Murray N. Rothbard
Money in a Free Society
12. Money Warehouses
Suppose, then, that the free market has established gold as money (forgetting again about silver for the sake of simplicity). Even in the convenient shape of coins, gold is often cumbersome and awkward to carry and use directly in exchange. For larger transactions, it is awkward and expensive to transport several hundred pounds of gold. But the free market, ever ready to satisfy social needs, comes to the rescue. Gold, in the first place, must be stored somewhere, and just as specialization is most efficient in other lines of business, so it will be most efficient in the warehousing business. Certain firms, then, will be successful on the market in providing warehousing services. Some will be gold warehouses, and will store gold for its myriad owners. As in the case of all warehouses, the owner's right to the stored goods is established by a warehouse receipt which he receives in exchange for storing the goods. The receipt entitles the owner to claim his goods at any time he desires. this warehouse will earn profit no differently from any other--i.e., by charging a price for its storage services.
There is every reason to believe that gold warehouses, or money warehouses, will flourish on the free market in the same way that other warehouses will prosper. In fact, warehousing plays an even more important role in the case of money. For all other goods pass into consumption, and so must leave the warehouse after a while to be used up in production or consumption. But money, as we have seen, is mainly not "used" in 3&3 the physical sense; instead, it is used to exchange for other goods, and to lie in wait for such exchanges in the future. In short, money is not so much "used up" as simply transferred from one person to another.
In such a situation, convenience inevitably leads to transfer of the warehouse receipt instead of the physical gold itself. Suppose, for example, that Smith and Jones both store their gold in the same warehouse. Jones sells Smith an automobile for 100 gold ounces. They could go through the expensive process of Smith's redeeming his receipt, and moving their gold to Jones' office, with Jones turning right around and redepositing the gold again. But they will undoubtedly choose a far more convenient course: Smith simply gives Jones a warehouse receipt for 100 ounces of gold.
In this way, warehouse receipts for money come more and more to function as money substitutes. Fewer and fewer transactions move the actual gold; in more and more cases paper titles to the gold are used instead. As the market develops, there will be three limits on the advance of this substitution process. One is the extent that people us these money warehouses--called banks--instead of cash. Clearly, if Jones, for some reason, didn't like to use a bank, Smith would have to transport the actual gold. The second limit is the extent of the clientele of each bank. In other words, the more transactions take place between clients of different banks, the more gold will have to be transported. The more exchanges are made by clients of the same bank, the less need to transport the gold. If Jones and Smith were clients of different warehouses, Smith's bank (or Smith himself) would have to transport the gold to Jones' bank. Third, the clientele must have confidence in the trustworthiness of their banks. If they suddenly find out, for example, that the bank officials have had criminal records, the bank will likely lose its business in short order. In this respect, all warehouses--and all businesses resting on good will--are alike.
As banks grow and confidence in them develops, their clients may find it more convenient in many cases to waive their right to paper receipts--called bank notes--and, instead, to keep their titles as open book accounts. In the monetary realm, these have been called bank deposits. Instead of transferring paper receipts, the client has a book claim at the bank; he makes exchanges by writing an order to his warehouse to transfer a portion of this account to someone else. Thus, in our example, Smith will order the bank to transfer book title to his 100 gold ounces to Jones. This written order is called a check.
It should be clear that, economically, there is no difference whatever between a bank not and a bank deposit. Both are claims to ownership of stored gold; both are transferred similarly as money substitutes, and both have the identical three limits on their extent of use. The client can choose, according to this convenience, whether he wishes to keep his title in note, or deposit, form. 
Now, what has happened to their money supply as a result of all these operations? If paper notes or bank deposits are used as "money substitutes," does this mean that the effective money supply in the economy has increased even though the stock of gold has remained the same? Certainly not. For the money substitutes are simply warehouse receipts for actually-deposited gold. If Jones deposits 100 ounces of gold in his warehouse and gets a receipt for it, the receipt can be used on the market as money, but only as a convenient stand-in for the gold, not as an increment. The gold in the vault is then no longer a part of the effective money supply, but is held as a reserve for its receipt, to be claimed whenever desired by its owner. An increase or decrease in the use of substitutes, then, exerts no change on the money supply. Only the form of the supply is changed, not the total. Thus the money supply of a community may begin as ten million gold ounces. Then, six million may be deposited in banks, in return for gold notes, whereupon the effective supply will now be: four million ounces of gold, six million ounces of gold claims in paper notes. The total money supply has remained the same.
Curiously, many people have argued that it would be impossible for banks to make money if they were to operate on this "100 percent reserve" basis (gold always represented by its receipt). Yet, there is no real problem, any more than for any warehouse. Almost all warehouses keep all the goods for their owners (100 percent reserve) as a matter of course--in fact, it would be considered fraud or theft to do otherwise. Their profits are earned from service charges to their customers. The banks can charge for their services in the same way. If it is objected that customers will not pay the high service charges, this means that the banks' services are not in very great demand, and the use of their services will fall to the levels that consumers find worthwhile.
We come now to perhaps the thorniest problem facing the monetary economist: an evaluation of "fractional reserve banking." We must ask the question: would fractional reserve banking be permitted in a free market, or would it be proscribed as fraud? It is well-known that banks have rarely stayed on a "100%" basis very long. Since money can remain in the warehouse for a long period of time, the bank is tempted to use some of the money for its own account?tempted also because people do not ordinarily care whether the gold coins they receive back from the warehouse are the identical gold coins they deposited. The bank is tempted, then to use other people's money to earn a profit for itself.
If the banks lend out the gold directly, the receipts, of course, are now partially invalidated. There are now some receipts with no gold behind them; in short, the bank is effectively insolvent, since it cannot possibly meet its own obligations if called upon to do so. It cannot possibly hand over its customers' property, should they all so desire.
Generally, banks, instead of taking the gold directly, print uncovered or "pseudo" warehouse receipts, i.e., warehouse receipts for gold that is not and cannot be there. These are then loaned at a profit. Clearly, the economic effect is the same. More warehouse receipts are printed than gold exits in the vaults. What the bank has done is to issue gold warehouse receipts which represent nothing, but are supposed to represent 100% of their face value in gold. The pseudo-receipts pour forth on the trusting market in the same way as the true receipts, and thus add to the effective money supply of the country. In the above example, if the banks now issue two million ounces of false receipts, with no gold behind them, the money supply of the country will rise from ten to twelve million gold ounces--at least until the hocus-pocus has been discovered and corrected. There are now, in addition to four million ounces of gold held by the public, eight million ounces of money substitutes, only six million of which are covered by gold.
Issue of pseudo-receipts, like counterfeiting of coin, is an example of inflation, which will be studied further below. Inflation may be defined as any increase in the economy's supply of money not consisting of an increase in the stock of the money metal. Fractional reserve banks, therefore, are inherently inflationary institutions.
Defenders of banks reply as follows: the banks are simply functioning like other businesses?they take risks. Admittedly, if all the depositors presented their claims, the banks would be bankrupt, since outstanding receipts exceed gold in the vaults. But, banks simply take the chance--usually justified?that not everyone will ask for his gold. The great difference, however, between the "fractional reserve" bank and all other business is this: other businessmen use their own or borrowed capital in ventures, and if they borrow credit, they promise to pay at a future date, taking care to have enough money at hand on that date to meet their obligation. If Smith borrows 100 gold ounces for a year, he will arrange to have 100 gold ounces available on that future date. But the bank isn't borrowing from its depositors; it doesn't pledge to pay back gold at a certain date in the future. Instead, it pledges to pay the receipt in gold at any time, on demand. In short, the bank note or deposit is not an IOU, or debt; it is a warehouse receipt for other people's property. Further, when a businessman borrows or lends money, he does not add to the money supply. The loaned funds are saved funds, part of the existing money supply being transferred from saver to borrower. Bank issues, on the other hand, artificially increase the money supply since pseudo-receipts are injected into the market.
A bank, then, is not taking the usual business risk. It does not, like all businessmen, arrange the time pattern of its assets proportionately to the time pattern of liabilities, i.e., see to it that it will have enough money, on due dates, to pay its bills. Instead, most of its liabilities are instantaneous, but its assets are not.
The bank creates new money out of thin air, and does not, like everyone else, have to acquire money by producing and selling its services. In short, the bank is already and at all times bankrupt; but its bankruptcy is only revealed when customers get suspicious and precipitate "bank runs." No other business experiences a phenomenon like a "run." No other business can be plunged into bankruptcy overnight simply because its customers decide to repossess their own property. No other business creates fictitious new money, which will evaporate when truly gauged.
The dire economic effects of fractional bank money will be explored in the next chapter. Here we conclude that, morally, such banking would have no more right to exist in a truly free market than any other form of implicit theft. It is true that the note or deposit does not actually say on its face that the warehouse guarantees to keep a full backing of gold on hand at all times. But the bank does promise to redeem on demand, and so when it issues any fake receipts, it is already committing fraud, since it immediately becomes impossible for the bank to keep its pledge and redeem all of its notes and deposits.  Fraud, therefore, is immediately being committed when the act of issuing pseudo-receipts takes place. Which particular receipts are fraudulent can only be discovered after a run on the bank has occurred (since all the receipts look alike), and the late-coming claimants are left high and dry. 
If fraud is to be proscribed in a free society, then fractional reserve banking would have to meet the same fate.  Suppose, however, that fraud and fractional reserve banking are permitted, with the banks only required to fulfill their obligations to redeem in gold on demand. Any failure to do so would mean instant bankruptcy. Such a system has come to be known as "free banking." Would there then be a heavy fraudulent issue of money substitutes, with resulting artificial creation of new money? Many people have assumed so, and believed that "wildcat banking" would then simply inflate the money supply astronomically. But, on the contrary, "free banking" would lead to a far "harder" monetary system than we have today.
The banks would be checked by the same three limits that we noted above, and checked rather rigorously. In the first place, each bank's expansion will be limited by a loss of gold to another bank. For a bank can only expand money within the limits of its own clientele. Suppose, for example, that Bank A, with 10,000 ounces of gold deposited, now issues 2000 ounces of false warehouse receipts to gold, and lends them to various enterprises, or invests them in securities. The borrower, or former holder of securities, will spend the new money on various goods and services. Eventually, the money going the rounds will reach an owner who is a client of another bank, B.
At that point, Bank B will call upon Bank A to redeem its receipt in gold, so that the gold can be transferred to Bank B's vaults. Clearly, the wider the extent of each bank's clientele, and the more the clients trade with one another, the more scope there is for each bank to expand its credit and money supply. For if the bank's clientele is narrow, then soon after its issue of created money, it will be called upon to redeem--and, as we have seen, it doesn't have the wherewithal to redeem more than a fraction of its obligations. To avoid the threat of bankruptcy from this quarter, then, the narrower the scope of a bank's clientele, the greater the fraction of gold it must keep in reserve, and the less it can expand. If there is one bank in each country, there will be far more scope for expansion than if there is one bank for every two persons in the community. Other things being equal, then, the more banks there are, and the tinier their size, the "harder"--and better--the monetary supply will be. Similarly, a bank's clientele will also be limited by those who don't use a bank at all. The more people use actual gold instead of bank money, the less room there is for bank inflation.
Suppose, however, that the banks form a cartel, and agree to pay out each other's receipts, and not call for redemption. And suppose further that bank money is in universal use. Are there any limits left on bank expansion? Yes, there remains the check of client confidence in the banks. As bank credit and the money supply expand further and further, more and more clients will get worried over the lowering of the reserve fraction. And, in a truly free society, those who know the truth about the real insolvency of the banking system will be able to form Anti-Bank Leagues to urge clients to get their money out before it is too late. In short, leagues to urge bank runs, or the threat of their formation, will be able to stop and reverse the monetary expansion.
None of this discussion is meant to impugn the general practice of credit, which has an important and vital function on the free market. In a credit transaction, the possessor of money (a good useful in the present) exchanges it for an IOU payable at some future date (the IOU being a "future good") and the interest charge reflects the higher valuation of present goods over future goods on the market. But bank notes or deposits are not credit; they are warehouse receipts, instantaneous claims to cash (e.g., gold) in the bank vaults. The debtor makes sure that he pays his debt when payment becomes due; the fractional reserve banker can never pay more than a small fraction of his outstanding liabilities.
We turn, in the next chapter, to a study of the various forms of governmental interference in the monetary system--most of them designed, not to repress fraudulent issue, but on the contrary, to remove these and other natural checks on inflation.
A third form of money-substitute will be token coins for very small change. These are, in effect, equivalent to bank notes, but "printed" on base metal rather than on paper.
See Amasa Walker, The Science of Wealth, 3rd Ed.(Boston: Little, Brown, and Co., 1867) pp. 139-41; and pp. 126-232 for an excellent discussion of the problems of a fractional-reserve money.
Perhaps a libertarian system would consider "general warrant deposits" (which allow the warehouse to return any homogeneous good to the depositor) as "specific warrant deposits," which, like bills of lading, pawn tickets, dock warrants, etc., establish ownership to certain specific earmarked objects. For, in the case of a general deposit warrant, the warehouse is tempted to treat the goods as its own property, instead of being the property of its customers. This is precisely what the banks have been doing. See Jevons, op. cit., pp. 207-12.
Fraud is implicit theft, since it means that a contract has not been completed after the value has been received. In short, if A sells B a box labeled "corn flakes" and it turns out to be straw upon opening, A's fraud is really theft of B's property. Similarly, the issue of warehouse receipts for non-existent goods, identical with genuine receipts, is fraud upon those who possess claims to non-existent property.