How Banks Create Money From Nothing

On Friday, March 10, 2023, the staff at Silicon Valley Bank watched their institution die in the space of a morning. The day before, depositors had tried to pull roughly $42 billion out of the bank, the largest single-day run in American history and the first to play out at the speed of a group chat. By the time the sun was high over the Pacific, the Federal Deposit Insurance Corporation had been appointed receiver and the bank was gone, before noon, before lunch, before most of the country had finished its first coffee. The lines that destroyed previous banks had been queues of anxious people standing outside marble lobbies. This time the queue was a Slack channel and a few thousand venture capitalists tapping at their phones.

What makes the story stranger is what was actually disappearing. The depositors were not pulling out stacks of bills that the bank had been guarding in a vault. The overwhelming majority of the money in those accounts had no physical form at all, and a great deal of it had been brought into existence by banks like Silicon Valley Bank in the ordinary course of lending. To understand why the run happened so fast, and why deposit insurance and central banks exist at all, you have to start with a fact that sounds like a conspiracy theory but is mainstream monetary economics: banks create most of the money supply, and they create it more or less out of nothing.

What a bank is really doing behind the counter

Strip away the branding and a commercial bank performs one core function. It takes in short-term deposits from savers, money that can be withdrawn on demand, and it uses those funds to make longer-term loans to borrowers, holding back only a fraction of the deposits in reserve to cover the everyday trickle of withdrawals. That single sentence hides almost everything interesting about banking, because it contains a tension that never fully resolves. The deposits are short-term and the loans are long-term, and the bank is sitting in the gap between them.

The intuitive picture most of us carry is that a bank is essentially a warehouse with a lending desk attached. Savers bring in their money, the bank stacks it somewhere safe, and when a borrower walks in, the bank reaches into the stack and hands some of it over. In this view, the bank is a passive middleman moving existing money from people who have it to people who want it. The deposits come first, and the loans are carved out of them.

That picture is wrong, or at least it gets the causation backward, and seeing exactly how it is wrong is the hinge on which modern monetary economics turns.

How a loan quietly conjures a deposit

Consider what happens when a bank approves a mortgage. The borrower does not leave with a sack of cash that the bank withdrew from someone else's account. Instead, the bank simply credits the borrower's account with the loan amount. It types a number into a ledger. On one side of the bank's balance sheet sits a new asset, the loan the borrower now owes, and on the other side sits a new liability, the deposit the borrower can now spend. The two are created in the same instant, with the same keystroke.

This is the part that trips people up. The deposit did not come from another saver. It did not exist a moment earlier. The act of lending created it. Each loan a bank makes brings into being a matching deposit, and that deposit is money in the fullest sense, spendable, transferable, and indistinguishable from any other balance in the account. When the borrower uses the loan to buy a house, the seller receives money that was, in effect, summoned into existence by the loan agreement.

In 2014, the Bank of England made this official in a way that surprised even some economists. Its quarterly bulletin stated plainly that banks do not merely lend out deposits that savers have placed with them; in the act of lending, banks create new deposits, and in doing so they create new money. The "loanable funds" framing taught in older textbooks, in which a fixed pool of savings is rationed out to borrowers, gets the direction of causation backward. Loans create deposits, not the other way around. The vast bulk of what we call money, the figures in checking and savings accounts across the economy, is not government-printed currency at all. It is bank-created deposit money, and it comes into being when banks make loans.

The chain reaction that multiplies a single deposit

None of this means banks can create money without limit. A constraint sits in the system, and it works through a chain that economists call fractional reserve banking. A bank is required, by regulation or prudence, to keep a fraction of its deposits in reserve rather than lending all of them out. That fraction is the brake.

Trace what happens to a single deposit. Suppose someone deposits $100 in a bank, and the reserve requirement is ten percent. The bank keeps $10 in reserve and lends out the remaining $90. The borrower spends that $90, and it lands as a deposit in some other bank, which keeps $9 in reserve and lends out $81. That $81 becomes a deposit elsewhere, which supports a further loan of about $73, and so the chain continues, each link a little smaller than the last as reserves are skimmed off at every stage.

Add up the entire sequence and something remarkable emerges. An initial $100 deposit can support up to $1,000 of deposits across the banking system as a whole, ten times the original sum. The relationship has a clean formula. The maximum expansion, known as the money multiplier, equals one divided by the reserve ratio. With a ten percent reserve requirement, the multiplier is one divided by 0.10, which is ten. Lower the reserve ratio and the chain runs further; raise it and the chain runs shorter. The reserve requirement, in other words, is the dial that sets how much money the banking system can manufacture from a given base.

It is worth being honest about the limits of this model. The textbook multiplier is a clean illustration rather than a precise description of how central banks operate today. Many modern central banks, including the Federal Reserve since 2020, have moved away from binding reserve requirements and instead steer the quantity of money through interest rates and the demand for loans. But the underlying insight survives intact. Lending expands the money supply, and something in the system, whether reserves, capital rules, or the price of borrowing, governs how far that expansion can go.

Borrowing short, lending long, and the danger built into the model

Return now to the tension buried in the bank's core function. The bank funds long-term loans, mortgages that run for decades and business loans that take years to repay, with short-term deposits that any saver can demand back at a moment's notice. This is called a maturity mismatch, and it is not an accident or a flaw to be engineered away. It is the business.

In calm periods, the mismatch is a license to print profit. Only a small fraction of depositors want their money on any given day, so the bank can comfortably lend out the rest at higher long-term rates while paying little or nothing on deposits. The gap between the two is the bank's bread and butter. But the same arrangement that is so comfortable in good times becomes lethal in bad ones. If enough depositors decide they want their money at once, the bank cannot oblige, not because it is dishonest or even insolvent, but because the money is tied up in loans that cannot be called back overnight. The assets are long, the liabilities are short, and a panic collapses the distinction.

Why a rational person joins the run

The truly unsettling feature of a bank run is that it can be perfectly rational for each individual even when the bank is fundamentally sound. Douglas Diamond and Philip Dybvig demonstrated this formally in a celebrated 1983 model, work that earned them a share of the 2022 Nobel Prize in Economics alongside Ben Bernanke.

Their insight runs like this. A bank holds enough good assets to pay everyone in full, eventually, but not enough cash to pay everyone today. As long as depositors trust that the bank is fine, only the usual handful withdraw, and the bank is fine. But if you come to believe that other depositors are about to run, then the smart move is to run first, because the bank pays out on a first-come basis and the cash runs out before the line ends. Your fear that others will run gives you a reason to run, and your running gives them a reason too. The expectation of a run produces the run. This is what makes bank runs self-fulfilling, and it is why a solvent bank can be destroyed by nothing more than a shift in mood. Silicon Valley Bank had real losses on its bond portfolio, but it was the coordinated rush for the exits, accelerated by venture capitalists messaging each other in real time, that finished it within hours.

The institutional answers that keep the system standing

If runs are rational and self-fulfilling, no individual bank can reassure its way out of one. The fix has to come from outside the bank, and the twentieth century built two of them.

The first is deposit insurance. The Banking Act of 1933, passed in the wreckage of the wave of bank failures that deepened the Great Depression, created the Federal Deposit Insurance Corporation. The FDIC insures deposits up to a cap, currently $250,000 per depositor, per bank. The logic is psychological as much as financial. If your money is guaranteed whatever happens, you have no reason to join a run, and if nobody runs, the run never starts. The insurance works precisely by making itself rarely necessary. Most developed economies adopted comparable schemes over the following decades.

Deposit insurance has a hard edge, though. It only covers balances up to the cap, and above that ceiling depositors still have every rational reason to flee. Silicon Valley Bank's customers were overwhelmingly tech companies and venture funds with balances far above $250,000, which is exactly why insurance alone could not stop their run. This is where the second institutional answer comes in. The central bank acts as a lender of last resort, standing ready to lend freely against good collateral to banks caught in a panic, so that a temporary shortage of cash does not cascade into a system-wide collapse. Over the SVB weekend in 2023, the federal government went further still, invoking a systemic-risk exception to guarantee even the uninsured deposits, an extraordinary measure meant to halt contagion before Monday's opening bell.

The pattern is older than any of these institutions. The Knickerbocker Trust panic of 1907 helped motivate the founding of the Federal Reserve. The bank failures of the early 1930s produced the FDIC. And the SVB collapse of 2023 revealed how violently fast a modern run can move once group chats replace the queues outside the branch doors. Each crisis rebuilt the safety net for a banking system that, by its very design, can never be entirely safe.

Key Takeaways

A commercial bank is not a warehouse that lends out money other people have deposited; it intermediates short-term deposits into long-term loans and, in the act of lending, creates new deposit money, which is why the Bank of England confirmed in 2014 that loans create deposits rather than the reverse and why bank-created money makes up most of the money supply. Fractional reserve banking lets a single deposit ripple into a far larger sum across the system, capped by the money multiplier of one divided by the reserve ratio, so a ten percent reserve can in principle support tenfold expansion. The price of this money creation is a permanent maturity mismatch, short liabilities funding long assets, which makes banks profitable in calm times and fragile in panicked ones, since Diamond and Dybvig's 1983 model (a 2022 Nobel) showed that runs can be rational and self-fulfilling even at a solvent bank. The institutional answers, deposit insurance up to $250,000 through the FDIC born in 1933, and the central bank's lender-of-last-resort function for everything above the cap, are what stand between an ordinary Friday and another Silicon Valley Bank.