Front Page Titles (by Subject) 8: The Supply of Currency - The Theory of Free Banking: Money Supply under Competitive Note Issue
The Online Library of Liberty
A project of Liberty Fund, Inc.
Search this Title:
8: The Supply of Currency - George A. Selgin, The Theory of Free Banking: Money Supply under Competitive Note Issue 
The Theory of Free Banking: Money Supply under Competitive Note Issue (Lanham, MD.: Rowman & Littlefield, 1988).
About Liberty Fund:
Liberty Fund, Inc. is a private, educational foundation established to encourage the study of the ideal of a society of free and responsible individuals.
This title has been put online with the permission of the author.
Fair use statement:
This material is put online to further the educational goals of Liberty Fund, Inc. Unless otherwise stated in the Copyright Information section above, this material may be used freely for educational and academic purposes. It may not be used in any way for profit.
The Supply of Currency
Money may be held in either of two forms: deposits, from which payments can be made by check, or currency—hand-to-hand money. This chapter examines the capacity of free banks and central banks to accommodate changes in currency demand. To simplify the problem it assumes that the total demand for money is unchanging, so that the central bank can treat it as known. Then the only changes in currency demand that need to be accommodated are those arising from decisions to alter the composition of money holdings—their division between currency and deposits.1
Suppose, for example, that a shift from deposit demand to currency demand occurs in a central-banking system in which deposit banks have no excess reserves. Although the total demand for money has not changed, people want to exchange their deposit balances for currency balances. How can a central banking system provide the needed adjustment? Can it change the relative mixture of deposits and currency in circulation without disrupting monetary equilibrium? How, in this regard, will its performance compare to that of a free-banking system?
Before answering these questions we must carefully distinguish currency demand, which is simply a demand for circulating means of payment, from outside-money demand, which is demand to hold a form of money that does not involve granting credit to banks. A rise in currency demand is a routine occurrence which does not involve any loss of confidence in banks; it can in theory be satisfied by a circulating form of inside money. In contrast, a rise in outside-money demand means a demand to exchange inside money for outside money, the ultimate money of redemption. In a closed system this implies either a loss of confidence in banks issuing inside money (which contradicts the assumptions of the present part of this study) or a failure of the banking system to provide enough inside money for use as currency.2
The Relative Demand for Currency
The public’s division of its demand for money between deposit demand and currency demand is not arbitrary. Particular sorts of plans call for holding currency rather than checkable deposits. Currency is more useful for making change; but more importantly the demand to hold currency reflects the degree to which sellers more readily accept currency than checks. One reason for the greater acceptability of currency is that sellers of goods and services may wish to avoid the inconvenience of depositing or cashing checks. More significantly, acceptance of a check requires a level of trust beyond what is required in the acceptance of currency of equal face value: the acceptor of a check has to have confidence not just in the bank upon which the check is drawn, which may or may not be good for the transferred sum, but also in the drawer of the check himself, who may or may not possess an adequate deposit balance.
Nor is the relative demand for currency constant. As Agger notes (1918, 85), it changes along with “basic changes in the economic life of the community” and with “changes in the disposition that is to be made of . . . borrowed funds.” In the United States until the 1930s the historical trend was toward less reliance upon currency and greater use of checks and other means for direct transfer of deposit balances. This was due mainly to improvements in deposit banking, which were spurred-on in part by the suppression of competitive note issue. In the last fifty years or so the trend has changed, and the demand for currency relative to total money demand has grown substantially.3
Other factors have historically caused the currency-deposit mixture to alter in a less regular way. An increase in retail trade relative to wholesale trade favors greater use of currency, because the former involves smaller, anonymous exchanges where less trust is possible, whereas the latter involves larger exchanges among previously acquainted parties. In the past, when wage payments were more often made in currency, payroll requirements caused weekly and quarterly cycles in currency demand. The demand for currency also increased during the autumnal expansion of agricultural activity, and there are still seasonal peaks in demand due to holidays (such as Christmas) which involve a burst of retail trade. Besides these influences Phillip Cagan, in his study of “The Demand for Currency Relative to Total Money Supply” (1958) lists the following: (a) expected real income per capita; (b) interest rates available on demand deposits (a measure of the opportunity cost of holding currency); (c) the volume of travel; (d) the degree of urbanization; (e) the advent of war; (f) the level of taxes and incentives for tax evasion; and (g) the extent of criminal and black-market activities. Changes in the currency ratio due to these and other factors since the turn of the century are shown in figure 8.1.
A final factor already alluded to which affects the relative demand for currency is the extent of business confidence. According to Agger (1918, 86), a decline in confidence “lessons the acceptability of the check as an instrument of exchange and usually involves an increase in the demand for media of more general acceptability.” Except during panics a loss of confidence extends only to individuals and not to banks so that, although it causes an increase in currency demand, it does not necessarily involve any increase in outside-money demand; that is, it does not involve a desire to remove outside money from the banking system. “Ordinarily,” Agger notes, “the shifting of demand is rarely so complete [and] it is only isolated banks that suffer a complete loss of confidence.”4 Pressure is more likely to be exerted by depositors desiring currency, including bank notes, than by holders of notes seeking to redeem them in outside money.5
This variety of influencing factors makes the relative demand for currency highly variable and sometimes unpredictable.6 In consequence, a central bank may have difficulty accommodating changes in the relative demand for currency even when the demand for inside money as a whole does not change. Yet it is essential that the public be able to acquire media of exchange in a mixture that suits its needs. Holders of inside money want to be able to switch from deposits to currency or vice versa depending upon which means of payment or combination of means suits its circumstances. If the public’s wants are not satisfied significant inconvenience and reduced opportunities for making desired purchases result.
Bank borrowers also may need to receive credit in one rather than the other form (checkable deposits or currency), so that a relative deficiency of either form will cause credit-market stringency just as if the the total availability of loanable funds were reduced. As Agger puts it:
Inability to meet an expanding demand or impediments in the way of issue of either form of bank credit may entail serious consequences. For those desiring credit in either form and unable to obtain it [for want of the desired media] the situation is alarming. The normal conduct of their business may depend upon obtaining bank accommodations of an acceptable form. Stringency in the market for such accommodation is . . . bound to be costly and a source of anxiety.7
The amount of credit granted in a well-working banking system should not depend on the form of payment medium wanted, so long as there is no special demand for the ultimate money of redemption. A well-working system should also permit the unrestricted interconversion of deposits and currency once either is already outstanding.
Currency Supply under Free Banking
When banks are unrestricted in their ability to issue bank notes each can meet increases in its clients’ demands for currency without difficulty and without affecting its liquidity or solvency. Under such free-banking conditions the “transformation of deposits into notes will respond to demand,” and banks will be able to supply credit in the form that borrowers require (Agger 1918, 154). The supply of currency is flexible under unrestricted note issue because bank note liabilities are, for a bank capable of issuing them, not significantly different from deposit liabilities.8 “In the absence of any restriction,” Agger writes, “it is a matter of indifference to the [note-issuing] bank which form its credit takes” (1918, 154). When the customers of a note-issuing bank—borrowers or depositors—desire currency, the bank offers them its own notes instead of a deposit balance. The issue of notes in exchange for deposit credits involves offsetting adjustments on the liability side of the bank’s balance sheet, with no change on the asset side. Suppose, for example, that a deposit-holder having a balance of $500 wants to withdraw $100 in currency.9 The bank simply hands him $100 of its notes. Then, on its balance sheet, it reduces the entry for “liabilities-deposits” $100 and increases the entry for “liabilities-circulation” $100. The composition of the bank’s liabilities changes but their total amount stays the same; and this is all that ought to happen, since by hypothesis the total demand for inside money has not changed.
It is even possible, as far as currency demand is concerned, for a note-issuing bank to hold enough notes on hand (in its vault and tills) to meet currency demands of its creditors and borrowers to the full amount of its outstanding demand liabilities. The only cost involved would be the cost of the notes themselves—an investment in paper and engraving. The notes are not obligations of the bank and pose no threat to its solvency until they start circulating. On the other hand, as long as notes are in a bank’s vault or tills they cannot be treated by it as a reserve asset capable of supporting its outstanding credits: unlike outside money, they are useless for settling clearing balances with other banks. A competitive bank’s own notes serve one purpose only, which is to satisfy that part of its clients’ demands for inside money which consists of a demand for currency. When no longer needed in circulation the notes return to the issuing bank, which may use them again the next time the demand for currency increases.10
In practice free banks would generally not keep on hand notes equal to 100 percent of their deposits, because the likelihood of demand for inside money shifting entirely into currency is minuscule. Even though only minor costs are involved, excessive note stockpiles would be wasteful. Banks would, instead, keep on hand as many notes as they would be likely to need to cover unusual, but not extremely freakish, demands. In the truly exceptional event of demands exceeding available note supplies relief would be no further away than the printing press.11
Freedom of note issue thus ensures the preservation of an equilibrium money supply as demand shifts from deposits to currency and vice versa. It assures that credit offers of persons willing to hold inside money are exploited even when the offerers want to hold bank promises in a form useful in circulation. It also assures that a growing demand for inside money that involves an absolute increase in currency demand is readily accommodated, instead of going unsatisfied because of a shortage of currency.12 The ability of free banks to function smoothly as intermediaries even in the face of changing currency demand stems from their note-issuing powers.
Monopolized Currency Supply
Under monopolized currency supply the ability of non-note-issuing (deposit or commercial) banks to convert deposits into currency is restricted. Deposit banks are not able independently to fulfill currency demands. They have to draw instead on their holdings of notes or fiat currency (or deposits convertible into notes or fiat currency) of the monopoly bank of issue.13 In doing so they reduce their reserves of high-powered money. It follows that, unless the monopoly bank of issue adjusts the amount of its credits to the deposit banks to offset their reserve losses due to currency demand,14 their lending power decreases. The banks will have to contract their liabilities. A change in the form in which the public wishes to hold money balances causes a disequilibrating change in the total supply of money.15
The same conclusion holds for uncompensated reductions in the relative demand for currency, which in a system with monopolized currency issue results in a return of currency to the deposit banks, who add it to their reserve holdings and use it as a basis for credit expansion. A fall in the relative demand for currency results in monetary overexpansion even though the demand for money has not fallen and even though there is no expansion of credit by the monopoly bank of issue. If changes in the relative demand for currency are not to result in monetary disequilibrium under central banking, the central bank must engage in continual “reserve compensation.” It has to adjust the supply of base or high-powered money in response to changes in the amount of base money needed in circulation.
This result, that changes in the relative demand for currency will affect total money supply under central banking unless offset by reserve compensation, is well recognized in the literature on central banking.16 But past writers have tended to view the problem as one inherent in all fractional-reserve banking, whereas the truth is that it is inherent only to systems where the issue of currency is monopolized.17
The amount of reserve compensation needed under central banking to maintain a constant money supply as the demand for currency changes can be calculated using a simple formula. Let
Then, under central banking
whereas, under free banking
Under central banking the relation between the supply of base money and total money supply is a function of the relative demand for currency, whereas under free banking this is no longer the case. The reason for the difference is that under central banking the public’s use of central bank notes as currency competes with the banks’ use of them as reserves. Under free banking high-powered (outside) money is not usually used as currency.
The amount by which the supply of base money needs to be adjusted under central banking to offset a change in currency demand can be derived by assuming that Mf is equal to the demand for money. We wish, in this case, to have Mc = Mf. If r is the same in both systems, this requires that
Bc - Cp (1 - r) = Bf
Bc = Bf + Cp (1 - r).
Thus, for example, if Cp rises $1000, and r = .10, then the monetary base under central banking would, other things being equal, have to increase
($1000) - ($1000) × (.10) = $900.
Put generally, the base needs to be adjusted by the amount of the increase in the relative demand for currency (the shift from deposit demand to currency demand) minus this value multipled by the reserve ratio.
Notice that this procedure is not equivalent to one of maintaining constant the level of deposit-bank reserves. This is because deposit banks only need to hold reserves against that part of the money supply that consists of deposit balances; they do not need to hold reserves against currency in circulation because it is not part of their outstanding liabilities. If, in the above example, the initial level of reserves of the deposit banks was $10,000 then, after the growth in Cp and corresponding reserve compensation, the new, equilibrium level of reserves is $10,000 - $1000 + $900 = $9,900. It follows that a central banking policy of maintaining constant the level of bank reserves, which might be desirable if the relative demand for currency were unchanging, will not preserve monetary equilibrium when the relative demand for currency flutuates.
Instruments for Reserve Compensation
Having seen the precise adjustments in the monetary base needed to compensate changes in deposit bank reserves due to changes in the relative demand for currency, we can ask how such reserve compensation might actually be undertaken by a central bank. Let us assume still that the total demand for money (currency plus deposits) is unchanging, and that only its division between currency demand and deposit demand alters. To simplify the problem even further, let us assume as well that the relative demand for currency is known to the monetary authority. We thus put aside for the moment the greater part of the knowledge problem discussed in the last chapter, which has to do with how the monetary authority could know how much currency it ought to supply, to consider whether the authority can actually make desired adjustments. Our concern is to examine the efficacy of various instruments for reserve compensation—statutory reserve requirements, open-market operations, and rediscount policy—in accommodating a known currency demand.19
The first instrument we have to consider, statutory reserve requirements, highlights the significant distributional impact of certain approaches to reserve compensation: although a correct adjustment of statutory requirements preserves monetary equilibrium on the whole, the uneven distribution of changes in liquidity brings welfare losses or gains to particular banks.20 Since changes in the relative demand for currency do not affect all banks simultaneously or uniformly, an ideal policy would have to make continual adjustments in statutory reserve requirements, bank by bank. This poses an impossible administrative problem. It requires, moreover, that the central authority know, not just the total extent of the public’s shift into (or out of) currency, but also which banks are affected by the shift.
Second, for adjustments in statutory reserve requirements to be adequate to accommodate substantial shifts into currency, the supply of base money held up in “free” statutory reserves would have to be large: a reduction in statutory reserve needs frees up more base money for use in circulation, but this is of no avail if the total supply that can be released is less than the increase in demand. Finally, phasing out statutory reserve requirements is obviously impossible if they are needed for reserve compensation.
The last point is important since statutory reserve requirements are themselves a barrier to automatic adjustments in the supply of deposit money. The significance of this becomes apparent when the assumption that the total demand for money is unchanging is (momentarily) relaxed. Of course the monetary authorities, if they knew the extent of changes in total money demand, could make the necessary modifications in their reserve-requirement adjustments, but this would just add another layer of complexity to an already tremendous administrative and calculational burden.21
A second vehicle for reserve compensation is open-market operations. The fundamental problem with this instrument is also distributive. Although it allows direct control of the total amount of base money created or withdrawn, it does not provide any means of ensuring that base money issued goes to banks that are experiencing currency withdrawals or, alternatively, that base money withdrawn is withdrawn from banks experiencing redeposits of currency. Lauchlin Currie draws attention to this in his Supply and Control of Money in the United States (1934, 117):
If the reserve banks should buy bonds to the amount of the increase in cash in circulation, less the amount of the reserve formerly held against withdrawn deposits, it would appear that the composition of money has changed but not its volume. This would be true if the reserve bank funds, arising from the purchase of bonds, go to those banks . . . experiencing withdrawals.
But, Currie observes (ibid., 114), the banks receiving the new base money from open-market sales will probably be different from those stricken by currency withdrawals. Those banks receiving new base money that do not need it to offset reserve drains will employ it like any other increment of excess reserves, by increasing their loans and investments.22
Opposite consequences follow attempts to offset by open-market sales an inflow of currency due to a shift in demand from currency to deposits. “Here again,” Currie writes, “the difficulty is that the bank gaining reserves from the deposit of cash [currency] may not be the same bank losing reserves from the selling operations of the reserve banks” (ibid.). As with adjustments in statutory reserve requirements monetary equilibrium in the gross sense will be preserved, but with substantial welfare effects.
To some extent inter-bank lending might reduce these welfare effects from reserve compensation. But this possibility is limited by the fact that banks receiving excess base money will not necessarily lend it to other banks in need of reserve compensation: this may or may not be the most profitable avenue of employment for the surplus funds. Banks suffering reserve losses from currency drains might not offer to pay a high enough interest rate to attract emergency loans, for fear that the currency withdrawals may be permanent ones, which would make it difficult to repay the loans. Or, if banks losing reserves do offer to pay higher rates, other banks may still be reluctant to lend to them because they fear that the rates represent increased risk that the borrowing banks are suffering, not just temporary currency withdrawals, but a permanent loss of business.
A final instrument for reserve compensation is rediscount policy. This seems to offer the advantage of automatically channeling emergency supplies of base money only to banks in need of them, without requiring the monetary authorities to make decisions on a bank-by-bank basis. Murray Polakoff, who is generally critical of rediscount policy, writes (1963, 203) that it “is particularly well suited to supply a portion of reserves for seasonal needs and reserve losses and supplying them directly and immediately to the points where they are most needed.” He adds that “this is not true of open-market operations” (ibid.).23 A defect of rediscounting, however, is that it relies on deposit banks’ knowing whether currency is being withdrawn from them because of (a) an increase in their clients’ demand for currency or (b) dissaving (a fall in the demand for inside money).24 In general it is not possible for banks to know which of these causes is behind some withdrawal of currency by their depositors. Banks may mistakenly borrow base money from the central issuer (by rediscounting) to offset drains of the second type, forestalling the credit contraction needed in such cases to preserve monetary equilibrium. Distributing emergency base money by the rediscount mechanism does not guarantee that it goes to banks suffering from currency drains due solely to changes in the relative demand for currency.
All this assumes that banks, if they knew how, would borrow from the central issuer only the precise amount needed to compensate their losses caused by changes in the relative demand for currency. But the extent of borrowing depends on the rate of rediscount that the central bank charges. A rate below the market rate encourages borrowing, not merely for reserve compensation, but for acquiring excess reserves to relend at a profit. Furthermore, even if banks borrow from the central bank only to offset reserve losses due to currency withdrawals, the return of currency from circulation when the relative demand for it declines may not lead to offsetting repayment of borrowed reserves. If the rediscount rate is too low, the surplus base money will be re-lent instead. Winfield Riefler (1930, 161) cites an example of this in the United States just after World War I. Commercial banks had borrowed heavily from the Federal Reserve during the war to offset reserve losses due to an increased relative demand for currency. At the close of the war, when demand shifted back to deposit balances, returning Federal Reserve notes “were not used to repay member bank borrowings in any corresponding amounts.” Instead, redeposited currency “went in considerable part to build up member bank reserve balances”:
Member banks as a group . . . were content to maintain their indebtedness [to the Federal Reserve banks] at about the level it had previously attained, using funds released from circulation . . . to expand their loans, for which there was an active demand at attractive rates.
The resulting expansion of the money supply undoubtedly contributed to the boom-bust cycle of 1920-1921.
An above-market rediscount rate, a “penalty” rate, also does not ensure proper borrowing for reserve compensation. A penalty charge for funds borrowed to be kept in reserve leads to a less than optimal amount of reserve compensation, since a bank that pays penalty rates for its reserves is not, at the margin, better off than one that contracts its liabilities to make do with reserves it already has. Therefore a penalty rediscount rate is likely to lead to insufficient reserve compensation at times of expanded relative demand for currency. This conclusion applies with greatest force when increases in the relative demand for currency are expected to be long lasting or permanent.25
All of this assumes that a penalty rate or below-market rate of interest can at least be identified. In truth the variety of interest-earning assets available to banks, all with somewhat different nominal rates of interest, makes it difficult to choose a measure for “the market rate” against which the rediscount rate may be compared. As Polakoff notes (1963, 192), the existence of a distinct market rate is a peculiarity of English banking not present in other systems:
In Great Britain, it is the bill dealers and not the commercial banks that borrow directly from the Bank of England. Since the former specialize in a particular kind of asset—formerly commercial bills and now Treasury bills—and since the Bank rate is higher than the rate on bills, the discount rate in that country truly can be considered to be a penalty rate when dealers are forced to seek accommodation at the central bank.
Finally, even where some reasonable rule for setting it does exist, the rediscount rate has to be continually adjusted to reflect changes in the market rate.26
History offers many episodes of banks failing to respond to changes in the relative demand for currency. Most have been due to the failure of monopoly banks of issue to supply deposit banks with supplementary reserve media so that the deposit banks could withstand their depositors’ temporary withdrawals of currency. A few examples will help to illustrate points made in the previous section.
We have already noted an episode, which occurred in the United States in 1919, where an uncompensated shift of money demand from currency to deposits resulted in an excess supply of inside money taken as a whole. More notable and frequent, however, have been cases in which the supply of inside money has been allowed to contract excessively due to insufficient issues of currency to accommodate depositor withdrawals. In London, for instance, the Bank of England has been the sole supplier of currency since it was established in 1694. Other London banks rely upon their reserves of Bank of England notes to supply depositors’ currency needs. Through most of the first one and a half centuries of its existence the Bank of England felt no obligation to assist other bankers when they found themselves stripped of cash by a shift of demand from deposits to currency. Partly in consequence of this a series of financial crises occurred in 1763, 1772, 1783, 1793, 1797, 1826, 1836, and 1839. Every one was marked by a significant increase in the demand for currency for making payments in and around London: confidence in Bank of England notes was not lacking, and there were few demands to redeem these in specie. Nor was there any evidence of a rush to redeem country bank notes or to exchange them en masse for Bank of England notes. The problem was that country bank notes were not suitable for use in London where their issue and redemption was prohibited. A drop in the acceptability of checks and other noncurrency means of payment therefore translated entirely into greater requests for the notes of London’s sole issuing bank.
Henry Thornton (1802, 113), referring to the crisis of 1797, observed that “the distress arising in London . . . was a distress for notes of the Bank of England”:
So great was the demand for notes, that interest of money, for a few days before the suspension of payments of the bank, may be estimated . . . to have been about sixteen or seventeen per cent. per ann.
If other London banks had been allowed to issue notes the pressure might have been significantly reduced, since customers might simply have converted their deposits into notes which were also liabilities for the banks making the conversion. Then Bank of England notes would not have occupied a privileged position in bank portfolios; they would have been routinely returned to their issuer for redemption like other competitively issued liabilities. The public, in turn, would have had no special reason to demand Bank of England notes, since notes of other banks would probably have been just as useful for making payments around the city of London. As matters stood, however, the extraordinary demand for currency in London could only result in an extraordinary demand for Bank of England notes. The directors of the Bank of England were, however, concerned only with keeping it solvent; they did not manage its issues to protect other London banks or to prevent a general contraction of credit. Instead, observing the prevailing state of panic and confusion, and fearing that bank closings would generate a general loss of confidence which would threaten the Bank’s (specie) reserves, they actually contracted its issues, making matters even worse. This action was perhaps not calculated even to serve the interests of the Bank of England, but then the extent of that Bank’s involvement in the monetary affairs of the rest of the country was not fully appreciated. Indeed, although changes in the relative demand for currency were a frequent cause of what later became known as “internal drains” upon the resources of the London banks, Hayek observes in his introduction (p. 39) to Thornton’s Paper Credit of Great Britain that “it took some years . . . for the Bank of England to learn that the way to meet such an internal drain was to grant credits liberally.”
The Bank Act of 1844, although it restricted the ability of the Bank of England to generate excessive quantities of base money (as the Bank had, according to the Bullion committee, in the years following the suspension of 1797), also prevented it from making needed adjustments to the supply of currency in response to greater demand. Furthermore, by limiting the note issues of the country banks the Act caused them to employ Bank of England notes to meet depositors’ demands where before they might have been able to rely exclusively upon their own issues.
Thus after 1844 episodes of credit stringency were as frequent as before, with interest rates fluctuating in response to the periodic ebb and flow of the relative demand for currency. Rates rose every autumn—when currency was used instead of checkbook money for agricultural transactions—and also at the close of every quarter when stock dividends were paid (often in cash). Jevons was so impressed by this pattern that he devoted a lengthy article to an analysis of it (1884, 160-93). He observed the growing tendency of the London and country banks “to use the Bank of England as a bank of support, and of last resort” (ibid., 170-71).27 He remarked, in addition, that freedom of note issue along the lines of the Scottish banks was an inviting alternative means for English banks to accommodate their clients’ demands for currency, especially since additional currency issued this way would “return spontaneously as the seasons go round” (ibid., 179).28 In spite of his observations, however, Jevons did not recommend that England adopt free banking; on the contrary, he ended his article by defending the Bank of England’s quasi-monopoly of note issue, suggesting incoherently that proponents of free banking were guilty of “confusing” free banking with freedom of trade (ibid., 181).
In many ways Jevons’s opinions, except for his opposition to free banking, anticipated29 those of Walter Bagehot who, in Lombard Street (1874, 235-53) drew so much attention to the “lender of last resort” function that it came to be regarded as an official responsibility of the Bank of England and as a rationale for centralizig reserves and note—issue. Because of his influence Bagehot is sometimes viewed as the first champion of scientific central banking. Yet the truth is that Bagehot preferred in principle the “natural system” of competitive note issue—the kind of system that “would have sprung up if Government had let banking alone.” His formula for central banking was not a recommendation of monopolized note issue but an attempt to make an “anomalous,” monopolized system work tolerably well. Bagehot did not want to “propose a revolution” (1874, 67ff). Nor would he have seen any need for one—or for a lender of last resort—save for the fact that monopolization of note issue prevented banks other than the Bank of England from using their own notes to fulfill depositors’ demands for currency.30
Bagehot was aware of the true connection between monopolization of note issue and the need for a centrally directed monetary policy. Unfortunately, many of those who followed him, including later advocates of central banking, forgot it. Ralph Hawtrey wrote (1932, 285): “When a paper currency is an essential part of the monetary circulation and one bank possesses a monopoly of note issue, that bank can secure to itself the position of central bank. It can cut short the supply of currency and drive other banks to borrow directly or indirectly from it.” Precisely. Yet Hawtrey, who more than anyone saw the lender of last resort function as the principal rationale for central banking, did not see how central banks’ ability to cut short the supply of currency once they possess a monopoly of note issue creates the need for them to serve as lenders of last resort in the first place. If there is competitive note issue, the traditional argument for a lender of last resort carries much less weight.
The case of England is only the most notorious example of problems arising from a lack of currency under monopolized note issue. In Germany a law similar to the Bank Act was passed in 1875. It placed a ceiling on the note issues of the Imperial Bank, Germany’s monopoly bank of issue. According to Charles Conant (1905, 128) “high discount rates became the rule . . . as soon as the business of the country grew up to the limit of the note issue.” If banks other than the Imperial Bank had been free to issue notes this might have been avoided, because there would not have been any shortage of media to supply the growing demand for currency. There would not have been any great danger of inflation, either (as there was when the Imperial Bank took advantage of its monopoly privilege) because competitively-issued notes would not have served as base money. If issued in excess by any bank, notes would have been returned to their issuer for redemption. As it happened, interest rates in Germany did not return to normal until 1901, when “the limit of the ‘uncovered circulation’ [of Imperial Bank notes] was raised to conform to the increased needs for currency growing out of the expansion of business.”31
Another significant financial stringency caused by an uncompensated drain of currency from bank reserves was the “great contraction” in the United States from 1930 to 1932. This involved a large-scale movement from deposits to currency, which was only partly offset by Federal Reserve note issues. The result was a drastic decline in the total money stock followed by a terrible banking collapse.32
According to James Boughton and Elmus Wicker (1979, 406), this particular shift from deposits to currency was triggered by the “massive decline in income and interest rates” that began in the fall of 1929.33 That meant an increase in the relative frequency of small payments combined with a reduced opportunity cost of holding currency. Also encouraging the shift from deposits to currency were a 2 percent federal tax on checks (enacted in June 1932) and an increase from two to three cents in the postal rate for local letters (from July 1932 to June 1933), which increased the cost of paying local bills by check (ibid., 409). Finally, when state authorities began declaring bank holidays in response to insolvencies caused by currency withdrawals and loan losses, they unwittingly provoked even greater withdrawals of currency by depositors. When banks go on holiday, deposits are immobilized, and checks become practically useless in making payments. Currency can, however, still circulate while banks are temporarily closed. Therefore, any suspicion by the public that their banks will go on holiday will lead to a wholesale flight to currency as consumers rush to protect themselves against the risk of being stuck without any means for making purchases.34 The failure of the Federal authorities to provide adequate reserve compensation during this flight to currency contributed significantly to the severity of this phase of the Great Depression. It caused interest rates, which for a decade before were probably below their “natural” level, to suddenly rise substantially above it.
The American crises under the pre-Federal Reserve National Banking System were also aggravated—and in some cases perhaps caused—by restrictions on note issue by deposit banks. Here, however, the problem was not monopolization of the currency supply, since note issue was still decentralized. Instead, the currency supply was restricted by special bond-collateral requirements on National bank note issues.35 When eligible bond collateral was in short supply and commanded a premium, note issue became excessively costly. In consequence, banks sometimes met their depositors’ requests for currency by allowing them to withdraw greenbacks, a form of currency issued by the Treasury that also functioned as a reserve medium. The consequence was a contraction of total bank liabilities equal to a multiple of the lost reserves.36 That greenbacks were sometimes not available in desired, small denominations also added to the inconvenience suffered by the public.37
The problems of the pre-Federal Reserve National Banking System would have been much less severe had note issue been unrestricted, that is, had banks been able to issue notes on the same terms as they created demand deposits. Free note issue would have satisfied most customers’ currency requirements while leaving banks’ reserves in place. It also might have made it unnecessary to resort to an agency for reserve compensation such as finally emerged in the shape of the Federal Reserve System.38 As Friedman and Schwartz note in their Monetary History of the United States (1963, 295fn), the troubles of the National Banking System “resulted much less from the absence of elasticity of the total stock of money than from the absence of interconvertibility of deposits and currency.” To achieve the latter, free note issue would have been, not only adequate, but more reliable than centralized note issue.39
In all of these historical episodes undesirable changes in the total supply of money occurred as a result of changes—sometimes merely seasonal changes—in the relative demand for currency. Had it not been prohibited freedom of note issue would have gone far in eliminating this problem, and where note issue was relatively free, as it was in Scotland and Canada, the problem did not arise.40 Reliance upon a lender of last resort, on the other hand, does not get to the root of the problem, since it generally involves a monopolized currency supply which is also “inelastic” and which can be managed properly only with great difficulty, if at all.
The findings of this and the previous chapter contradict the claim that central banking is superior to free banking as a means for guaranteeing monetary equilibrium and general economic stability. But before any broad conclusion can be reached concerning the relative merits of free and central banking we must consider some other, potential shortcomings of unregulated banking. In particular we must consider the possibility that free banking may be unstable due to causes not dealt with in preceeding pages, along with the possibility that it may be inefficient. These issues are taken up in chapter 9.
Appendix: Reserve-Compensation Formulae
A. CENTRAL BANKING
The reserve ratio is defined as:
By definition, since Sp = O, (currency held by the public). Also, (since Sp and Sb = 0)
B. FREE BANKING
The reserve ratio is:
Since Bf (high-powered money) = Sp + Sb = S and Sp is assumed = 0, it follows that
By substitution, .
[* ] It is assumed that under central banking commercial banks convert all specie holdings into deposits or notes of the central bank.
[1.] Throughout this chapter it is assumed that the variance of bank clearings is not affected by a change in the form (notes or deposit-credits) of outstanding liabilities so long as the average holding time (turnover period) of the liabilities is the same. This is equivalent to assuming that precautionary reserve demand for outstanding note liabilities will be the same as for an equal amount of demand-deposit liabilities with the same average rate of turnover.
[2.] Under central banking with fiat money the distinction between currency demand and outside-money demand is blurred: there is no observable difference between the two, since the ultimate money of redemption is also the only currency in the system. Nevertheless it is still possible conceptually to distinguish the desire to acquire hand-to-hand media from the desire to withdraw savings from the banking system. Under central banking with a commodity standard, the former manifests itself in increased demand for the notes of the central bank, whereas the latter involves redemption of those notes for the money commodity.
[3.] According to Bowsher (1980, 11-17), the ratio of currency to demand deposits rose in part because of a fall in the importance of demand deposits relative to savings accounts. Nevertheless the trend is surprising in view of the development of alternatives to currency, such as credit cards, and of the substantial increase in interest rates which are a measure of the opportunity cost of holding cash. Many economists attribute this growth in demand for currency to the expansion of the “underground” economy.
[4.] Agger, p. 87. To consider only currency demand and not outside-money demand is not to neglect the usual consequences of a falling off in business confidence. Historically, when a general decline in confidence has led to increased outside-money demands it has been because of banks’ failure to meet depositors’ increased demands for currency through increased issues of inside money. For evidence of this see below. The problem of banking panics is dealt with later in this chapter.
[5.] Somers (1873, 204-25) writes with regard to conditions in 19th-century England that “when the situation is so bad that distrust or panic sets in it is the withdrawal of deposits [by their conversion into currency], and not the cashing in of notes, that gives the fatal blow to a tottering establishment.”
[6.] See Cagan (1958); also Agger (1918, 78-86), and Frank Brechling, (1958, 376-393). Brechling investigates fluctuations in the relative demand for currency in twelve countries and reports both significant across-country variation and significant short-run fluctuations within individual countries. He concludes that the assumption of a constant short-run cash preference ratio, which is determined predominantly by custom and institutional factors, is not supported by the empirical evidence and should be abandoned. Another study, also involving twelve countries, which reaches similar conclusions is Joachim Ahrensdorf and S. Kanesthasan (1960, 129-132).
[7.] Agger (1918, 87). The surrounding general discussion (pp. 76-90) is one of the best on this whole subject.
[8.] See ibid. 76; and also Francis Dunbar (1917, 17-18).
[9.] By assumption only the composition of money demand is varying; the depositor is not seeking to reduce his average money holdings or to take his business to some other bank. For present analytical purposes—and not necessarily because it is realistic to do so—it is desirable to abstract from these other possibilities. I have already shown how a free banking system would respond to them.
[10.] To repeat, the notes are not useful as a basis for credit expansion, so that their return to their issuer should not provoke any addition to loans.
[11.] Thus the equilibrium stockpile of notes on hand depends on (1) the probability density function over levels of currency demand; (2) the difference between the price of notes ordered in normal course and that of notes ordered on a “rush job” basis; and (3) the interest cost of paying the former price sooner. The setup is identical to the choice of optimal (base money) reserves and investments. An expansion of note issue will result in clearing debits—debits that, under free banking, have to be settled in outside money—unless it is consistent with the currency needs of the public.
[12.] “Generally speaking, an increase in the supply of money in the form of check-currency [deposits] must normally appear as part of a composite supply, in which other types of currency are represented; . . . the absence of these other types may effectually prevent the issue of check-currency itself.” See Arthur Marget (1926, 255).
[13.] According to Somers (1873, 207-8), “when [the unrestricted right of note issue] is stopped, and notes are only authorized from a central source, the facility a bank may enjoy in supplanting itself with currency for the uncertain demands upon it can only be in proportion to its proximity to the Issue Department.”
[14.] The amount of “reserve compensation” needed will be less than the actual increase in currency demand.
[15.] Thus McLeod writes (1984, 65-66fn) that a system of competing banks of issue (where no distinction is made between note and deposit liabilities as far as reserve needs are concerned) “has certain practical advantages if, as is usually the case, there are seasonal fluctuations in the public’s demand for notes relative to deposits. In [a system with monopolized currency supply] the peak seasonal demand for notes withdraws reserves from the banks and causes a seasonal credit stringency, and in a managed money system the central bank or other monetary authority must consciusly act to offset any such tendency.” The same is true concerning cyclical but nonseasonal changes in the relative demand for currency.
[16.] See for example Cagan (1958) and Milton Friedman (1959, 66-67).
[17.] Friedman revealed an awareness that the problem stems from monopolization of the currency supply when he noted (1959, 69) that it might be solved by allowing competition in note-issue. At the time, however, Friedman was less sympathetic (and, one might add, less understanding) toward free banking than he is today, and he described the solution of competitive note-issue as “the economic equivalent to counterfeiting.” Compare Friedman (1953, 220).
[18.] The formulae assume that commodity (outside) money does not circulate. Derivations appear in an appendix to this chapter.
[19.] By “known” I mean that the total quantity of currency demanded is known; I do not mean that the distribution of this demand—how changes in it will affect the reserve position of particular deposit banks—is known. To assume otherwise would be to grant too much in favor of the case for central banking. I have also chosen to deal only with the three more popular instruments of control. I leave it as an exercise to the reader to contemplate the practicability of other procedures not considered here.
[20.] Obviously these welfare changes affect not just the banks but also their borrower customers. In the event of a severe currency drain, depositors at some banks may also become victims of a restriction of payments.
[21.] For further comments on the shortcomings of statutory reserve requirements as instruments for monetary control see Friedman (1959, 45-50).
[22.] Compare Caroline Whitney (1934, 159-60).
[23.] For a general discussion of the disadvantages of rediscount policy as a means for monetary control see Joseph Aschheim (1961, 83-98).
[24.] This possibility does not violate the assumption of a fixed total demand for money so long as there is an equal increase in money demand elsewhere in the system.
[25.] See Currie (1934, 113).
[26.] See Friedman (1959, 40ff).
[27.] Particularly significant is Jevons’s finding that, in the course of the “autumnal drain,” coin and bank notes—including notes issued by “country” banks—moved together. This confirms the view that there was no rush for gold or Bank of England notes as such, but rather a rush for all types of currency. The pressure upon the Bank of England came when the other banks had exhausted their own authorized note issues.
[28.] Unfortunately Jevons believed as well that emergency currency supplied by the Bank of England would also be withdrawn from the system once it was no longer needed in circulation. This was incorrect. Bank of England notes might eventually return to those banks from which they were withdrawn (assuming no change in banks’ shares in the deposit business); but having come this far they went no further—they were retained as vault cash instead of being returned to the Bank of England for redemption and so their total supply would not fall to its original level. Instead, the notes were once again used as reserves to support further lending until the Bank of England made some conscious effort to contract their supply.
[29.] Jevons’s article first appeared in the Journal of the Statistical Society of London, vol. 29 (June 1866), pp. 235-53.
[30.] See the discussion of Bagehot’s views in Vera Smith (1936, 121ff) and in L. White (1984d, 145).
[31.] Conant (1905, 128). But compare Paul M. McGouldrick (1984, 311-49), who claims that the Reichsbank carried on a successful, countercyclical policy throughout most of the period in question.
[32.] For a general discussion of this episode, see Milton Friedman and Anna Jacobson Schwartz (1963, chapter 7).
[33.] See also James M. Boughton and Elmus R. Wicker (1984).
[34.] In view of this, it would have been better had state authorities declared a mere restriction of payments, prohibiting withdrawals of currency and coin, instead of outright holidays. This would have allowed checking-account transactions to continue, and would not have provoked as complete a flight to currency in neighboring states.
[35.] State bank note issues had ceased following a prohibitive 10 percent Federal tax on them in 1865.
[36.] See Friedman and Schwartz (1963, 169). Forced par collection, lack of branch facilities for convenient redemption, and the fact that bond-secured notes were perceived as being a lien on the Federal government rather than on their nominal issuers encouraged National banks to hold and reissue notes of their rivals instead of seeking actively to redeem them. Thus these notes were, unlike bank notes in an unregulated system, a kind of high-powered money. Their supply would not contract in response to any fall in the demand for currency, and their issue on more liberal terms (short of complete deregulation) might have led to serious inflation. This was, however, not a problem of practical concern in the latter part of the 19th century. On the downward-inelasticity of National bank notes see Charles F. Dunbar (1897, 14-22).
[37.] On this see Richard H. Timberlake, Jr. (1978, 124-31).
[38.] See Vera Smith (1936, 133-34).
[39.] Although many contemporary writers saw free note issue as a potential cure for the problems of the National banks, most believed that some agency was needed for supplying the system with emergency reserves. See for example Victor Morawetz (1909); Alexander Dana Noyes (1910); and John Perrin (1911). These writers, as well as O. M. W. Sprague (1910), tended to view reserve losses (and consequent monetary contraction) as a distinct problem rather than as a consequence of restrictions on note issue.
The evidence contradicts the view that media not backed by bonds or not centrally issued would have been unacceptable for supplying depositors’ demands during crises. For example, Canadian bank notes flowed readily into northern states to fill the void created by insufficient National bank note issues, even though Canadian notes were not backed by any special collateral. [See Joseph F. Johnson (1910, 118).] Also, clearinghouse certificates and loan certificates were issued in various places and were accepted even though they were of questionable legality. Finally, cashier’s checks and payroll checks of well-known firms were issued in small, round denominations to serve as currency. The only shortcoming of such emergency currency was that there was not enough of it. Nonetheless what there was showed every sign of being acceptable to the public, and there is every reason to think that freely issued bank notes would also have been accepted. On emergency currencies issued during the Panic of 1907 see A. Piatt Andrew (1908). On clearinghouse note issues see Richard H. Timberlake, Jr. (1984).
[40.] On the supply of currency in the (pre-1935) Canadian banking system see James Holladay (1934) and L. Carroll Root (1894).