Standard Open Market Operations: How the Fed and Commercial Banks "Create Money"
[This article is part of the Understanding Money Mechanics series, by Robert P. Murphy. The series will be published as a book in 2021.]
In this chapter we will define some of the conventional “monetary aggregates,” such as M1 and M2. Then we will summarize the textbook description of how the Federal Reserve and commercial banking system “create money” when the Fed buys assets and the commercial banks extend new loans.
Although the operations we describe in this chapter are somewhat simplistic, this type of baseline description is necessary for anyone who wants to understand how money is created (and destroyed) in modern economies. In chapter 7 we will discuss the new techniques that central banks have been using since the 2008 financial crisis, while in chapter 13 we will address critics who argue that the textbook approach given in this chapter doesn’t accurately reflect the causal relationship between bank reserves and new deposits.
Various Measures of “How Much Money” Is in the Economy
As we explained in chapter 2, a standard definition of money is that it’s a medium of exchange that is (nearly) universally accepted in trade among a given community of people. However, in practice there are different ways of applying this definition, because of the special economic nature of claims on money.
Think back to our discussion of the historical goldsmiths. In a town where everyone agrees that gold is the money, how should we treat a paper note issued by a reputable goldsmith that is an airtight and immediate redemption ticket entitling the bearer to a gold coin? If all of the merchants in town are just as willing to sell merchandise in exchange for these paper notes as they are for actual gold, then doesn’t that render the notes issued by the goldsmith a “universally accepted medium of exchange”? So if we’re trying to count up “how much money” is held by the townsfolk, shouldn’t we count the physical gold and the total number of paper notes issued by the reputable goldsmiths?
These are the complications that give rise to different monetary aggregates. The following list defines some of the most popular ones, with an application to the United States today.1
M0: The narrowest definition of money, M0 refers to the actual physical items, such as $20 bills and coins. (Note that some classifications consider M0 equivalent to the monetary base.)
Monetary Base: The monetary base includes paper currency and coins, as well as commercial banks’ (electronic) deposits at the Federal Reserve. Under current regulations, commercial banks in the US are required to keep some money “in reserve” in order to satisfy the demands of their customers who might show up to pull some cash out of their checking accounts. These “reserve requirements” can be satisfied by either literal paper currency in the banks’ vaults, or by commercial banks’ deposit balances with the Federal Reserve.
For example, suppose a particular bank had customers with total checking account balances of $1 billion. If the reserve requirement were 10 percent, then the bank would need to hold $100 million in reserves. It could satisfy this legal requirement if it held (say) $30 million in physical US currency in its own vaults on location and the Fed’s own computer system said that the bank had $70 million in its own account with the Fed.
M1: When going from the monetary base to M1, we need to be careful, because we don’t merely add another component, but also subtract two. Specifically, M1 consists of official US paper currency and coins held by the general public (but not in bank vaults, to avoid double counting), plus demand deposits and other checkable deposits (e.g., negotiable order of withdrawal (NOW) accounts), plus traveler’s checks issued by nonbank institutions. That means M1 does not include commercial bank reserves, whether they consist in notes and coins in the vault or electronic entries on the Fed’s books.
The intuition behind this classification is that M1 measures the amount of money and “very close money substitutes” held by the general public. A money substitute, as the name suggests, is an immediately redeemable claim on actual money that everyone in the market expects to be honored at par.
M2: Everything in M1, plus most savings account balances, so-called money market account balances, the balances on retail money market mutual funds, and small denomination time deposits (including bank certificates of deposit—CDs—of less than $100,000).
There are other popular aggregates, such as MZM (money of zero maturity) and M3, which of course is M2 plus some additional items that are claims on actual money that are not as “economically equivalent” to money as the components in the previous categories. (For example, the implied dollar value of certain repurchase agreements, or “repos,” is included in M3 but not in M1 or M2.) Fans of the Austrian school will be interested in the “true money supply” (TMS) aggregate developed by Murray Rothbard and Joseph Salerno, which corresponds to the Austrian theoretical definition of money.2
To avoid confusion, we should stress that moneyness is not the same thing as liquidity. If someone owns a $200,000 house and also has $200,000 in stocks, we would typically say that the stocks are more liquid than the house. What we mean is that the person can fairly quickly convert the stocks into $200,000 in actual currency (if so desired), whereas several months would be required to convert a house that’s “worth $200,000.”
Yet even though shares of corporate stock (especially those listed on major exchanges) are very liquid, we don’t include them in the definition of money. This is because a share of stock is a claim on ownership of the corporation, not a claim on a certain amount of dollars. The price of a share of stock—quoted in dollars—can fluctuate rapidly, meaning that your “$200,000 in stocks” could fall to zero depending on the news. In contrast, if you have traveler’s checks, those are claims denominated in dollars. They are not literally the same thing as money—if you’re trying to pay a cab driver, it’s better to have a $50 bill than a traveler’s check entitling you to a $50 bill—but traveler’s checks are nonetheless much better money-substitutes than shares of stock.
The following chart from the St. Louis Fed’s website displays the monetary base, M1, and M2 for the United States since 1984. The rapid expansion of the base and M1 following the financial crisis in 2008 is evident:
Source: St. Louis Federal Reserve. (monthly data, not seasonally adjusted, May 1984–December 2019)
How Commercial Banks “Create Money” under Fractional Reserves
As explained in chapter 3, our current monetary system is based on fiat money; there is nothing “backing up” the US dollar. The ability of the federal government/Federal Reserve to create new money simply by printing up green pieces of paper—or nowadays just through electronic activities that don’t even involve physical currency—might lead some people to believe that it is only in a fiat money system that this type of money creation “out of thin air” is possible. However, if they maintain less than 100 percent reserves on their checking accounts (demand deposits), commercial banks also have the ability to create money through their lending decisions.
To see how this works, let’s first imagine a town where the banks keep 100 percent reserves. Suppose there are 100,000 gold coins held by the townsfolk. Out of concerns for safety and convenience, the people deposit (say) 80,000 of the gold coins with the bankers, for which they receive paper notes entitling them to their 80,000 coins.
Now suppose the banks do not practice fractional reserve banking, but instead maintain 100 percent reserves. That is to say, for every paper banknote held by someone in the town, there is an actual gold coin in a bank vault to “back it.”
In this arrangement, notice that the public’s decision to hold some of their money in the form of banknotes rather than physical gold coins does not affect the total amount of money in the town. The townsfolk still hold 20,000 of the gold coins in their direct physical possession, and they also have 80,000 banknotes entitling them to gold coins. So, if each person reports how many gold coins he or she effectively has, their answers will sum up to 100,000 gold coins, which is the same amount they would have reported before using the banks.
Incidentally, we should point out that 100 percent reserve banking is possible, whether or not one thinks that it is desirable. Banks can charge a fee for the warehousing of their customers’ money, just as the owners of storage units manage to stay in business even though they don’t rent out their clients’ furniture. Furthermore, remember that we are here talking about demand deposits (think checking accounts), where the depositors believe they are entitled to obtain their money upon demand. If instead a customer buys (say) a one-year bank certificate of deposit (CD), the bank can lend that money out to a borrower even while practicing 100 percent reserve banking, because the CD is not a promise for immediate redemption.
But now suppose that the bankers in our hypothetical town don’t maintain 100 percent reserves but instead practice fractional reserve banking. The bankers realize that the public has come to trust the redeemability of the banknotes, and that most of the 80,000 in gold coins in their vaults will just sit there. Perhaps the bankers look at the history of transactions and conclude that so long as they always have enough gold coins in the vault to satisfy just 10 percent (say) of their total outstanding banknotes, they should be safe. In other words, the bankers reason that it would be very unlikely that the public would show up at the same time to demand more than 10 percent of the total paper notes that they’d issued.
In this case, the bankers see a great new way to earn income. Rather than “uselessly” keeping so many gold coins in their vaults, they lend some of the coins out to new borrowers. The borrowers then spend the money in the town, and the recipients in turn deposit the coins back into their own checking accounts at the banks. The process plays out until each gold coin sitting in a bank vault “backs up” ten paper banknotes held by people in the town. (See the endnote for links to methodical explanations of this process.3)
In this new scenario, in which the banks only keep 10 percent reserves, what happens to the “total amount of money” in our town? If we calculate M0, the answer is still the same: there are 100,000 gold coins in the town, period. Issuing paper notes and making loans doesn’t alter that fact.
However, if we use a broader aggregate such as M1, then the banks’ actions do affect the total. Specifically, there are 20,000 gold coins still held by the public, plus 800,000 banknotes held by the public, each entitling the holder to a gold coin. In other words, the public’s decision to keep 80,000 gold coins in the banks’ vaults, combined with the bankers’ decision to issue additional loans until the point at which they only held 10 percent reserves, caused M1 to grow from 100,000 gold coins to 820,000 gold coins. (Note that the actual unit of money would be something like a gold ounce rather than “gold coin.”)
We have deliberately worked with an example of commodity money—in our example, gold—in order to isolate the role played by fractional reserve banking. Because the broader monetary aggregates (M1, M2, etc.) include not just the base money but also very reliable and quick claims on it, the actions of banks can expand or contract the total amount of money when measured in the broader sense of these aggregates. In the modern United States, the base money is actual US dollars. But if someone has $100 in a checking account at Citibank, she really thinks she has $100, even though Citibank might only be holding (say) $10 in its vault (proportionate to each customer) to back her checking account balance.
How the Central Bank Can Affect the Total Quantity of Bank Reserves
For a community whose base money is hunks of gold, the reserves held in bank vaults would of course be determined by how much of the yellow metal had been mined (and fashioned into bars or coins). Yet today in the United States, because the underlying base money is the US dollar itself—meaning that it currently has no redemption option but is simply fiat money—the reserves held in bank vaults are green pieces of paper featuring US presidents. In addition, a commercial bank in the US can also satisfy its reserve requirements by having (electronic) balances on deposit with the Federal Reserve. Legally speaking, a commercial bank can itself hold a “checking account” with the Fed, and its deposit balance is “as good as” currency that the commercial bank holds in its own vaults.
Because of this situation, the Federal Reserve is able to affect the monetary base through its actions. Suppose that Fed officials want to adopt an “easier” policy that increases the quantity of money in the system and also (other things equal) tends to push down short-term interest rates. To accomplish these goals, the Fed can simply buy assets, writing checks on itself.
To give a specific example, suppose the Fed buys $10 million worth of Treasury bonds originally held by a dealer in the private sector. The Fed obtains the $10 million in bonds, adding them to its balance sheet. The seller of the bonds, in turn, receives payment in the form of a check written on the Federal Reserve. Legally, the Fed can’t “bounce a check”—there are no limits operationally on how much it can spend. When the dealer that sold the bonds deposits the check into its own bank account (at Citibank, say), the dealer’s checking account balance goes up, of course, by $10 million.
Now here is the important part of the story: Citibank passes along its customer’s deposited check to the Fed, which then credits Citibank’s account with the Fed by $10 million as well. At this initial stage, Citibank itself is just treading water; its liabilities have gone up by $10 million (because the bond dealer now thinks it has an extra $10 million in its checking account with Citibank), but its assets have also gone up by $10 million—represented by Citibank’s higher account balance with the Fed.
Yet look at what has happened. From Citibank’s perspective, a customer effectively just deposited $10 million in new base money that entered the financial system at the moment the Fed wrote the initial check. It is as if new gold coins had suddenly entered our hypothetical town from the earlier discussion and customers had deposited the new coins with the bankers. As we saw earlier, an influx of newly deposited base money sitting in the vaults of the commercial banks allows for new lending by the banks.
The same process happens here. Because Citibank’s reserves have gone up by $10 million, while its total outstanding customer deposit balances have also gone up by $10 million, it is now holding more than it needs to. Citibank can effectively lend out some of the newly deposited money, because it doesn’t need to hold the entire $10 million in new reserves to back up the $10 million in extra checking account funds now held by its customers.
If the commercial banks follow a 10 percent reserve rule and the system becomes “fully loaned up” after the Fed’s injection of $10 million, then the total increase in M1 will ultimately be $100 million. To sum up: the Fed’s decision to buy $10 million in bonds created $10 million in new (base) money, but then the banking system itself effectively creates $90 million in new (broader) money on top of it.
As before, we point interested readers to the endnotes for further reading that spells out this process more exhaustively. For our purposes here, there are two crucial takeaway messages:
- In our current fiat money system, the Federal Reserve creates new base money when it buys assets by writing checks on itself. Going the other way, the Federal Reserve destroys base money by selling assets (or by letting its assets mature and refraining from rolling over the proceeds). These actions do not require a literal printing press, as they can be achieved through electronic operations.
- When the Fed injects new base money into the system, it will often be deposited into commercial banks, where it will add to reserves. Under fractional reserve banking, the new reserves give the commercial banks the ability to pyramid new money (as measured by M1, M2, etc.) on the system through the process of granting new loans. Going the other way, when the commercial banks restrict their loan portfolios or the public withdraws base money from the banks, it causes the broader aggregates (M1, M2, etc.) to shrink.
- 1. The description of various monetary aggregates is a condensed version of this article by the same author: Robert P. Murphy, “The Definition of Various Monetary Aggregates,” Mises Wire, Sept. 1, 2016, https://mises.org/library/definition-various-monetary-aggregates.
- 2. Joe Salerno explains the “true money supply” aggregate, and compares it with other popular measures, in this 1987 article: Joseph T. Salerno, “The ‘True’ Money Supply: A Measure of the Supply of the Medium of Exchange in the U.S. Economy,” Austrian Economics Newsletter, Spring 1987, pp. 1–6, https://cdn.mises.org/aen6_4_1_0.pdf.
- 3. For a more methodical explanation (including balance sheet analysis) of fractional reserve banking and central bank open market operations, see Murray N. Rothbard, The Mystery of Banking, 2d ed. (Auburn, AL: Ludwig von Mises Institute, 2008), chaps. 7, 9, 10, and 11, https://cdn.mises.org/Mystery%20of%20Banking_2.pdf. (Note that Rothbard is hostile towards fractional reserve banking and central banking, but his explanation of how these processes actually work is still very helpful even to readers who do not share his attitude.) For a video presentation of similar material, see Robert P. Murphy, "The Theory of Central Banking," Mises Academy, lecture presented on Jan. 16, 2011, YouTube video, https://youtu.be/6HAEPSt_12U.