds. People will value the pound at 1/4 oz, even though gold convertibility is delayed and uncertain. Next, the BOE prints up another 4 pounds and uses them to buy bonds that are themselves worth 4 pounds. No problem. The bonds don't have to be denominated in gold. They can be denominated in pounds. The following year, the BOE sells those 4 pounds worth of bonds on the open market in exchange for 4 of its own paper pounds, and burns the pounds it receives. They are back where they started, and at no point would the value of the pound change from 1/4 oz of gold.

        • George Selgin

          BoE notes during the restriction were "credit money," and, as I explained, that meant that "backing" did matter for them.

          But it didn't matter in anything like the mechanical way you suggest: the prospects for peace and, hence, for an eventual resumption of gold payments, were a far more important determinant of the paper pound's value relative to gold than the Bank's actual gold holdings.

          • Mike Sproul

            Once again you see black and white when there are only shades of gray. The only difference between BOE notes and FR notes is that gold convertibility has been suspended for a longer period for FRN's. The fed holds gold, loans, and bonds as assets against its notes, just like the BOE did. The fed uses its loans and bonds to buy back its notes, just like the BOE did. Most of us expect that if the Fed were liquidated, then its assets (including gold) would be used to redeem its notes. The same was true of the BOE.

            And yet you keep insisting that the BOE notes were "credit" while the FR notes are "fiat".

          • elriel


            I think you're beating a dead horse here with this black and white thing.

            If people think there's only a 10% chance that FR will go back to actually converting their notes to something else, it's different enough from people thinking there's 90% chance that it warrants using a different term. If people think there's a 90% chance that something is credit, then it's still overwhelmingly valued like credit. If people think there's a 10% chance that something is actually credit, that will still affect it's valuation a little, but hardly enough that it'd act like credit anymore.

          • Mike Sproul


            There are many kinds of convertibility, and each of those kinds has varying degrees. The mistake that quantity theorists make is to think that "inconvertible into gold"="unbacked". A bank might issue notes that are gold-convertible during business hours, but over the weekend those notes are gold-inconvertible for 2 days, but they are still backed by the bank's assets. If the bank lost half its assets on saturday night, then by sunday its notes will lose half their value.

            If gold inconvertibility is extended to 30 days, 30 years, or 200 years, but the assets are still there in the bank's vault, then the bank's notes are still backed but inconvertible. This is most obvious if the bank maintains bond convertibility and loan convertibility throughout, as all modern central banks do.

  • kduda

    George, thank you for a very nice article. I am always amazed that more people don't understand that the only thing that gives US dollars value is the fact that hundreds of millions (if not billions) of people believe they have value now and expect that they will have comparable value arbitrarily far into the future. It's one global-scale self-fulfilling prophecy. The only "faith and credit" of the US government that's required is a credible commitment that the Fed will faithfully regulate the supply of dollars so that they maintain that value.

    Fiat money is so beautiful, it brings a tear to my eye.

    It drives me nuts when people insist that a money issuer guarantee redeemability into commodities (gold). Or people who think that the Federal Reserve "liabilities" of currency are "backed" by "aseets" of government bonds… talk about not getting it.

    Anyway, thanks again.


    Kenneth Duda
    Menlo Park, CA

    • George Selgin

      Thanks for your remarks, Ken.

      Of course, redeemability into some real asset like gold isn't necessary to make paper money command value. Still, practically all modern fiat money owes its existence to the fact that it was once convertible, either into some other paper money, at a fixed rate, or into precious metal money. Had there been no gold standard money in the past, it is not clear how we could have gotten here! So when you next shed a tear for fiat money, think a nice thought for gold also!

      Personally, I'm inclined to go rather further than to value gold merely as a stepping stone to fiat money: I'm convinced that no fiat system has even come close to matching the overall success of the prewar, classical gold standard, which has a very impressive record of internal stability (especially if one distinguishes instability due to bad banking structures from that attributable to the workings of the gold standard per se) and an unparallelled one for international (exchange rate) stability. Today economists are too quick to dismiss gold as a "barbarous relic" or whatever: they forget that for decades after the close of WWI, which had left the classical gold standard in tatters, experts struggled mightily though unsuccessfully (and with tragic results in the case of their first, misguided approach) to replicate its generally-acknowledged success. Viewed against the backdrop of this history, nostalgia for gold today is far from being irrational or inexplicable.

  • Mike Sproul

    1) The backing theory implies that money is valued like bonds. When the assets backing those bonds drops below a certain level, then the bond's value follows asset values. But when assets values are sufficiently high, the bond value is unaffected by changes in asset value.
    2) How much do you trust monetary data coming out of Somalia and Iraq? How significant is bitcoin in relation to the US dollar? The bibliography to my paper entitled "There's No Such Thing as Fiat Money" references the studies by Cunningham et. al. The studies generally looked at irredeemable currencies that were considered "fiat" by most economists, and concluded that the backing theory explained their value better than the quantity theory.
    3) Currencies get value because their ISSUERS accept them as payment, and thereby back the currency with their own assets. Once that value is established, Walmart will also accept them. If we regard the Fed as part of the government, then the dollar is backed both by the fed's assets and by the government's assets, and the government's assets include its "taxes receivable".
    The BOE had no fixed rate commitment from 1797-1821, but you agreed that the pound would have been valued according to its assets.
    4) Hyperinflations happen because the amount of currency outruns the issuer's assets. For example, a government issues $100, and holds assets worth 100 oz of silver as backing for the dollars, so people value the dollar at $1=1 oz. Then the government gets on the losing side of a war and starts paying its bills by printing another $200, while assets remain constant at 100 oz. The dollar will fall to 1/3 oz. If it then loses half its assets (with the quantity of dollars still at $300), then $1=1/6 oz.
    5) You can cash in your FRN's either by buying a bond from the Fed, or by repaying a loan from the Fed. You might even cash them in by buying some used furniture from the Fed.

    • vikingvista

      "Hyperinflations happen because the amount of currency outruns the issuer's assets."

      What is the mechanism by which knowledge of the money issuer's assets reach general users of that money, to cause those users to demand and accept different prices?

      • Mike Sproul

        The same mechanism that makes holders of stocks and bonds behave the same way.

        • vikingvista

          Which is?

          • Mike Sproul

            Not sure what you mean. If a bank issues 100 bits of paper called dollars, and if the banks assets fall in value from 100 oz. to 60 oz, then how long of a story does it take to convince you that the dollars value will fall from 1 oz. to .6 oz.?

          • Koen

            Mike wrote:
            "Not sure what you mean. If a bank issues 100 bits of paper called dollars, and if the banks assets fall in value from 100 oz. to 60 oz, then how long of a story does it take to convince you that the dollars value will fall from 1 oz. to .6 oz.?"

            But what if the demand for those dollars has gone up in the meantime, for example because people in another country start to use these dollars as a medium of exchange? Or what if the economy has grown by like 10% in the meantime?

            Wouldnt it be that kind of supply and demand that determines the price of those dollars instead of or in addition to the relation between the total supply of dollars and the total supply of gold that the Fed has?

            Btw, do we know how much gold the Fed has?

          • vikingvista

            "If a bank issues 100 bits of paper called dollars, and if the banks assets fall in value from 100 oz. to 60 oz, then how long of a story does it take to convince you that the dollars value will fall from 1 oz. to .6 oz.?"

            So hyperinflation, as you imagine it, requires public knowledge of the total assets of the money issuer? Does that mean that a money whose issuer does not reveal is assets cannot suffer hyperinflation?

          • Mike Sproul

            Koen (7:09PM):
            A bond market analogy might help. Some bond is maturing and promises to pay $100 to its holder. The backing theory says that the bond would trade in the market for $100. You can't then come along and say that foreign demand for those bonds goes up by 10%, and so the bond rises to $110. The increase in demand won't happen in the first place, so the bond stays at $100.

          • Mike Sproul

            "Does that mean that a money whose issuer does not reveal is assets cannot suffer hyperinflation?"

            Such a bank would never get any customers to begin with. But here's another stock market analogy: Say GM's factories all burned down, but nobody knew it for 2 hours. Pretty clear that GM stock would hold its value for those 2 hours.

          • vikingvista

            "Say GM's factories all burned down, but nobody knew it for 2 hours. Pretty clear that GM stock would hold its value for those 2 hours."

            Are you saying that FRNs have market value because the market *doesn't* know the value of the FRB's assets (like GM's 2 hours), or because it *does*?

  • Matt Young

    IMHO, the currency is backed by the recruitment of smart member banks. I dunno how that works.
    The actual currency function is a spreadsheet that manages deposit and loan rates, and publishes the distribution of deposits and loans transaction. The spreadsheet guarantees that loan assets over deposit assets are unhedgable, and changes rates to make that happen. I don't see much else, except ability to recognize a smart banker, (do they wear shoes?).

  • Paul Marks

    Yes economic value is subjective.

    However, if a commodity does not have economic value BEFORE it is used as money (if it does not demanded apart from its monetary role) then things are not good.

    That is my problem with Bit "coin" – who actually valued these "special numbers" in a NON monetary role?

    As for "backing" generally – I am not interested.

    Either the commodity (gold, silver whatever….)is the money or it is not – I am not happy with 1920s style "backing".

    "backing" or "standards" do not stop "broad money" (bank credit) expansion – do not stop a boom-bust.

    If lending is not from real savings (the sacrifice of consumption) then there is an artificial credit bubble "boom" which must end in "bust" as the "broad money" (bank credit bubble) shrinks back down toward the monetary base.

    "backing" and "standards" miss the point.

    Either the money is a commodity or it is not.

    And either lending is from real savings or it is not.

    And, contra Keynes, credit expansion is not "saving – as real as any other form".

    • Koen

      "However, if a commodity does not have economic value BEFORE it is used as money (if it does not demanded apart from its monetary role) then things are not good."

      Why not? Why would it matter whether something originally had a non-monetary role or not?

      If it didn't have a prior non-monetary role then its first price will be 0+x and if it did have a prior non-monetary role its price will be >0+x. In both cases once it acquires a monetary role the price jumps by x, and it shouldn't matter what the starting point was, whether it was 0 or >0.

      Interestingly and somewhat off topic, the more successful a good becomes as a medium of exchange the less successful it will be in its prior non-monetary role (if it had one, that is). If a commodity starts being used as a medium of exchange, its price will go up. The more it's so used, the higher the price will be. This price increase is not a problem for that good's use as a money (in fact, it's a good thing), but it is for its original non-monetary use. Due to substitution effects the price increase of the good will cause a decrease in the demand for the original non-monetary function of the good, as people will switch to other goods performing the same service for a lower price.

      So over time this mechanism would lead to an ever smaller (in both absolute and relative terms) non-monetary role for whatever commodity is now being used as a medium of exchange / money.

      Exceptions are goods that:
      – actually becomes more popular the higher its price. Gold's use as jewelry is a good example, and in that sense gold would be one of the very few forms of money that could still have a non-negligible non-monetary use
      – never had a non-monetary use (or only a negligible one)
      – goods where the non-monetary and the monetary functions are inseparable (but where the unit-price for the non-monetary roles of/made-possible-by the good is adjustable downward independently of the unit-price of the good itself)

      Bitcoin arguably falls into one or both of the latter categories (and qua store of value, perhaps an argument could be made it falls into the first category as well)

    • vikingvista

      You diminish the fact that the monetary role–a medium of exchange–is extremely valuable to most people in and of itself. As we see with most of the world's currencies today, that value is sufficient to maintain a money. The hurdle for any money, including gold (historically) and bitcoin today, is achieving that value as a medium of exchange.

      But a money doesn't need to be valued for its use in two or more purpose (as is gold). A medium of exchange is a valued enough propose.

  • Luke M

    If central bankers say their money is backed (you provide quotes), there are two cogent ways of disagreeing:
    1) You're lying.
    2) You may intend for it to be backed, but your bank doesn't have the assets to make good.

    They are not merely stating an opinion. They run the bank. If they say it's backed, then it is.

    • George Selgin

      Luke, first of all, if you are saying that bankers (central bankers included) can't possibly make false claims about their financial condition, you need to bone up on your banking history–including recent experience! Second, the issue isn't whether there are assets to go with a bank's liabilities. It is whether the value of those assets determines the value of outstanding "liabilities" even when the liabilities are inconvertible and free-floating money.

      • Luke M

        I'm only arguing that central bank money is backed. Not that backing is absolutely required.

        If unbacked, then the value of money is at best metastable (some people use the term bubble, but under some definitions it's not a bubble). There is nothing preventing the value going to zero and staying there forever. Even if nothing bad happens!

        If backed, then the value won't go to zero unless the bank goes bust, or fails to perform its appointed duties. Something bad has to happen.

        • George Selgin

          I respectfully disagree, for reasons that I've tried to explain in this post. The assets of a central bank play no fundamental role in determining the value of its currency so long as the holders of that currency are not entitled to them! Although you assert the contrary, you offer no reason at all as to why people should care about assets to which they can under no circumstances have a stake. Every hyperinflated central bank money was "backed" by assets purchased by that bank in the process of expanding its paper and other liabilities. So much for such "backing" keeping a floor on an irredeemable currency's value!

          • Mike Sproul

            Once again you ignore the fact that those hyperinflating banks did NOT back their new money with new assets. The central bank would print a million lira and, rather than buying assets worth a million lira, they would spend them fighting on the losing side of a war.

          • George Selgin

            So, Mike: if the central bank paid the government's military (or reparation or whatever) expenses directly, there would be hyperinflation. But if instead it first traded new money for government securities, so as to allow the government to pay the same expenses itself, there would be no inflation at all, even though total spending, money creation, and real resources were the same in each case?

            I suspect your answer will be "yes." I also suspect that other readers will think, "no!"

          • Luke M

            Without assets how would the bank decrease the quantity of money (or earn interest which it would pay out)?

            Without that power the bank wouldn't be able to prevent spontaneous hyperinflation. You would have to trust that the government could maintain the value with regulations, e.g. laws forbidding transacting in a foreign currency. Which is a pretty shaky foundation of value I would say.

          • Mike Sproul

            " if the central bank paid the government's military (or reparation or whatever) expenses directly, there would be hyperinflation. But if instead it first traded new money for government securities, so as to allow the government to pay the same expenses itself, there would be no inflation at all, even though total spending, money creation, and real resources were the same in each case?"

            If the central bank pays the military expenses directly, and if this causes the quantity of the central bank's money to signifiicantly exceed the central bank's assets, and if there is no prospect that the central bank will get a bailout in the form of new assets, then we get hyperinflation.

            If the central bank instead bought government bonds, and if the government had sufficient assets to cover these bonds, then there would be no inflation.

            If the central bank bought government bonds, and the government could not cover those bonds, then we get hyperinflation.

        • George Selgin

          Mike writes, "If the central bank bought government bonds, and the government could not cover those bonds, then we get hyperinflation."

          You miss a crucial possibility open to governments and their central banks, but not to the rest of us mortals. Consider, first, a Ponzi scheme: debts paid off by further borrowing, in an upward spiral. Impossible to maintain.

          Now consider: with the help of a fiat money issuing central bank, a government can in fact keep paying the principle and interest on its debt by borrowing (and monetizing) more, and more, and more, at an ever-accelerating rate. And central banks have done precisely this in hyperinflations, monetizing debt to pay the interest on debt already monetized ad infinitum–or close enough. You get hyperinflation this way, without the government actually defaulting, just as surely as if the central bank buys oodles of goodies directly.

          Your suggestion that hyperinflation can only set in once the government defaults is simply untrue. Hyperinflation is not caused by default: it is a way for governments to avoid default by using money creation instead to reduce their real indebtedness to zero.

          These are notorious truths, denial of which is extremely irresponsible. The Central Bank of Zimbabwe, for example, was acquiring government debt throughout its hyperinflation. Ditto the Reichsbank in 1923. You can assert the contrary, but anyone who looks up the facts (as in this report on Zimbabwe's hyperinflation) will know what to make of such assertions.

          • Mike Sproul


            I didn't suggest that hyperinflation only happens when governments default. It can also happen when the central bank issues new money without acquiring new assets of adequate value. Or if a government simply prints new money to pay its bills, then of course the quantity of money will outrun the government's assets and inflation results. If the central bank of Zimbabwe is printing billions in cash, and overpaying for the bonds it buys, then once again money outruns backing and we get inflation.

            Sargent's "The Ends of Four Big Inflations" examined the European hyperinflations, and concluded that hyperinflations ended as a result of fiscal reforms, not monetary reforms. He concluded, as I do, that the backing theory gives a better explanation of these episodes than the quantity theory does.

            You never replied to my point about the paper pound during the suspension of 1797-1821. You claimed that the pound was credit money, even though gold convertibility was suspended for 24 years. At the same time you claimed that the dollar is fiat money, on the grounds that gold convertibility has been suspended for 80+ years. How long does gold convertibility have to be suspended for you to decide that a "credit money" becomes a "fiat money"?

  • Paul Marks

    Central Bankers claim their money is "backed"?

    What astonishing nonsense.

    For example the Pound has not been defined weight of gold (or any other commodity) since 1931. Even before this date there were far more Pounds than there was gold in the Bank of England (at the claimed exchange rate) the "Gold Standard" of 1925 to 1931 was just a wild fraud (the depression did not destroy it – on the contrary it just revealed what was already the truth).

    And the Dollar?

    American citizens were robbed of their gold in 1933, again there were far more Dollars than there was gold in the Federal Reserve system – at the declared exchange rate.

    Rather than admit the exchange rate was a lie (and divide the amount of gold the Fed actually held by the number of Dollars – in order to get the real exchange rate) the government just torn up the Constitution of the United States – a usurpation ratified (five votes to four) by the Supreme Court judgement of 1935.

    As Ben Franklin once said – the Constitutional Convention had given the people a "Republic – if you can keep it", as the people re elected Mr Franklin Roosevelt in 1936 (by a 60% versus 40% vote) it is clear that Americans do not want a Republic – do not want a Constitutionally limited government.

    Most people want a government that will do everything for them (give them everything they desire) – and they do not care that such a vast and unlimited government will enslave them.

    As for "backing" of fiat money.

    As the word "fiat" ("command" "order") indicates – there is no "backing".

    One might as well say that the Social Security system is sound because there is a "trust fund" with a lot a government "I.O.Us" in it.

    I had better stop here as I can feel the "red mist" coming down, and going into a rage does not do the right nodal blockage in my heart much good.

  • vikingvista

    What is the standard textbook teaching in college economics courses of why people value money, and in particular fiat money?

    • George Selgin

      Ah, vikingvista: I am afraid that I haven't read a textbook, on money or anything else, for some decades, precisely so that I would not have to know the answer to your question!

      • vikingvista

        It is just that money is less mysterious than than all the different ways people, including economists, are confused about it. I thought maybe there was a problem in this regard with a standard economics education.

  • dufus

    First of all well done for the article and debate.

    A few months ago I was following an interview to Medvedev (Russia's PM).
    At some point he argued; currently the Dollar is the world's reserve currency, but the world should have 5 or 7 Reserve Currencies to stabilize the global economy.

    – How is such a reserve currency established? Is it only by the fact of millions of people around the world willing to own that currency?

    – What effects would more reserve currencies bring about?



    • George Selgin

      Pierre, the answer to the first question is that this is indeed a matter of voluntary choices, though central bankers rather than ordinary persons obviously are making the choices.

      As for multiple reserve currencies, nothing presently prevents either central banks or private traders from diversifying. However, the overwhelming dominance of the dollar is to a significant degree self-reinforcing. We have, as usual with money, important network effects driving matters.

  • George Selgin

    Listen, folks: it is of course true, in a trivial sense, that a technically solvent (that is, positive net worth) central banks' liabilities are "backed" by its assets. But it is not true in the sense necessary to salvage the backing theory. Suppose, for simplicity's sake, that a central bank issues fiat "dollars" (forget about associating these with the U.S. or any particular country) only by exchanging them for some real and imperishable asset–say, gold, where the dollar price of gold is free to change–remember, this is an inconvertible currency, so there's no question of any commitment to redeem it in some real asset at a fixed rate of exchange.

    To make the argument as simple as possible, suppose that gold is the most important good in the economy in question, and that we treat its dollar price as a reasonable proxy of the general level of prices. (The assumption isn't essential: after all, gold producers will use the proceeds from gold sales to buy whatever other goods the economy makes, causing those goods' prices to move as gold demand increases, but more or less depending on relative elasticities of supply and demand.) Let P therefore stand for the dollar price of gold.

    Its trivially true that our central bank's dollars are always fully "backed" by gold, since it never issues any except by acquiring that commodity. Moreover, given our assumption if P changes for any reason the (dollar) value of the central bank's gold holdings changes proportionately, so changes in P don't themselves pose any threat to the central banks' ability to maintain full "backing" of its notes.

    But the very fact that the bank's assets in this case are bound to be worth as much as its liabilities proves why the "backing" theory, understood as a theory of how the purchasing power of a fiat money is determined, is entirely vacuous. For there's no limit to how much gold the central bank can acquire–it might try to buy the entire stock, and keep bidding up P so long as it has paper and ink at hand, and yet never reach a point at which its dollars are any less completely "backed" than before–that is, completely backed by gold possessing an equivalent dollar or "nominal" value. The fact that gold itself is scarce in real terms has absolutely no bearing on the central bank's capacity to create dollars and raise dollar-denominated prices. None.

    So the point isn't that fiat notes can't be said to be fully "backed" by assets. It is that their "backing" is irrelevant to the determination of the notes' purchasing power.

    My example involves a real commodity as "backing," but without convertibility. Allowing the "assets" purchased by the central bank to consist, not of such a commodity, but of financial assets denominated in its own currency unit, only makes the "backing" theory appear that much more irresponsible. In this case, dollars are supposedly "backed" by assets that are promises to pay…dollars! Dollar promises backed by promises to pay dollars = circularity = indeterminacy. The bonds are valuable so long as the dollars are valuable, and the dollars are valuable so long as the bonds are valuable. But both the bonds and the dollars can be worth nothing–and that is what happens in every modern episode of hyperinflation. Mikes suggestion to the contrary–that such hyperinflation can only happen because a central bank starts issuing money without acquiring valuable assets, e.g., in the equivalent of helicopter drops, isn't consistent with the reality of many hyperinflations, in which the central banks in question were monetizing their governments' securities. As I said before, the effects are in fact the same as they would be if the central bank just paid the governments (and its own) bills by printing more notes.

    I am, of course, beating a dead horse But this is monetary economics, and monetary economics is haunted by zombie horses.

    • vikingvista

      Is it not a tautology (or rather a mere redefinition of "backed") to say that as long as money is used (i.e. it trades for other assets) that it is therefore "backed"? Is there any even theoretical way to falsify such a claim? For a backing theory to be at all informative, does it not have to offer a theoretical example of non backed circulating money, or at least introduce a causal mechanism that contradicts other theories?

      By the way he redefines "backed", and the pure accounting rather than causal nature of his theory, it seems Mike's claim "money is valued to the extent it is 'backed'" reduces to "money is valued to the extent it is valued".

      • MichaelM

        I figured out a while ago that 'backing theory' is Mike's attempt to make double entry accounting into an economic theory.

        • Mike Sproul

          Actually, it's my discovery that the same accounting principles that govern the value of stocks, bonds, etc. also govern the value of paper money. There is no need for a "special" theory of money.

      • Mike Sproul

        It's tautological to say that 100 paper dollars are backed by 100 dollar's worth of assets. But that's not what the backing theory says. It says that if 100 paper dollars are backed by 100 oz of silver, then $1=1 oz.

        Take it a step further. Suppose $100 is backed by 10 oz. of silver, plus $90 worth of bonds (denominated in $). It remains true that $1=1 oz. If every dollar refluxed to the bank, then the first $90 could be redeemed for the $90 of bonds, and the last $10 could be redeemed for the 10 oz. The value of the dollar stays at 1 oz throughout the process.

        • vikingvista

          "It's tautological to say that 100 paper dollars are backed by 100 dollar's worth of assets. But that's not what the backing theory says. It says that if 100 paper dollars are backed by 100 oz of silver, then $1=1 oz."

          You've just said the same thing, assuming "100 paper dollars" are worth 100 dollars. Unless $100 doesn't equal 100 paper dollars, your backing theory appears to be a tautology. And if they aren't equal, then how exactly does a dollar differ from a paper dollar?

          • Mike Sproul

            I'm saying that (1 paper dollar)=(1 checking account dollar)=(1 oz of silver). That's not a tautology. All of the tautologies are to be found on the quantity theorist's side of this debate (MV=Py, Y=C+I+G, etc.)

          • vikingvista

            The tautology isn't the statement that $1=1oz. The tautology is the claim that because the 1oz=$1, therefore $1=1oz. Since that is the price at which the $ issuer acquires the oz's, or the oz-holder acquires the $s, how can it be otherwise? Or do you imagine a realizable test where the price might be different (without changing the relative amount of issued $ to issuer-held oz)?

          • vikingvista

            "All of the tautologies are to be found on the quantity theorist's side of this debate (MV=Py, Y=C+I+G, etc.)"

            Presumably, like every other theory, even your theory can make use of identities in its explanation. Quantity theorists know that the exchange equation is an identity. And the other income identity is merely expressing a preferred partition for further modeling. That isn't the issue. The issue is, does your fundamental claim actually say anything. Perhaps it does. Can you envision a test whereby it might be disproven? Does your theory claim casual mechanisms different than other theories (e.g. does it contradict the claim the inflation is caused by new money holders bidding up prices)?

          • Mike Sproul


            "Can you envision a test whereby it might be disproven?"

            How about the old Bank of Amsterdam, which maintained coin convertibility of its credit money, but people never actually converted for centuries. The Bank only issued new money in exchange for new coins deposited, but after a few centuries it lost assets, and its money lost value as a result, even without an expansion of its money-issue.

          • vikingvista

            "its money lost value as a result, even without an expansion of its money-issue."

            Are you saying that had that money not lost value, such an example would have contradicted your theory?

            Also, do you believe the market was aware of those lost assets?

          • Mike Sproul

            Yes, if the B of Amsterdam guilders had held their value even after the discovery of the lost assets, that would have refuted the backing theory. The fact that they did lose value (after the loss was discovered) supports the backing theory.

          • vikingvista

            Do you believe the market was aware of those lost assets?

          • vikingvista

            Well, it seems the quickest way to terminate a thread with you is to ask you if you think the market is aware of a money-issuer's assets. Perhaps it is because you're anticipating the follow-up, but since after multiple tries in different threads you've not allowed me to lead you to it, I'll take the initiative myself before I forget why I was ever even interested.

            For changes in issuer's assets to affect prices denoted in its money, then either the market is aware of the issuer's assets, or it is not. If it is not, then it seems you lack a causal mechanism for how changes in assets could realistically (i.e. excluding the supernatural) affect prices.

            If, on the other hand, your theory requires market knowledge of the issuer's assets in order for its asset value changes to affect prices, then you face another problem. This would mean that if an issuer simply keeps knowledge of its assets away from the market, it can print as much fiat money as it wishes, without affecting prices in the least. That an economy could literally be buried 5 feet deep in paper money without affecting prices, is hard to believe.

            As a way out of this dichotomy, you might argue that any issuer's attempt to mask or deceive the market about its assets would cause the "backing" and value of the money to collapse. But then you are left explaining how the market could either know it is being deceived, or make such a dramatic turn upon hearing news of lack of knowledge of assets. This would seem to leave any issuer, and particularly monopoly issuers, highly unstable in the presence of rumors.

            Notice, the quantity theory explanation of inflation doesn't suffer any of this. It requires only that extra money results in extra spending. Neither real nor perceived knowledge of issuer's assets makes any difference, aside from how it result in individuals having less or more money to spend.

          • Mike Sproul

            Vikingvista (3/11)

            My answer about awareness of assets would be the same for any bonds or stocks. People have varying degrees of awareness of an institution's assets and liabilities, and they value the institution's liabilities accordingly. People are reasonably aware of the fed's assets and liabilities. They are also aware of the possibility that the Treasury might some day bail out the fed, or loot the fed. But if you doubt the existence of any mechanism, then you'd be just as doubtful about the mechanism for valuing bonds and stocks.

          • vikingvista

            "My answer about awareness of assets would be the same for any bonds or stocks."

            I know that is your answer. That's why last time you replied thus, I asked (and did now receive) what you thought *that* mechanism was. For all I know, your notion of both is superstitious (or more likely, tautological). You have yet to articulate a mechanism to me. Can you think of one?

            "People have varying degrees of awareness of an institution's assets and liabilities, and they value the institution's liabilities accordingly."

            Okay, so your "backing" theory really says not that the market's valuation of an issuer's money equals the quantity of the assets of the issuer, but rather the market's valuation of an issuer's money equals what the market *suspects* is the quantity of the issuer's assets.

            So, given that people and markets are not omniscient, do you not suspect then that the market value of a currency will at times NOT be related to the issuer's assets?

        • Mike Sproul


          The mechanism that assures that stocks and bonds are priced according to their backing is arbitrage. Any over/underpriced stock or bond creates an arbitrage opportunity, and as arbitragers pounce, the price is driven back into line. The same is true of money. If the dollar is worth more than its backing, then arbitragers short the dollar (Think of the 1997 run on the Thai central bank.) If the dollar is worth less than its backing, then arbitragers go long and drive the dollar up.

          • vikingvista

            "The mechanism that assures that stocks and bonds are priced according to their backing is arbitrage."

            No. Arbitrage is what levels prices. Arbitrage doesn't determine what prices the arbitrageurs target (to reason so is circular). My question to you, is what determines the prices that arbitrageurs observe as profit opportunities?

            "If the dollar is worth more than its backing, then arbitragers short the dollar"

            Unless your backing theory reduces to a tautology, this isn't true. Arbitrageurs exploit existing price differences FROM WHATEVER CAUSE. If you want to claim that the difference between, say, a currency exchange rate and the ratio of goods & services priced in those differing currencies is the result of one money quantity exceeding its backing, then you have to explain how backing can matter to such a ratio. The ratio is the same whether one money exceeds its "backing", both exceed their "backing", or both rapidly decline relative to prior "backing". Arbitrageurs respond equally to the ratio regardless of any issuer's relative quantities of assets and money.

            I would like to know what you think would happen to the USD if the Federal
            Reserve chose tomorrow to permanently to keep all its sales and purchases, as well as its current holdings and net USD production, completely secret. Your theory depends upon the market having a valuation of the established fiat issuer's assets. The quantity theory requires no such knowledge or guesswork.

    • Mike Sproul


      It is indeed possible to back dollars with promises to pay dollars, provided there is some amount of real assets anchoring the whole thing.

      Stock market analogy: GM shares are worth $60 each. GM then issues 10 new shares for $600, and uses the $600 to buy call options on GM. GM stock would then be partially backed by call options that are payable in…GM shares. It works, though of course the potential volatility of GM shares would be increased. If, for example, GM lost some assets, then GM shares would fall. This would make the calls less valuable, and since those calls are GM's assets, that would make the shares fall still further, etc.

      Same with the dollar. The Fed has gold, buildings, foreign currencies, etc, which anchor the dollar's value. The fed then issues another $100 of FRN's in exchange for 100 DOLLARS WORTH of bonds, effectively backing the dollar with bonds payable in dollars. It works just like the GM stock example.

      By the way: The old Bank of Amsterdam held 100% coin backing for the credit "guilders" that it issued. Credit guilders were legally redeemable in coin, but the coins in fact sat in the vault for centuries without ever being paid out. When the bank eventually lost its assets, its guilders lost value too, just as the backing theory implies. What matters to customers is that the coins are actually there in the bank. Convertibility can be delayed for centuries, as long as backing remains.

    • Mike Sproul

      "For there's no limit to how much gold the central bank can acquire–it might try to buy the entire stock, and keep bidding up P so long as it has paper and ink at hand"

      Yeesh! Echoing Lloyd Mints is not going to do the trick. You, like Mints, assume your own conclusion. The whole point of the backing theory is that if money increases at the same rate that backing increases, the price level will not be bid up. Mint's "Self-perpetuating cycle of more loans, more money, and higher prices" never gets off the ground.

  • Paul Marks

    The "assets" of Central Banks are mostly I.O.Us. – I repeat for people to take them seriously is as absurd as taking the Social Security "trust fund" seriously.

    As for Bit "coin" – it is a complicated mathematical "box".

    With nothing inside the box.

  • Martin Brock

    The definition of "fiat money" omits the definitive characteristic of fiat money accounting for its value. Fiat money is valuable because a state commands its subjects to collect the money to pay taxes and other debts imposed by the state. Bitcoin is not fiat money for this reason. Backing or intrinsic value has nothing to do with it.

    That fiat money is valued for this reason does not imply that the supply of fiat money relative to the demand for money is unrelated to the value of the money (the price of goods denominated in the money). The state theory of money does not assert that increasing the supply of fiat money relative to demand for money has no effect on prices. It rather asserts that inflation is one of many possible state policies. That a state hyperinflates its currency does not contradict this assertion. States policies may be foolish, and hyperinflation may begin as a deliberate policy.

    Fiat money is not the only possible money. That Somalians continued to use Somali shillings as money after the collapse of the issuing state and the looting of its central bank does not contradict the state theory of money, because Somali shillings were not fiat money following this collapse, i.e. they were not money because a state decreed them money. They were money because Somalis had grown accustomed to using them as such, and their supply relative to the demand for indirect exchange affected their value as it had before.

    The backing theory does not apply to fiat money at all, seems to me, and it does assert that the supply of money relative to demand is unrelated to the value of money. It rather asserts that the value of backed money is determined by its backing and that the supply of backed money is elastic. If demand for money increases, the supply of backed money expands to satisfy the demand without changing the value of the notes, i.e. if people need more money, they circulate more notes as money, or they create new notes to satisfy the demand.

    I would not say that Bitcoin cannot be money because it is not backed. I would rather say that it cannot be money because its value cannot be sufficiently stable, so it cannot be a unit of account and will never be widely accepted. Bitcoin has value in black market transactions, including shadowy international remittances, but this value doesn't make it money any more than the value of gold makes gold money. [Shadowy remittances may be the work of the angels, but this virtue doesn't make Bitcoins money.]

    • vikingvista

      "Fiat money is valuable because a state commands its subjects"

      Is it inconceivable that the common medium of exchange function itself is valued enough? Must there necessarily be some other reason?

  • elriel

    I think you could say that the main purpose of a central bank is to try to keep the price mechanism in quantity theory from affecting the value of it's currency. That's why they hold assets in the first place. This is the purpose of making noise about backing their currency, to create a trustworthy bottom for the value.

    The problem lately, though, is that people trust the currency too much. So, now they are scrambling to also create a ceiling for the value. Let's hope they can manage that without upsetting the bottom too much.

    Anyway, the point I wanted to make here is that fiat currency is not free from effects of quantity theory. It only looks like that when the central bank is successful in it's struggle to keep it that way. Eventually it's bound to slip out of their control.

    • vikingvista

      "Eventually it's bound to slip out of their control."

      People value FRNS now. If there is going to be a change the creates a problem with FRNs, it seems more likely to result from being *in* the central bank's control, not out of it. To keep their dollars valuable, all they have to do is nothing.

  • Matt Young

    I am getting brave here!

    "Listen, folks: it is of course true, in a trivial sense, that a technically solvent (that is, positive net worth) central banks' liabilities are "backed" by its assets."

    I think the currency banker has to have a slight negative net worth, measured in its own money. The currency banker pays for its own measurement error, price variance due to money jitter. The utility of its money may be greater than its own induced error, but in a strict actuarial sense the money variance exists and has to be accounted for. If not then the banking network would shrink over time.

    • Matt Young

      I didn't want to get verbose, let add a detail.
      The liabilities of any banker in the chain projected two periods out, minus its assets projected two periods out must be equal to one unit of money jitter which the banker introduces. The two periods is necessary so that the money users can have two sample periods to cover the jitter, the banker allows the user to avoid the jitter. This is the no arbitrage condition. The banking system works because the banker guarantees the no arbtrage, either via competition or by published spreadsheet. The banker then makes money by consistently outperforming the natural money jitter, finding more accurate borrowers and more of them.

      • Matt Young

        Wait, you say. I am dumb, I have assets and liabilities switched. Or I have debt and credit flowing the wrong directions, and so on. You are likely correct. But a switching variables does not help, the equations of 'state', as someone mentioned, or the equations of flow remain constrained to the no arbitrage situation. Banks can still continually make money when the banking network optimally covers the economy; and the banking network is more accurate then the economy.

  • Lawrence Kramer

    The "backing theory" reminds me of a conversation with a two-year-old:

    Billy, I believe this man is not a stranger.


    Because the dog didn't bark at him.


    Because I believe the dog barks at strangers.


    Because I believe he was trained to do that.


    Because I trust his trainer,


    Because I expect the trainer to pay me if the dog isn't reliable.


    Because he promised to pay me and I trust his promise.

    Why? Because he posted collateral in gold.

    Oh, I see.



    Billy, I believe this man is not a stranger.


    Because the dog didn't bark at him.


    Because I believe that the dog barks at strangers.


    Because he has always barked at strangers. So shut up and sit down.


    The first stranger is "backed" by the the silence of the dog, which is "backed" by the claim of the trainer, which is "backed" by the guaranty of the trainer, which is "backed" by the gold in the owner's hands. The second stranger is also "backed" by the dog's silence, but that silence is backed by the dog's prior performance, as in "Money is used because users have observed that it can be used." Money is as money does. As I said in an earlier comment, backing is only a metaphor for the source of our confidence.

  • Paul Marks

    I repeat there is very little (if any) "backing".

    What Central Banks mostly have is piles of I.O.Us.

    Even if Central Banks had all the gold and so on they claim to have (and they do not – it is not really all in the New York Federal Reserve, it just does not add up) it would be very little "backing".

    The rest is just government and other I.O.U.s

  • tymoignee

    Hello, Eric Tymoigne here. I just stumbled on your blog and I see I am kind of late in the game here but what the heck…
    There is a lot here so let me just pick on of your comments: "The assets possessed by a fiat-money-issuing central bank–that is, by a bank that issues inconvertible paper, with no promise to ever redeem it in a specific amount of some other money–play no part in determining its currency's purchasing power."

    There are two issues here:
    1- There is a redeeming clause even for inconvertible modern monetary instruments: the issuer will take them back at face value in payments due to him/her at any time. That's what the issuer owes the bearers. Put in terms of present value we have a fair price that is P = sum discounted nominal cash flows = sum of discounted nominal incomes + discounted face value. For a monetary instrument the term to maturity is zero and incomes promised are zero, so P = face value. Now, in the past, governments did issue inconvertible monetary instruments and never promised to take them back under any circumstance, then in this case P = 0. But today that is not the case, banks accept bank money at anytime (zero term to maturity) to service debt owed to them, the government accepts government money at any time in payments owed to the government. And both banks and government accept their money at face value.
    2- There are two things that affect negatively the purchasing power of a monetary instrument: a- if it trades at a discount in nominal terms(a $20 bill circulates at $15, which used to be quite common in the past) b- if there is inflation. The fair price only deals with the first one, it is not concerned with inflation, it is about the nominal value of monetary instruments. Everybody seems to be concerned with reason b without paying any attention to reason a. That's where bitcoins failed too.

    • George Selgin

      Hi Eric, and thanks for chiming in.

      Your claim that a fiat-money issuers' willingness to receive its own liabilities in payments to itself amounts to a sort of redemption obligation seems to me quite mistaken. The normal notion of redeemability, and the only one that matter's for making the "backing" of an issuer's assets a crucial determinant of its money's purchasing power, is redeemability into some outside asset (that is, one the issuer itself cannot create) at a fixed rate. A CB that offers to accept its own IOUs in payments doesn't make any such commitment. Consequently the "offer" places no effective constraint on how much money it may create, and supplies no reason for anyone to care about the value of its assets or to value its liabilities according to the value of their asset "backing." In short, that Janet Yellen accepts FR dollars in payment of her salary plays no greater role in determining those dollars' purchasing power than the fact that Walmart accepts them in exchange for its products.

      The situation is quite different from that of a commercial bank offering to redeem its deposits into Federal Reserve Notes, or a central bank that offers, say, to convert its IOUs into gold in unlimited amounts at a fixed rate. There you have a meaningful commitment that restrains the issuing institution's asset choices, and makes those choices a determinant of the public's confidence in its convertibility offer.

      • Lawrence Kramer

        "Consequently the "offer" places no effective constraint on how much money it may create…."

        What is the test – that the currency be "backed," or that there be an "effective constraint on how much money [its issuer] can create"? Seems to me you're admitting that "backing" is merely a special case of "effective constraint." Good for you!

        • George Selgin

          If I gave that appearance, Lawrence, I certainly didn't intend to: I meant to imply that the offer to receive its own money in payments doesn't constrain an issuer's portfolio choice, does not give the public any reason to worry what choice it has made in that regard, and, finally, does not give the public any reason to adjust its valuation of the worth of the money in question according to the nature of the portfolio actually held.

          I will say it again: the fact that Janet Yellen is willing to receive dollars in payment of her salary does not make the Fed's asset "backing" any more relevant a determinant of the dollars' purchasing power than the fact that Walmart accepts dollars makes Walmart's assets a determinant of the dollar's purchasing power. All who accept a currency in payment increase the demand for it and to that extent contribute to its value. But that's all. "Backing" has nothing to do with this.

          • Lawrence Kramer

            "If I gave that appearance, Lawrence, I certainly didn't intend to…"

            Why not? You chose "effective constraint" as your test of the currency. I agree that "backing" is not necessary, and I have not put forth the MMT "you can pay your taxes with it" argument. (Why do you trivialize the argument by reference to Dr. Y's salary, when you know the MMT position relates to taxes?) But backing has always been an effective constraint on issuance, and I don't see why you would reject the idea that it is just a special case of the phenomenon to which you turned when trying to put down the MMT view. (BTW, Wal-Mart pays taxes, too, and so do its vendors. I agree that public acceptance is what matters, but it's not turtles all the way down; maybe the fact that dollars can keep be redeemed for the blessings of liberty is the bottom turtle…

          • George Selgin

            "Why do you trivialize the argument by reference to Dr. Y's salary, when you know the MMT position relates to taxes?"

            My remarks were in direct response to Eric Tymoignee's comment, in which he writes: "The issuer will take them back at face value in payments due to him/her at any time. That's what the issuer owes the bearers." "The issuer" in this context is the Fed, not the U.S. Treasury. The reference to Yellen reflects nothing more than my general preference for the concrete over the abstract: it stands for the sum of all the payments made to the Fed.

            And I still see no argument in what you say for claiming that the assets "backing" Fed dollars play any crucial part in determining their purchasing power. As for tax receivability itself doing so, I have addressed the point above. That Walmart has to pay taxes in dollars, is not uniquely responsible for the fact that Walmart accepts dollars: Walmart, like everyone else, is influenced by dollars' general acceptability. Some firms that do business abroad owe U.S. taxes for some of their sales, payable in dollars. Yet that doesn't determine what money they accept in payment overseas.

            Again, the real demand for X as an exchange mediumcontributes to the value of X. No one denies this. What I deny is that either the assets backing X, where X is not a convertible fixed-rate claim to anything else, or the receivability of X in payment of taxes, serves as a distinct or indispensable determinant of X's purchasing power.

      • tymoignee

        In my opinion, viewing redemption and convertibility has similar is too narrow because it does not capture other important aspects of monetary mechanics. If tomorrow the Fed and Treasury declared that they would only accept Fed currency (FRNs or reserve balances) at half their face value in payments due to them, I can guarantee that a 50% discount will be applied very quickly in the rest of the economy (one would go to a shop, hands a $5 bill and that would cover only $2.5 worth of purchase). Kings used to do this kind of thing quite a bit when they cried down or up the coinage, and they did it so much that it created a mess. Recently when some European states declared they would not accept their currency in payments and instead move to the euro, the fair value of their currency failed to zero (I have a friend who learned this lesson dearly when he found 10,000 French franc in cash that his now-diseased wife had previously hoarded without telling him…). Also be mindful that I am here focusing only on the nominal value, i.e. one can assume no inflation for the sake of argument. To my knowledge, the first person who brought up that aspect of redemption very clearly is Thomas Smith in 1832, then came Macleod, Knapp and Innes in 1913, all of whom focused on the financial mechanics at play in a monetary system (the "credit view").
        Convertibility is not a constraint on banks’ ability to create their currency as long as there is an elastic supply of what they promise on demand. With an elastic supply of inconvertible gov. currency, banks can always get all the gov. currency they want. What puts a constraint on monetary creation by banks is their underwriting criteria, the willingness of the non-bank sectors to get bank credit to finance economic activity (banks can't create monetary instruments in a meaningful way if there are no customers who want bank credits), and capital regulation (at least given the state of financial arbitrage). The quantity of reserves is irrelevant.
        For government, convertibility is a constraint on monetary creation indeed but it is not the only one and not a necessary one. One can do without: budgetary procedures and political process determine the size of government in the economy. Ability to spend an unlimited amount does not mean it ought to happen or that it will happen (just look at the US today which has one of the smallest government size among developed economies).
        On a final note, I (and nobody in the MMT camp) am not of the opinion that only gov. can issue monetary instruments. In fact anybody can do so (just issue a zero-coupon, zero-term to maturity bond), the problem is to find someone willing to accept these monetary instruments. A sufficient condition is to have debtors and to tell them they can pay you with these monetary instruments. State does it by imposing a tax liability (starting with Mass. Bay colonies in the U.S. for which it worked well for a while: notes circulated at par for a while after a period of distrust), banks do it by imposing debt when issuing monetary instruments. I can issue E.T. notes it in my class, via in-class pretend exercises; by imposing a significant debt on my students (say, half of their grade involves paying a tax that must be paid in E.T. notes), they very quickly accept E.T. notes at par for any kind of things I want to purchase from them (computer, labor, etc.). They won’t accept it if I don’t impose a tax. Again the relevance of the fair value shows.

        • Mike Sproul

          A key element of the backing theory is that there are many kinds of convertibility. Money-issuers might redeem their money for gold, for bonds, for loans or taxes owed to them, for land, used furniture, etc. A bank can suspend gold convertibility for 24 days, 24 years, or 80+ years, and as long as it maintains the other forms of convertibility, the suspension of gold convertibility can be irrelevant. Even George is willing to admit that the British pound was still a "credit money" during its 24 year suspension period, but somehow he thinks that the 80+ year suspension of the dollar causes the dollar to be "fiat money".

          It's clearly wrong to claim that "inconvertible"="unbacked", especially when the backing assets are right there on the issuer's balance sheet. As long as there are open channels through which money can reflux to its issuer, it does not matter whether the reflux occurs in exchange for a physical amount of gold, or for a "dollar's worth" of bonds.

          (A short and simple illustration of the Law of reflux is provided below)

          • tymoignee

            "It's clearly wrong to claim that 'inconvertible"="unbacked'": indeed I agree. Creditworthiness of the issuer may be the only thing that backs a financial claim.
            Yep reflux mechanism are central.

          • George Selgin

            I never claimed that "inconvertible=unbacked." I'm aware as anyone that the Fed's balance sheet includes assets as well as liabilities! My claim has always been that backing plays no crucial part in determining the value of an inconvertible currency. What's more I've been perfectly clear about it.

          • Mike Sproul


            Then can you explain your contradictory statements that the pound was "credit money" during its 24-year suspension, while the dollar (suspended 82 years and counting) is fiat money? The balance sheet of both of those banks looked quite similar, with assets as well as liabilities.

            And if, as you say, backing plays no part in determining the value of an inconvertible currency, then what is the difference between your position and the "inconvertible=unbacked" statement that you now disown?

          • George Selgin

            Difference hinges on expectation of future redemption; see post.

          • Mike Sproul

            I've seen the post. "Expectation of future redemption"? Both the BOE and the fed had/have assets with which to redeem in the future, so there is some non-zero probability of redemption in both cases. How low does the probability of gold redemption have to fall before you decide to stop calling something "credit money" and start calling it "fiat money"?

            And, of course, redemption might be in gold, bonds, loan or tax payments, etc. There is nothing special about gold convertibility, but that seems to be the only factor that you think relevant when using the words "credit" or "fiat".

          • George Selgin

            Mike, I have before me a slip of paper bearing my notarized signature, on which I have written as follows:

            "It is possible that I, George Selgin, will pay to Mike Sproul (or whosoever is the bearer of this note), some amount of something some day–I'm not saying how much or what or when."

            I am prepared to give you this instrument, together with a copy of my personal balance sheet, indicating my assets, liabilities, and net worth (the last of which will show me to be happily and considerably in the black), also bearing my signature and certified by a licensed and reputable accountant. I am prepared as well, should you so desire, to make a binding commitment to not issue any like instruments to any other person, or to otherwise encumber my possessions. You will, in short, be the only person holding any sort of new claim against me.

            I trust that you will admit that the possibility to which my note refers is no less real than the one to which you appeal in claiming that the value of an apparently irredeemable fiat money depends not on supply and demand but on the value of the issuers' assets. So, if you will please indicate what positive amount you will send me in exchange for my note, I will happily agree to treat whatever amount you offer as a measure of the worth of the backing theory.

          • Mike Sproul


            If you were a central banker, publicly committed to maintaining the value of that slip of paper, possibly with a 2% inflation rate built in, and if those slips were a first and paramount lien against your assets, then I'd gladly accept those slips of paper at par.

          • George Selgin

            Which means, in effect, that you value the central banks' money, not because of the asset it holds, but because you trust it to honor a commitment to restrict its quantity and thereby maintain its purchasing power, for the "lien" etc. have their precise counterparts in my offer to you.

          • Mike Sproul

            No, it means I trust the central bank to honor a commitment to maintain a certain amount of backing per dollar issued. Purchasing power is maintained, not by a limitation of the quantity of money issued, but by assuring that the issuer's assets move in step with its money.

            Everyone agrees that the backing theory is true of stocks and bonds. Everyone agrees that it is true of money that is both backed and convertible. You even agree that it applies when convertibility is delayed, but expected to resume. Yet somehow you think that an 82-year suspension of gold convertibility changes everything, and suddenly backing counts for nothing, while supply and demand count for everything.

          • elriel

            I'm not sure why this is something to argue about. Backing,__if trusted__, will affect how people value the thing that is backed. If that trust is lost, things will revert to supply and demand. It's merely a question of "do enough people believe in the backing?". It's not really relevant if the backing really exists except for how it affects the trust.

            So, the way I read George's statement "backing plays no crucial part in determining the value of an inconvertible currency" is that very few people trust the backing and therefore the backing theory has very little effect on the value.

          • Chris LeRoux

            Only counterfeit needs backing. Money is a commodity.

        • Mike Sproul


          The trouble with the money supply/money demand theory is the old saying "You can't ride a claim to a horse, but you can trade with a claim to money." For example, as foreign moneys start to be used, or as checking accounts, credit cards, etc are invented, the demand for the base money falls and it loses value, with no stable solution short of zero value. The backing theory does not suffer from this problem. As rival moneys are introduced, base money holds its value because it has just as much backing per dollar as before.

  • Lawrence Kramer

    "'The issuer' in this context is the Fed, not the U.S. Treasury."

    But you are a teacher. You know that the argument ET was making was a version of the MMT claim, and I think you owe it to your readers to read his claim as if "issuer" meant the Platonic "monetary sovereign" and not the particular legal entity to which the function of creating money has been assigned.

    "And I still see no argument in what you say for claiming that the assets "backing" Fed dollars play any crucial part in determining their purchasing power. "

    What assets? What claim? If you have read my earlier comments, you would know that I don't believe backing is important, that it is just an obsolete form of "effective constraint" on issuance. Where have I said that there are assets "backing" Fed dollars, much less that they are crucial?

    "Some firms that do business abroad owe U.S. taxes for some of their sales, payable in dollars. Yet that doesn't determine what money they accept in payment overseas."

    They also owe taxes in their local currency, which they do accept. Surely you can do better than that. And why is "uniquely responsible" the test of relevance. If I am wearing suspenders, my belt is not uniquely responsible for my pants staying up. But if I take off the belt, confidence in my not be depantsed decreases. It all comes down to the test you articulated but now seem to want to disown, or at least control: an effective constraint on issuance. So, I repeat, if I may, why isn't backing a special case of effective restraint? Is it not an effective restraint?

    • George Selgin

      I honestly don't know how to reply further to you without repeating myself. "Backing," we seem to agree, doesn't determine purchasing power. But neither does it constrain quantities at all. I have seen no good argument anywhere here to prove otherwise. An issuer of an irredeemble fiat money may buy assets till the cow comes home, so long as he or she has paper and ink at hand, and by so doing can bid up prices until the stuff becomes worthless. And the progress of the inflation will not depend in any way on what assets the issuer acquires in the process of issuing more units, or on whether the units are issued in exchange for anything at all, as opposed to being dropped from helicopters (or whatever). Public receivability of a currency in tax or other payments likewise poses no barrier to hyperinflation–as example after example demonstrates. Finally, public receivability contributes to the demand for a currency, and thereby to its purchasing power, precisely in the same way that receivability in private payments does so, and so is neither necessary nor sufficient to secure its purchasing power. Were the IRS to decide tomorrow that payments to it should henceforth be made in Euros or Yen, that would cause people to trade dollars for Euros or Yen in the FX market for the sake of paying taxes, just as they must now trade them in that market for the sake of importing goods from Europe or Japan. That would of course imply a reduced demand for dollars, just as an increased demand for cars made in Germany or Japan relative to cars made in the U.S. would mean a reduced demand for dollars. But it would not by itself necessarily trigger wholesale "Euroization" or "Yenization" of the U.S. economy. The Treasury differs from others here in being a big player, to be sure. But this is a difference in degree, not in kind.

  • Lawrence Kramer

    "But neither does it constrain quantities at all. An issuer of an irredeemble fiat money may buy assets till the cow comes home, so long as he or she has paper and ink at hand, and by so doing can bid up prices until the stuff becomes worthless."

    But avoiding worthlessness is a constraint. That's why backing works: the issuer WILL NOT bid up the price of assets to the point of rendering the money worthless, because then the issuer will be out of a job. You seem to believe that a constraint on issuance matters. But is it the constraint, or the public's faith in it, that makes the constraint useful? If the latter, as I believe, then ANY constraint, including a political constraint, including a process that delegates the issuance function to people who will not "buy assets till the cow comes home," is sufficient. By limiting the issuer's asset choices through law, the populace can know by the market price of the backing material – gold or silver, historically – whether the backing mechanism is in fact acting as a constraint on issuance. If the backing mechanism is so acting, then the currency remains in favor. If not, then not. I have no doubt that any sovereign can destroy its currency by over-creation; but in a functioning democracy, where the price of gold is a known datum, a currency backed by gold will not be debased if the populace does not want it to be debased. Thus, backing, with sufficient dissemination of information, is a special case of an effective constraint on issuance. It is, in my view, no longer the best constraint, but I don't believe we can deny that has been at other times and places been the best constraint available.

    There is no ceteris paribus world in which the US abandons the dollar for the payment of taxes. I wouldn't speculate on what would happen in a real world in which that could happen.

    • George Selgin

      "the issuer WILL NOT bid up the price of assets to the point of rendering the money worthless, because then the issuer will be out of a job." If the issuers job depends on protecting a money's purchasing power, that is a constraint on overissue. But what has "backing" to do with it? Why drag "backing" in when the firing is what is doing the work? Honestly I don't see any connection here at all!

      A fiat money issuing central bank can even inflate a great deal solely by acquiring gold; the price of gold will steadily rise as the CB acquires it, just as the price of any asset might. In the limit, the price rises toward infinity; yet every unit remains fully "backed" by it in nominal terms. The price of gold would rise faster than other prices because of the change in relative demand and the fact that the gold supply schedule is rising. But other prices would also tend to rise, and might continue to do so until the issuer is no longer able to buy any more gold even by offering to pay an infinite price for it!

      • Lawrence Kramer

        "Why drag 'backing' in when the firing is what is doing the work?"

        We want to fire the guy before he can destroy the currency. Otherwise, we won't trust the currency to begin with. How are people to know that a currency is not being overissued before it's too late? The price of the mandatory backing is the dog that barks in the night, or doesn't. In a world in which TV has not been invented, not to mention CNBC or the briefcase indicator, the price of the backing is the transmission device for the information that the people need.

        Backing has always been nothing more than a form of information, the auditor's "objective verifiable evidence" that the currency is not being overissued. It's a tool. It's not an essential tool, but it once was a very good tool. Now, not so much.

      • Mike Sproul

        Rather than spinning your wheels on these fanciful "What if the Fed buys everything?" thought experiments, let's just imagine that the Fed issues 10% more dollars, with each new dollar being issued for a dollar's worth of bonds. The backing theory says that the value of the dollar will not change, because the Fed has the same amount of backing per dollar as before. A few observations:
        1. If the public's demand for money rose by 10% at the same time, then the quantity theory agrees that there will be no inflation, not because backing kept pace with money, but because money demand kept pace with money supply.
        2. If the public's demand for money were unchanged, then the unwanted 10% of dollars would pile up in bank vaults. The unwanted dollars would want to reflux to the fed, probably through the bond channel or the loan channel. If the fed allows this reflux (Maybe because it is targeting free reserves), then the extra 10% of money refluxes as fast as it is issued, with no price effects.
        3. But suppose the Fed got no new assets in exchange for the new money. The backing theory says there will be 10% inflation, even if money demand rose 10%. The quantity theory says that as long as money demand rose 10%, there would be no inflation. This is where the quantity theory gets in trouble, because the fed has 10% too few assets to buy back all the money it has issued, and yet somehow the dollar supposedly holds its value. The backing theory, in contrast, implies that when the amount of money outruns the fed’s assets by 10%, the money will lose 10% of its value. At this value, the fed’s assets are still enough to buy back all the money it has issued.

        • George Selgin

          Mike, when confronted with a scenario designed to demonstrate why your theory doesn't work, you typically reply, not by showing why the logic of the example isn't correct, but with an alternative scenario you think more congenial. That's one reason why I find arguing with you quite futile.

          In this particular instance your argument at least has the virtue of making clear that your position is that the supply of and demand for money play no part at all in determining its market value. I will leave it to my readers to decide whether they find it as easy as you do to embrace this view.

          • Mike Sproul


            Supply and demand works fine for commodities–things that are produced using scare resources. Supply and demand does not work for securities–things that can be instantly and costlessly issued and retired in infinite amounts (like money). That's why accountants use balance sheets when valuing securities, not supply and demand curves.

          • George Selgin

            But fiat money and bitcoin aren't securities. See post. (It is ridiculous, by the way, to claim that something (1) is a security, and therefore (2) can be priced according to the usual present-value formula while defending (1) by observing that, to qualify as a security, a financial instrument need not consist of a certain commitment to pay any particular amount of a particular thing, but instead can consist of something that could potentially be redeemed for some unknown amount of some unknown thing at some unknown time! One can either toy with the definition of "security" this way, or appeal to the normal means of pricing a security by taking the present value of future returns. But one can't simultaneously do both–as several persons have tried to do now on this forum–without being guilty of the most blatant sort of casuistry.)

  • Chris LeRoux

    Money is the most voluntarily saleable good. Anything else is counterfeit, fraud, and embezzling. You seem to think legal tender "laws" are not necessary to get people to use the counterfeit, and I suppose you also believe the empire's theft of the people's gold was superfluous to the adoption of counterfeit, and I suppose you also think compulsory tax-funded indoctrination was also superfluous. Very amusing nonsense you have there.