In this paper I argue that the real bills doctrine has been wrongly discredited, and that it ought to displace the quantity theory as the dominant theory of money. The discussion begins with the observation that the issue of backed money will not be inflationary as long as central banks follow the real-bills rule of only issuing money to those customers who offer good security in exchange. I then contend that modern paper currencies, which we normally think of as unbacked fiat money, are actually backed by the assets of the central bank. If correct, this would imply that the real bills doctrine, and not the quantity theory, is a satisfactory model of the value of modern money. Monetary theorists have mistakenly viewed the real bills doctrine as a rule that would make the quantity of money move in step with aggregate output of goods, but the correct view is that the real bills doctrine assures that the money supply moves in step with its backing.
When the Directors of the old Bank of England were accused of having allowed the pound to depreciate between 1797 and 1810, their defense was based on the real bills doctrine. They stated that they had only issued money to those customers who offered good security in exchange for the money. Therefore, they claimed, the Bank had only issued as much money as the legitimate needs of business required. The Bullion Committee appointed by the House of Commons in 1810 denounced this defense as "wholly erroneous in principle" (Gilbart, 1882, p. 53). Sixty-three years later, the bankers' answers were still derided as "almost classical by their nonsense." (Bagehot, 1873, p. 86) It would be difficult to count the number of times that similar debates over the real bills doctrine have flared over the centuries. A few episodes are summarized by Mints (1945, p. 9.):
The real-bills doctrine has been a most persistent one. Given its most elegant statement in all its history by Adam Smith in the Wealth of Nations, it has since served as the defense for the directors of the Bank of England during the period of the Restriction. With some changes it re-appeared as the banking principle; it was the main reliance of the agitators for banking reform in the United States before 1913; it was as comforting to the Federal Reserve Board following the depression of 1921 as it had been a century earlier to the directors of the Bank of England; more recently it has re-emerged as the doctrine of "qualitative" control of bank credit; and, quite aside from these special uses to which it has been put, it has been consistently defended throughout all these years by a large proportion of bankers and economists.
Since Mints' time, a dissident tradition opposed to the quantity theory (and sometimes favorable to real-bills principles) has been evident in the writings of Tobin (1963), Black (1970), Samuelson (1971), Wallace (1982), and Sargent and Wallace (1982). Still, most economists' attitudes toward the real bills doctrine have remained far from charitable. G. A. Selgin (1989, p. 489.), for example, comments that
The dead horses of economic theory have a habit of suddenly springing back to life again, which is why it is necessary to beat them even when they appear lifeless.
In what follows I hope to revive this dead horse.
II. Backed Money
Empirical studies by Sargent (1982), Smith (1985a, 1985b, 1988), Calomiris (1988), Siklos (1990), Bomberger and Makinen (1991), and Cunningham (1992) have found that the value of money is more accurately predicted by a real-bills type "backing theory" than by the quantity theory. Opposing views favoring the quantity theory have been presented by Bordo and Marcotte (1987), Laidler (1987), Michener (1987), and McCallum (1992). It is not my purpose to comment directly on these debates. Instead, my goal is to clarify the elements of the real bills doctrine, and to present it as a plausible alternative to the quantity theory. At the same time, I hope to correct certain errors that have crippled past inquiries.
Three main ideas form the basis of my discussion:
1. It is possible that fiat money does not exist. A true fiat money would be unbacked, meaning that the issuing entity would hold no assets against the money, would not recognize the money as its liability, and would be unable (not just temporarily unwilling) to redeem the money for anything of value. None of these conditions is true of the U.S. dollar or any other money, and the moneys that have lost all backing (such as the Assignats and the Continental dollars) have simultaneously lost all value.
2. If so-called fiat money is actually backed, then its value is correctly described by the real bills doctrine. That is, banks that follow the real-bills rule of only issuing money in exchange for good security will automatically acquire additional backing every time new money is issued, thus assuring that backing moves in step with the quantity of money.
3. Past discussions of the real bills doctrine have assumed that the value of money is maintained by assuring that the quantity of money moves in step with the aggregate output of goods, and the real bills rule has been wrongly viewed as a means to this end. The correct view is that the real bills doctrine would maintain the value of money by assuring that the quantity of money moves in step with its backing.
A. The Real Bills View of Backed Money
The real bills doctrine holds that money issued in exchange for sufficient security (usually meaning short-term commercial bills) will not cause inflation. For example, Figure 1 represents a bank which has taken in 100 ounces of gold on deposit and issued 100 'credits' (either bank notes or deposits), each of which is a claim to one ounce of gold.
100 oz. of gold 100 credits
I assume for simplicity that this bank has zero net worth, and that each credit gives its holder the right to claim one ounce of gold from the bank on demand. Assuming the bank always honors these claims, each credit will be worth one ounce of gold. But should the bank lose some of its gold, the credits would become an equity claim to whatever assets the bank has left. For example, if the bank lost 30 ounces of its gold, the 100 credits could lay claim to only 70 ounces of gold, so the credits would fall in value to 0.7 ounces each.1
Granting for the moment that this money is backed, its value would not depend on any of the following factors:
(1) the quantity of money,
(2) the convertibility of the money,
(3) money demand,
(4) the quantity of derivative moneys,
I will discuss each of these in turn.
1.The Quantity of Money
Suppose that one of the bank’s customers requests a loan of 100 credits, over and above the 100 credits already issued (Figure 2, line 2), but instead of offering gold in exchange he offers IOU’s (his own or someone else’s) with a current market value of 100 ounces of gold. The banker would have no reason to refuse this offer, and so he would issue 100 more credits, thus doubling the money supply.
1) 100 oz. of gold 100 credits
- - - - - - - - - - - - - - - - - - - - - -
2) IOU’s worth 100 credits
100 oz. of gold
Figure 2 There are now 200 credits laying claim to assets worth 200 ounces of gold, so each credit must still be worth one ounce. The banker can safely issue any amount of money the public desires, provided that he only issues credits to those customers who offer sufficient security (i.e., resources worth one ounce of gold). This rule is nothing but the real bills doctrine, except that the security need not be "short-term commercial bills" (i.e., real bills). Anything worth 100 ounces of gold would serve equally well.
While it is true that money-creation will not affect the credits’ value relative to gold, it is still possible that the issue of credits might reduce the monetary demand for gold and thus reduce its value. This may seem to support the quantity theory proposition that money-creation, even on sufficient security, will cause inflation. However, if gold’s value drops because of competition from the bank's credits, the drop would reflect increased economic efficiency, as monetary gold is released for other uses. But this is the effect of an improvement in monetary technology--not of an increase in the quantity of money.
Two views of money can clearly be set apart at this point: The first is that money's value is determined by supply and demand--by how much money is chasing a given quantity of goods. This is the quantity theory view. The other view, which I am advocating, is that the value of money is determined by its backing. This is the real bills view or, perhaps more accurately, the backing theory view. On this view, token money is valued according to the same principles as any other financial security. For example, suppose that General Motors stock sells for $60 per share. If GM issues additional shares, which it sells for $60 each, then GM's assets would increase in step with the number of shares. There would be no change in the price of GM stock, despite the fact that there would be 'more GM stock chasing a given quantity of goods'. The contention of the backing theory is that token money is itself a financial security, and as such is valued according to its backing, not its quantity. The opposing view of the quantity theory is that token money is like a commodity, and that its value is therefore governed by supply and demand, not by backing.
The backing theory yields the testable implication that the value of token money will be more accurately described by the amount of backing per unit of money than by any model based on the supply and demand of money. Unfortunately, this is a difficult implication to test. While central banks are careful to publish numerous tables describing the behavior of the money stock, they publish little data by which one might guess the true value of their assets.
Suppose that the bank in Figure 2 closes over the weekend, thus making its notes temporarily inconvertible. Then, while the bank is closed, the value of the IOU's drops to 50 ounces of gold (perhaps because of a default by their issuers). The credits would then trade for 150/200=.75 ounces for the rest of the weekend. If the bank restored convertibility at one ounce per credit on Monday morning, it would face a run. The first 150 depositors would get their gold (or something of equivalent value) and the last 50 would get nothing. As the run progressed the expected value of the credits would fall, so that, for example, after 80 credits had been redeemed at one ounce each the value of each remaining credit would be 70/120=.58 ounces. If the bank continued to offer one ounce per note, customers would see it as an empty promise, and they would value the notes at only .58 ounces. Clearly, it is backing that matters, not convertibility. Put another way, convertibility requires backing, but backing does not require convertibility.
If banks can suspend convertibility for a weekend, they can suspend it for a hundred years. For example, a banker might make this offer to his depositors: "Give me resources worth one ounce of gold today, and in 100 years I will return your deposit plus a competitive interest yield." Each credit issued on these terms would initially be worth one ounce of gold, and its value would grow at the rate of interest. If customers preferred the credits to have roughly constant value, then the banker could make periodic interest payments, say by adding .05 credits per year to the account of each credit-holder. Note that the banker need not specify the exact date of redemption, or even that he will pay with certainty. All that matters to the customers is that the credits are a claim to something of value.
We are now in a position to make an important observation: It is possible that what we think of as unbacked fiat money is in fact money that is backed but inconvertible. Consider the usual justification for asserting that the dollar is fiat money:
You cannot convert a Federal Reserve Note into gold, silver, or anything else. The truth is that a Federal Reserve Note has no inherent value other than its value as money, as a medium of exchange. (Tresch, 1994, p. 996.)
Observing that the dollar is inconvertible, economists conclude that it is unbacked. The most remarkable thing about this simple non-sequitur is that it has survived virtually unquestioned for centuries. If we want to show that the dollar is not just inconvertible, but unbacked, it is not enough to say that the Federal Reserve does not pay out gold on demand. Yet economists' belief in fiat money, and in fact the better part of monetary theory, is founded on nothing but this obviously flawed premise. Add to this the facts that the Federal Reserve (like all central banks) does in fact hold assets against the money it issues, that no dollar is ever issued except in exchange for valuable assets, and that the Federal Reserve's own balance sheet plainly identifies those assets as "Collateral Held Against Federal Reserve Notes", and we have good reason to wonder if fiat money is no more real than the phlogiston, ether, and caloric of early physical sciences.
Consider the Bank of England’s suspension of convertibility from 1797-1821. Prior to 1797, the British pound had been convertible into gold at the rate of about 3.9 pounds to the ounce. On February 26, 1797, the Bank suspended convertibility. As Table 1 (Cannan, 1969, p. xliii) shows, the immediate effect of the suspension upon the value of the pound was negligible--a fact which surprised most observers.2
Table 1. British Price Index (1782=100) Year Prices Year Prices
1792 93 1807 132
1793 99 1808 149
1794 98 1809 161
1795 117 1810 164
1796 125 1811 147
1797 110 1812 148
1798 118 1813 149
1799 130 1814 153
1800 141 1815 132
1801 153 1816 109
1802 119 1817 120
1803 128 1818 135
1804 122 1819 117
1805 136 1820 106
1806 133 1821 94
During the suspension, the pound was not a true fiat money. Not only did the Bank maintain assets as collateral against the pound, not only did it continue to recognize the pounds as its liability, but it actually did resume convertibility after a delay of 24 years. There is a difference between a money that is temporarily inconvertible and one that is truly unbacked, but this difference was lost on the major writers of the period (Ricardo, 1811, p. 114; Thornton, 1802, p. 244.), who consistently claimed that the value of the pound was not determined by its backing, but by a limitation of quantity. By way of comparison, the U.S. dollar has been inconvertible to individuals since 1933, and to foreign governments since 1971. Like the old Bank of England, the Federal Reserve has maintained its assets throughout, and has always recognized Federal Reserve dollars as its liability. Unlike the Bank of England, it has not resumed convertibility, and has given no indication that it will. But the fact that the Federal Reserve maintains assets against the dollar means that convertibility could be restored. This would not be true of a bank that simply had no assets.
3. Money Demand
Figure 3 shows two sources of demand for gold. D--use shows industrial and decorative demand, D--money shows monetary demand, and D--total is their horizontal sum. As shown, an ounce of gold is worth 100 bushels of wheat. If there were no monetary demand for gold it would be worth 85 bushels, while if gold were demanded only as money it would be worth 60 bushels.
Starting from equilibrium at point A, the introduction of derivative moneys (such as bank notes, checks, and credit cards) reduces monetary demand for gold, ultimately
reducing its price to 85 bushels (point B)3. Once this point is reached, further creation of derivative moneys cannot affect the value of gold. Furthermore, if the derivative moneys are convertible into an ounce of gold (either instantly or subject to delay), then any amount of derivative moneys can be issued without affecting their value--provided only that their issuers follow the real-bills rule of always acquiring sufficient additional backing whenever additional money is issued. The value of the derivative moneys is not determined by supply and demand, but strictly by backing. Note also that convertibility can be delayed--by a weekend or by one hundred years--without affecting this result.
Now suppose that gold has no industrial or decorative uses, so that monetary use is its only source of demand. (This appears to have been the case for wampum, and for stone money on the island of Yap.) In this case, gold would be worth 60 bushels, and it would fit the definition of a true fiat money, in that it would be unbacked, and valued only because of its usefulness as money. As long as no rival moneys existed, gold could maintain its value of 60 bushels. But as soon as rival moneys began to be issued, the demand for gold would drop, and its value would ultimately reach zero if rival moneys drove gold out of circulation entirely. This kind of fiat money, which has no issuer, can only have value when society is too primitive for rival moneys to be created.
What of fiat money that does have an issuer? If modern textbooks are to be believed, the dollar falls in this category. But paper dollars are no more immune to rival moneys than are shells, stones, or gold. It is clearly possible that rival moneys could drive the value of fiat dollars to zero; yet it is also possible that man-made restrictions on the issue of rival moneys might somehow maintain the dollar's value in the same way that shells and stones can maintain a purely monetary value when there are natural barriers to rival moneys. If we could find examples of paper moneys that have had value in excess of their backing, or were not recognized as the liability of their issuers, then we could claim empirical support for the existence of fiat money. But since there is no obvious historical example of such a money, and since all modern paper moneys are recognized as the liability of their issuers, it seems that the limited evidence available to us favors the belief that fiat money does not exist, and that its existence is precluded by the presence of rival moneys.
If the banker in Figure 1 has resources worth 100 ounces of gold, as backing for 100 credits, then those credits will be worth one ounce each regardless of the public's demand for them3. If their value exceeded one ounce by (say) 2 percent, then rival bankers could earn easy profits by issuing credits for 1.02 ounces of gold, keeping 1 ounce as backing, and spending the seignorage of .02 ounces on their own consumption. This profit opportunity will exist as long as there is any seignorage, so the only stable solution is for the seignorage to be driven to zero. The same reasoning implies that there can be no such thing as fiat money, since fiat money is money whose whole value is seignorage. The proposition that private, competitive banks are unable to issue fiat money does not admit of dispute. That the same thing could also be true of government banks has, to my knowledge, never even been suggested.
Depending on who is talking, we hear that fiat money has value because other people value it (Samuelson, 1980, p. 261), because the government accepts it for taxes4 (Wicksteed, 1910, p. 619), because it is useful for making exchanges and limited in supply (Marshall, 1922, p. 49), because the government requires banks to hold it (Fama, 1980, p. 56), or because it allows us to transfer wealth to our children (Wallace, 1980, p. 50). The trouble with these theories is that they fail to consider rival moneys. Each theory begins by asserting that there is some force (e.g., liquidity services) that creates a demand for intrinsically worthless pieces of paper. They then assert that it would only be necessary to limit the supply of these pieces of paper in order to give them value. But this gives a free lunch to the issuer of those pieces of paper, and creates an incentive for rivals to issue their own pieces of paper, until competition drives their value to zero. Of course, no one believes that private, competitive banks can earn this kind of free lunch. Furthermore, if a private bank could issue notes on which it paid no interest, while investing the proceeds at 5%, then competitors would issue rival notes until the interest spread just covered costs of printing, periodic redemption, controlling counterfeiting, etc. Given this, it is strange to see how easily economists accept the proposition that central banks earn seignorage on their note issue, and that note issue therefore gives a free lunch to the Federal Reserve, especially if the dollars go to foreign countries. Since most of us are trained to be suspicious of free lunches, this idea deserves some skepticism.
The only reason to believe that the Federal Reserve earns seignorage is that it has the power to suppress rival bank notes. But governments cannot suppress commodity money, credit, foreign bank notes, or barter. There are also traveler’s checks, gift certificates, and scrip, all of which are bank notes issued by non-bank institutions. (In point of fact the only entities barred from issuing bank notes are banks themselves.) Given this rivalry, it is hard to believe that note issue could yield abnormal profits, even to government banks. Where countries are small, weak, and close together, it seems impossible.
But assume for the sake of argument that a country is strong enough to erect significant barriers to rival bank notes. The government notes will still face rivalry from derivative moneys. (By 'derivative money', I mean money that is a claim to some other money, in the sense that a dollar in a checking account is a claim to a Federal Reserve note.) For example, a farmer might pledge $10,000 of wheat to a banker, and the banker in turn will lend the farmer $10,000 by crediting that amount to his checking account. By this exchange the banker will have effectively coined wheat into dollars. If we accept the assertion that the dollar has value because of the liquidity services it provides, then the creation of the new wheat-backed derivative dollars would reduce the demand for Federal Reserve dollars, and thus would reduce their value. If there were no constraint on the issue of derivative dollars, the value of Federal Reserve dollars would be driven to zero.
One might argue that banks are constrained by reserve requirements, but these only apply to conventional bank accounts-- not to credit cards, eurodollars, scrip, and so on. In light of this limitless potential for the issue of rival moneys, fiat money seems implausible. In contrast, the view that the dollar is backed but inconvertible only requires us to believe that money is valued for the same reason that any other financial security is valued.
A stock market analogy might help to explain the role of reserve requirements. Just as bankers issue checking accounts that are claims to Federal Reserve dollars, stock market traders routinely issue derivative securities that are claims to GM stock. Suppose that those traders were required to hold "reserves" of genuine GM stock against the derivative shares that they issue. Would this requirement increase the value of GM stock? The answer is no, since this requirement would not affect GM's assets or its liabilities. If one accepts the idea that the dollar is backed, then the same reasoning implies that reserve requirements are irrelevant to the value of the dollar.
A reasonably skeptical reader could still argue that a constraint on rival moneys could cause the dollar to sell for a few points above its backing. However, one could also argue that GM stock could be raised a few points (or lowered!) by a constraint on the issue of rival stocks. But this argument would not persuade economists to abandon the theory that stocks are valued according to their backing. When applied to money, the same argument is clearly an inadequate reason for believing that the dollar is a pure fiat money.
Why does the Federal Reserve (and every other central bank) bother to hold gold and financial securities if the dollar does not get its value from backing? How could fiat money ever come into circulation in the first place? Why issue dollars through an expensive central bank instead of just printing them and spending them? Why do even the smallest and weakest countries seem to be able to maintain "fiat" money in circulation? These questions and many more have inspired a mountain of convoluted monetary theories. But if fiat money is in fact an illusion--if it is actually backed but inconvertible, then these questions do not even arise.
4. The Quantity of Derivative Moneys
Checking accounts issued by private banks entitle depositors to claim Federal Reserve notes on demand. Thus we could call the accounts 'derivative money' (a term I prefer to 'inside money') since they are claims to Federal Reserve dollars. The dollar, in turn, is an inconvertible claim to the assets of the Federal Reserve, and is itself a derivative money, even though we commonly think of it as base money. By analogy, there are derivative financial securities (options, warrants, etc.) that are claims to GM stock. The GM shares, in turn, are a claim (generally inconvertible) against GM's assets. Thus the base stock is itself a derivative security.
The issue of derivative shares of GM stock does not change GM's assets or liabilities, and therefore does not depreciate GM stock. Similarly, if the dollar has value because of its backing, then the issue of derivative dollars does not reduce the value of the dollar. The quantity theory, however, implies that derivative dollars reduce the demand for base dollars and thus cause inflation. On this view, a legitimate banker is no different from a counterfeiter: Both increase the quantity of money, so both cause inflation! This belief has led to a number of proposals to require all banks, public and private, to maintain 100% reserves against the money they issue (e.g., Friedman, 1948, p. 372.). This idea, besides being out of character for libertarian economists, ignores the fact that banks recognize their money as their liability, while counterfeiters do not.
Derivative moneys raise the question of what is 'money' and what is not. Should we include only gold? Notes issued by the central bank? Private bank notes? What about checking accounts, credit cards, traveler’s checks, eurodollars, overdrafts, and gift certificates?5
These questions could only matter to someone who believes that the value of money depends on its quantity. No one bothers to wonder whether derivative shares of stock (options, hypothecated shares, etc.) should be counted along with genuine shares. We simply recognize that derivative shares can take many forms, and that each share will be valued in accordance with the resources backing it. If economists understood money as well as they understand stock, they would recognize that derivative moneys also take many forms, but that their quantity is irrelevant to the value of the dollar. Misunderstandings of this point are widespread. For example, Salin (1984, p. 13.) worries that eurodollars reduce the Federal Reserve's control over the world supply of dollars. On this view, the issuer of a eurodollar is violating the 'brand name' of the dollar. The real bills view, however, is that a foreign bank which issues a eurodollar is analogous to a foreign brokerage house which issues a derivative share of GM stock. Viewed in this way, we see that eurodollars are no cause for concern to Americans, and in fact are likely to improve the efficiency of the market for dollars.
This view of derivative moneys has at least two testable implications: (1) The introduction of new derivative moneys (e.g., credit cards, electronic payment systems) will have no effect on the value of base money, and (2) The value of money should be more closely correlated with the quantity of base money than with broad aggregates such as M1, which includes various derivative moneys.
B. Inflation, Money Shortages, and Inflationary Feedback
Return to the case of a bank that holds miscellaneous goods worth 100 ounces of gold as backing for 100 outstanding credits. Now suppose that the bank follows an 'easy money' policy and begins issuing credits in exchange for securities worth only .95 ounces of gold. If the banker issues 50 new credits on these terms, then the value of each credit will drop to (100+(50x.95))/150=.983 ounces. We get the familiar result that easy money leads to inflation. Note that the inflation results from a drop in the bank's ratio of assets to money, not from an increase in the quantity of money relative to aggregate output of goods.
When depositors see the bank handing out credits worth one ounce of gold in exchange for securities worth only .95 ounces, they will rush to buy these credits. Thus the quantity of bank money will expand as long as the bank follows its easy money policy. At the same time, the value of the money will fall because of the drop in the bank’s ratio of assets to money. This point has confused economists for centuries. The true cause of inflation is that the bank issues money for insufficient security. The apparent cause of inflation--the increase in the quantity of money--is really just a side effect of the bank's easy money policy.
This analysis provides an interesting view of credit rationing and money shortages. In discussing an English credit rationing episode, Friedrich Hayek commented that:
This recourse to a rationing of credit caused renewed stringency in the money market in the spring of 1796 and evoked loud protests from the City (London).
It is not easy to reconcile these complaints about the continued scarcity of money during this period with the no less insistent complaints about high prices, and with the continued unfavorable course of the exchanges." (Hayek, 1933, p. 40)
From a real-bills perspective, it is easy to reconcile the two sets of complaints. The fact that the bank was rationing its notes indicates that it was selling the notes too cheaply. Any issue of money on these terms would reduce the amount of backing per note. Thus the value of the pound would fall, even if the Bank were issuing fewer notes than usual (and thereby generating complaints of a shortage of money).
Economists often scoff at complaints of money shortages, since the quantity theory implies that a reduction in the quantity of money will simply cause a commensurate increase in the value of the remaining money. But on real bills principles, a reduction of the money supply--accompanied by an equal reduction of backing--leaves the value of money unchanged, while leaving the community without enough cash to conduct business conveniently. Thus, a reduction in the money supply would normally be recessionary. The direction of the effect on prices, however, would depend on whether the amount of backing per unit of money was rising, falling, or constant.
A problem arises when the bank's assets are denominated in the bank's own credits. (e.g., The Federal Reserve's bondholdings are denominated in dollars.) If the bank's credits were to depreciate for any reason, then the bank's assets would also depreciate. The credits would then fall still further, and so on. Monetary theorists from Henry Thornton to Lloyd Mints have mistakenly believed that this kind of ‘nominal’ backing is no backing at all, but this is not the case for money or for any other financial security. For example, suppose that GM issues 100 new shares of stock, which it sells for $60 each. Then, it uses the cash proceeds to buy derivative shares of GM stock (which were issued, let us suppose, by Merrill Lynch). This transaction will not change the value of GM stock, since GM’s assets will have increased exactly in step with its new issue of stock. The derivative shares, however, are nominal backing—that is, they are denominated in GM stock. In exactly the same way, a central bank that issues liabilities called dollars, while holding assets denominated in those same dollars, is using nominal backing, but it is backing just the same.
Consider the following illustration: Let P represent the price, in gold, of a bank's credits. Assume that the bank initially has 100 outstanding credits, each worth one ounce of gold, and that its assets consist of 30 ounces of gold plus bonds, denominated in the bank's own credits, which have a market value of 70 credits (Figure 3). Now suppose
Figure 3 that the bank follows an easy money policy and issues 10 new credits in exchange for bonds with a true value of only 6 credits (line 2). If assets (30 ounces of gold plus bonds worth 76 credits) are to equal liabilities (110 credits), then it must be true that 30+76P=110P, which implies P=.888 ounces of gold. If the bank had instead started with 20 ounces of gold and bonds worth 80 credits, the same easy money policy would have reduced the value of the bank's credits to P=.833 ounces. Thus, the bank's holdings of gold or other 'real' assets would limit the force of inflationary feedback. The smaller are real reserves, the more volatile the bank's money, but nominal assets still back money in a meaningful way.
C.Adam Smith's View of the Real Bills Doctrine
Debates between quantity theorists and real bills adherents have flared repeatedly over the last three centuries. As a rule the quantity theory has come out on top, but often the controversies "have slumbered, rather from the exhaustion of the combatants than from the acknowledged defeat of either party." (Farrer, 1898, p. 78.) When a debate becomes as protracted as this, there is good reason to think that both sides are asking the wrong questions. I contend that two key errors lie at the root of the problem. First, both sides accepted the existence of fiat money, without exploring the possibility that paper money could be backed but inconvertible. Second, both sides held that stable prices would be achieved if the money supply rose and fell with the "needs of business" (i.e., real output). They ignored the role of the issuing bank's assets. Both of these errors can be traced to Adam Smith's treatment, which has strongly influenced subsequent discussions.
What a bank can with propriety advance to a merchant or undertaker of any kind, is not either the whole capital with which he trades, or even any considerable part of that capital; but that part of it only, which he would otherwise be obliged to keep by him unemployed, and in ready money for answering occasional demands. If the paper money which the bank advances never exceeds this value, it can never exceed the value of the gold and silver, which would necessarily circulate in the country if there was no paper money; it can never exceed the quantity which the circulation of the country can easily absorb and employ.
When a bank discounts to a merchant a real bill of exchange drawn by a real creditor on a real debtor, and which, as soon as it becomes due, is really paid by that debtor; it only advances to him a part of the value which he would otherwise be obliged to keep by him unemployed and in ready money for answering occasional demands. (Smith, 1776, p. 322)
The idea that a bank should only lend as much money as its customers would otherwise have kept in their tills is nonsense. We all know that a bank can, with propriety, lend to a merchant any amount reasonably short of the merchant's net worth. Such a bank will always have assets sufficient for its liabilities, and thus its money will maintain its value. Smith's proposition that the amount of bank money should correspond to the amount of gold and silver it replaces implicitly assumes that money's value will be maintained by a limitation of its quantity, and not by a matching of bank assets to the quantity of money. Having no clear idea of this distinction, Smith made a dangerous misinterpretation of central bank policies:
By issuing too great a quantity of paper, of which the excess was continually returning, in order to be exchanged for gold and silver, the bank of England was for many years together obliged to coin gold...at an average (of) about eight hundred and fifty thousand pounds. For this great coinage the bank (in consequence of the worn and degraded state into which the coin had fallen a few years ago) was frequently obliged to purchase gold bullion at the high price of four pounds an ounce, which it soon after issued in coin at 3 l. 17 s. 10 1/2 d. an ounce, losing in this manner between two and a half and three per cent. on the coinage of so very large a sum. (Smith, 1776, p. 286.)
By attributing the bank's loss upon the coinage to excessive note issue, Smith got the chain of causation precisely backwards. Note issue expanded because the bank issued money on insufficient security. But on Smith's interpretation, it was excessive note issue that forced the bank to accept insufficient security for its notes. When the bank pays out four one-pound notes for an ounce of gold that it reissues as coins worth 3 l. 17 s. 10 1/2d., (about 3.9 pounds) then customers will eagerly bring one ounce of gold to the bank to exchange for 4 pound notes. They will then return 3.9 of those notes to the bank in exchange for coins containing one full ounce. With the bank losing 2.5% on each exchange, the bank would soon exhaust its treasure.
What would get the bank started on such a ruinous course? Smith himself gave the answer: "the worn and degraded state into which the coin had fallen". When a new one pound coin was issued, it would have contained gold worth one pound. Because of wear and clipping, that same coin might soon contain gold worth 0.975 pounds. The heavy coins would disappear from the circulation and the value of the pound would drop by 2.5%. Thus, as Smith states, it would require four one-pound notes to buy the same amount of gold that it previously bought for 3.9 notes. But the effort to restore the pound to its old value will be hopeless. If the bank recoined the newly-purchased gold and sold it for 3.9 pounds per ounce, it would be buying its own notes on the market for 2.5% more gold than the notes were worth. The pound’s backing would fall, and the value of the pound would drop.
Modern central banks make the same mistake when they attempt to support their currency in world markets. Suppose, for example, that the pound trades for $1.60, but that the Bank of England wants the pound to trade for $1.80. The quantity theory prescription would be for the Bank of England to use its dollar reserves to buy pounds in the open market for $1.80. But by paying $1.80 for a British pound that is only worth $1.60, the Bank would lose $.20 on each purchase. Its ratio of assets to currency would drop, and the value of the pound would fall. Small wonder then, that efforts to support various currencies are so often followed by devaluation (Taylor, 1982, pp. 356-68). This empirical result is exactly what the real bills doctrine implies, and exactly opposite to the implications of the quantity theory.
The real bills doctrine holds that money issued for good security will not cause inflation. When money is backed, this doctrine is a simple and obvious truth: If every issue of money is matched by an equal increase in its backing, then the issue of new money will not change the amount of backing per unit of money, and so the value of the money will be stable. On the other hand, the real bills doctrine would not be true of unbacked (fiat) money. Quantity theorists contend that fiat money has value because it is limited in supply. Thus, even if new money is issued for good security, the simple fact that there is more money will cause inflation. The two views are clearly set apart: Any given money is either backed or unbacked. If it is backed, then its value is correctly described by the real bills doctrine. If it is unbacked, then its value is correctly described by the quantity theory.
I take the view that all money is backed, and that fiat money does not exist. I offer three observations in support of this view: (1) Fiat money would create a free lunch for its issuer, thus inducing rivals to issue their own money until competition drove the money's value to zero. (2) The fact that a currency such as the dollar is inconvertible does not imply that it is unbacked. Convertibility can be suspended for a weekend or for a hundred years, but as long as the issuing bank maintains valuable assets against its money, we cannot call that money unbacked. (3) All central banks recognize their money as their liability. They maintain valuable assets as collateral against that money, and they have, on many occasions, used those assets to resume convertibility.
The idea that so-called 'fiat' money is actually backed yields the testable implications that the value of money would be unaffected by the quantity of base money, by convertibility, by money demand, or by the quantity of derivative money. The backing theory also predicts (1) that we should seldom or never find examples of token moneys whose issuers are unable (not just temporarily unwilling) to redeem them; and (2) that a bank's attempt to support the value of its own money in foreign exchange markets will actually reduce the value of the money. Both of these predictions are consistent with experience.
When the central bank fails to take sufficient security for its money, inflation will follow from the resulting drop in its ratio of assets to units of money. At the same time, the easy money policy will lead to an increase in the quantity of money, and observers will wrongly conclude that the increase in the quantity of money caused the drop in its value. This faulty perception has allowed the quantity theory to become the dominant theory of money, while the real bills doctrine has been wrongly discredited.
Adam Smith's view was that as long as banks only issued money in exchange for real bills, the quantity of money would automatically move in step with the economy's real output. Ever since, the economics profession has unfortunately viewed the real bills doctrine as a rule for limiting the quantity of money, which is completely counter to a proper understanding of the real bills view. The proper view is that the real bills rule would lead to stable prices by maintaining sufficient backing for token money.
References Ashton, T. S., and Sayers, R. S., Papers in English Monetary History, London: Oxford University Press, 1953.
Bagehot, Walter, Lombard Street, 1873. Reprinted by Richard D. Irwin, New York, 1962.
Black, Fischer, "Banking and Interest Rates in a World Without Money," Journal of Bank Research, vol. 1, number 3, Autumn 1970, pp. 9-20.
Brock, Leslie V. The Currency of the American Colonies, 1700-1764: A Study in Colonial Finance and Imperial Relations. New York: Arno, 1975.
Bomberger, William A., and Makinen, Gail E. “The Hungarian Hyperinflation and Stabilization of 1945-1946” Journal of Political Economy, volume 91, number 5 (1983), pp. 801-24.
Bordo, Michael D., and Marcotte, Ivan A. “Purchasing Power Parity in Colonial America: Some Evidence for South Carolina 1732-1744: A Comment” Carnegie-Rochester Conference Series on Public Policy, vol. 27 (Autumn 1987), pp. 311-23.
Bosanquet, Charles, Practical Observations on the Report of the Bullion Committee, London: Printed for J. M. Richardson, 1810.
Calomiris, C. W., "Institutional Failure, Monetary Scarcity, and the Depreciation of the Continental." Journal of Economic History, volume 48, 1988, pp. 47-68.
Cannan, Edwin (Editor), The Paper Pound of 1797-1821: A Reprint of The Bullion Report, 1925. Reprinted by Augustus M. Kelley, New York: 1966.
Cunningham, Thomas J., "Some Real Evidence on the Real Bills Doctrine versus the Quantity Theory", Economic Inquiry, volume XXX, April 1992, p. 371.
Farrer, Studies in Currency, 1898. Reprinted by Augustus M. Kelley, New York: 1968.
Fullarton, John, Regulation of Currencies of the Bank of England (second edition), 1845. Reprinted by Augustus M. Kelley, New York: 1969.
Gilbart, J. W., The History, Principles, and Practice of Banking, 1882. Reprinted by Greenwood Press, New York: 1968.
Harley, [Robert] (or rather, Simon Clement), The Faults on Both Sides (1710), in Somers' Tracts, XIII. (Second Edition; London:1805-15).
Hayek, Friedrich A., Monetary Theory and the Trade Cycle, 1933. Reprinted by Augustus M. Kelley, New York: 1966.
Laidler, David, “Wicksell and Fisher on the ‘Backing’ of Money and the Quantity Theory: A Comment on the Debate between Bruce Smith and Ronald Michener.” Carnegie-Rochester Conference Series on Public Policy, vol. 27 (Autumn 1987), pp. 325-34.
McCallum, Bennett T. “Money and Prices in Colonial America: A new Test of Competing Theories.” Journal of Political Economy, volume 100, number 1 (1992), pp. 143-61.
Michener, Ronald. “Fixed Exchange Rates and the Quantity Theory in Colonial America.” Carnegie-Rochester Conference Series on Public Policy, vol. 27 (Autumn 1987), pp. 233-307.
Mints, Lloyd, A History of Banking Theory, Chicago: University of Chicago Press, 1945.
Ricardo, David, Economic Essays, 1811. Reprinted by G. Bell and Sons, Ltd., London: 1926.
Salin, Pascal, Currency Competition and Monetary Union, Boston: Martinus Nijhoff, 1984.
Samuelson, Paul A., Economics, eleventh edition, New York: McGraw-Hill, 1980.
Samuelson, Paul A., "Reflections on the Merits and Demerits of Monetarism," in Issues in Fiscal and Monetary Policy: The Eclectic Economist Views the Controversy, ed. James J. Diamond (DePaul University, 1971); quoted in Leonall C. Anderson, "The State of the Monetarist Debate" Federal Reserve Bank of St. Louis Review, volume 55, number 9 (September 1973), pp. 2-8.
Sargent, Thomas J., "The Ends of Four Big Inflations", in Robert Hall, editor, Inflation: Causes and Effects, Chicago: University of Chicago Press, 1982.
Sargent, Thomas J., and Wallace, Neil, "The Real Bills Doctrine versus the Quantity Theory: A Reconsideration.", The Journal of Political Economy, vol. 90 (December 1982), pp. 1212-36.
Selgin, G. A., "The Analytical Framework of the Real Bills Doctrine", Journal of Institutional and Theoretical Economics, volume 145, (1989), p. 489.
Siklos, Pierre L., "The Link Between Money and Prices Under Different Policy Regimes: The Postwar Hungarian Experience." Explorations in Economic History, volume 27, 1990, p. 468.
Smith, Adam, Wealth of Nations, 1776. Reprinted by Random House, New York: 1937.
Smith, Bruce D. "American Colonial Monetary Regimes: The Failure of the Quantity Theory and Some Evidence in Favour of an Alternative View." Canadian Journal of Economics volume 18, August 1985, pp. 531-65. (a)
Smith, Bruce D., "Some Colonial Evidence on Two Theories of Money: Maryland and the Carolinas", Journal of Political Economy, volume 93, number 6, December 1985, p. 1178. (b)
Smith, Bruce D., "The Relationship Between Money and Prices: Some Historical Evidence Reconsidered." Federal Reserve Bank of Minneapolis Quarterly Review, volume 12, Summer 1988, pp. 18-32.
Taylor, Dean, "Official Intervention in the Foreign Exchange Market, or, Bet Against the Central Bank, Journal of Political Economy, volume 90, number 2, April 1982, pp. 356-368.
Thornton, Henry, The Paper Credit of Great Britain, 1802. Reprinted by Augustus M. Kelley, New York: 1965.
Tobin, James, Commercial Banks as Creators of 'Money'", in Deanne Carson (ed.), Banking and Monetary Studies, Homewood: Richard D. Irwin, 1963.
Tresch, Richard, Principles of Economics, St. Paul, Minnesota: West, 1994.
Wallace, Neil, "A Modigliani-Miller Theorem for Open-Market Operations," The American Economic Review, vol. 71 (1982), pp. 267-274.
Wallace, Neil, "The Overlapping Generations Model of Fiat Money", in John H. Kareken and Neil Wallace (eds.), Models of Monetary Economies, Minneapolis: Federal Reserve Bank of Minneapolis, 1980, pp. 49-82.
Wicksteed, Phillip H., The Common Sense of Political Economy, volume II, 1910. Reprinted by Routledge & Kegan Paul, Ltd. London: 1957.
1The value of these credits resembles the payoff to the seller of a put option. If depositors believe that the bank might lose some of its gold, then it seems that they would value these credits at slightly less than one ounce. However, depositors might realize that gold held in their own pockets is also subject to loss, so they might be indifferent between the bank’s uncertain promise of convertibility and their own uncertain ability to hold on to the gold. Alternatively, the argument I make below would not be substantially changed if the bank were assumed to have positive net worth, which would act as a cushion against the loss of gold.
2”Looking back from the safety of 1798, ‘A Proprietor of Bank Stock’ thus summarized the transition: ‘In this desponding state, when all men dreaded, with the utmost anxiety, the event that was seen to be inevitable, and not far distant, and which it was supposed would involve the Kingdom in general bankruptcy and intire ruin, the 26th February, 1797, was the crisis that gave the happy turn, and almost instantly dismissed all the horrors and fears that surrounded us; restored complete confidence...'” (Ashton and Sayers, 1953, p. 19.)
3The release of gold from monetary uses will yield social gains approximately equal to the shaded area.
3Their value would also be independent of their velocity of circulation, and of the economy's output of goods and services. Just as with any other financial security, it is backing alone that matters.
4There have been many cases where acceptability for taxes has conferred value on paper money, but these have been cases where the taxes themselves constituted backing. For example, in the American colonial period, colonies would issue paper money either on loan or in direct payment for government expenses. These notes were declared acceptable for taxes, and they were officially rated in terms of some other money. A New York shilling, for example, was legally rated at 8 shillings to the ounce of silver from 1709-1718 (Brock, pp. 66-67.). Had these shillings been backed in an ordinary sense, the holder of 8 shillings could have claimed one ounce of silver from their issuer. But the tax backing used in the colonies gave the holder of 8 shillings the right to discharge a tax (or other debt to the colony) of one ounce of silver. Conventional backing thus depends on the ability of the issuer to pay silver, while tax backing depends on the issuer’s ability to take away silver.
5In 1710, people denied that bank notes were money, since they ultimately had to be paid off in coin (e.g., Harley, 1710, p. 42). In 1845, people denied that checking accounts were money, since they ultimately had to be paid in notes or coin (Fullarton, 1845, p. 32). Having learned nothing from this, modern economists deny that credit cards are money, since they must ultimately be paid off by check, notes, or coin.