Fractional Reserve Banking Explained: How Banks Create Money

At its simplest, a bank takes in deposits — money people and organizations leave with it for safekeeping, earning a small interest rate — and makes loans to borrowers who need money, charging them a higher interest rate. The difference between what the bank pays depositors and what it charges borrowers is called the spread; measured against all the assets the bank earns interest on, it is the net interest margin (NIM), and it is the core of how a bank makes money. If a bank earns $7 a year on every $100 it lends and pays $2 a year on every $100 of deposits, the $7 − $2 = $5 difference, before loan losses and running costs, is what keeps it in business (illustrative rates).

This sounds simple. In practice it is one of the most consequential functions in all of economics. When banks lend freely, businesses expand, homes are bought, infrastructure is built, and economies grow. When banks stop lending — as happened in 2008 — economic activity seizes up almost immediately, because modern economies depend on credit flowing continuously like blood through a body.

Behind that simple picture, a bank does three jobs at once. It runs the payment system its customers use, it judges who is likely to repay, and it performs maturity transformation: it funds long-term loans, such as a 30-year mortgage, with deposits its customers can withdraw at any moment. That mismatch is useful and it is also the source of bank runs (Section 3.9: Bank Regulation — and the SVB Case Study). One more fact surprises most people: a bank does not need your deposit before it can lend. Making a loan creates a new deposit, a mechanism Section 3.2: How Banks Create Money — Fractional Reserve Banking works through line by line.

Here is the part most people, and many textbooks, get backwards. When a bank lends you $20,000, it does not take $20,000 out of a vault or out of another customer’s account. It credits $20,000 to your deposit account and records a $20,000 loan as its own asset. Both sides of its balance sheet grow at the same moment, and money exists that did not exist a second earlier. The Bank of England put it plainly in its 2014 Quarterly Bulletin article “Money creation in the modern economy”: “Whenever a bank makes a loan, it simultaneously creates a matching deposit in the borrower’s bank account, thereby creating new money.” The balance in your checking account is, for the most part, money a commercial bank created, not money a central bank printed.

The old name of this chapter, fractional reserve banking, still describes one true thing: a bank holds only a fraction of its deposits as cash and as reserves (balances in its own account at the central bank). That is why no bank can survive every depositor asking for cash on the same morning, and why bank runs, deposit insurance and the lender of last resort exist (Section 3.9: Bank Regulation — and the SVB Case Study). What the old story gets wrong is the order of events. A bank does not wait for deposits and then lend a multiple of them. It lends first, and afterward finds the reserves it needs to settle its customers’ payments: from other banks overnight, from the repo market (Section 3.16: The Repo Market — The Financial System’s Overnight Plumbing), or from the central bank, which supplies reserves on demand at its policy rate.

So what stops a bank from lending without limit? Four things, none of them a reserve ratio:

• Capital and leverage rules. Every loan can go bad, so regulators require part of every loan to be funded by shareholders’ equity, which absorbs losses first (Section 3.13: The BIS and the Basel Framework — Global Banking Rules). A bank short of capital must stop growing its loan book, however many reserves it holds.
• Liquidity rules. Under Basel III, the Liquidity Coverage Ratio (LCR) requires a bank to hold enough high-quality liquid assets to cover its expected cash outflows over a 30-day stress, at a minimum of 100%, and the Net Stable Funding Ratio (NSFR) requires long-term assets to be funded by stable sources. A bank that lends aggressively loses deposits to other banks as its borrowers spend, and must replace that funding.
• Borrowers. A loan needs someone creditworthy who wants one. In a downturn, demand for credit falls and banks tighten their standards, so lending shrinks even when reserves are plentiful.
• The price of money. The central bank’s policy rate sets what banks pay for overnight funding and so the rates they charge. Higher rates make fewer loans profitable for banks and fewer loans attractive to borrowers (Section 2.10: The Federal Funds Rate: How the Fed Actually Sets the Price of Money).

Money is destroyed the same way it is made. When you repay a loan, the bank subtracts the payment from your deposit and reduces the loan by the same amount: both sides shrink and that money no longer exists. An economy whose credit is growing sees its money supply expand; a period of heavy debt repayment can shrink it.

The central bank creates the other kind of money: currency and reserves, together called base money. In quantitative easing (QE) it buys bonds, and when the seller is a non-bank, three balance sheets change at once. Say a pension fund sells $1 billion of Treasury bonds to the Fed: the Fed gains $1 billion of bonds and owes $1 billion of new reserves to the pension fund’s bank; the bank gains $1 billion of reserves and owes the pension fund a new $1 billion deposit; the pension fund swaps a bond for a deposit. Reserves and deposits rise together, by the same amount. Quantitative tightening (QT) runs the film backward: when the Fed lets a bond mature without replacing it, the Treasury refinances by selling a new bond to private investors, who pay out of bank deposits, so reserves and deposits fall together. The Fed ran QT from June 1, 2022, until December 1, 2025; on December 12, 2025, the New York Fed began reserve management purchases of Treasury bills to keep reserves at an “ample” level as demand for them grows.

Why did trillions of dollars of QE reserves not set off a lending explosion? Because reserves are not lent to households or firms: only banks (and a few other institutions) hold them, and they move between banks as payments settle. What anchors short-term rates in a system full of reserves is the rate the Fed pays on them, interest on reserve balances (IORB), which Congress authorized in 2006 and brought forward to October 1, 2008. A bank will not lend overnight to another bank for less than the Fed pays it, risk-free, to keep the money on account. The overnight reverse repurchase facility (ON RRP) does the same for money market funds, which cannot earn IORB. Section 2.10: The Federal Funds Rate: How the Fed Actually Sets the Price of Money shows how the two hold the federal funds rate inside its target range. The cost is real: at the current IORB of 3.90% on about $2,948 billion of reserves, $2,948 billion × 0.039 ≈ $115 billion a year if both stayed at that level for twelve months, an expense that reduces what the Fed would otherwise remit to the Treasury.