The bank that banks bank at
You have just watched ordinary banks pull off something close to magic: holding only a slice of your deposit in reserve and lending out the rest, the banking system as a whole performs money creation, so that most of the money supply is not printed cash at all but numbers in accounts, born the moment a loan is made. That raises an obvious question the earlier guides left dangling. Your bank promises *you* instant cash; but where does *your bank* keep its own money, and who promises *it* cash when it needs some? The answer sits one level up: the [[central-bank|central bank]] — the Federal Reserve in the US, the European Central Bank, the Bank of England, the People's Bank of China.
The cleanest way to picture it: the central bank is a bank, but its customers are not people — they are the commercial banks themselves, plus the government. Each commercial bank holds an account at the central bank, and the balance in that account is its reserves. When the earlier guides talked about a bank's reserves, this is largely where they live: not as banknotes in a vault, but as a deposit at the central bank, as electronic and abstract as the balance in your own checking account. The central bank is, quite literally, the banker to the banks.
This gives banks a second, hidden settlement system running beneath the one you see. When you pay a friend whose account is at a different bank, no truck of cash drives between them. Instead your bank's reserve account at the central bank is debited and your friend's bank's account is credited — the central bank clears the payment by shuffling reserves. Trillions move this way every day, invisibly, on the central bank's books. Reserves, in other words, are the money that *banks* use to pay *each other*.
The money only the central bank can make
Here is the crucial split. Commercial banks can create the *deposit* kind of money by lending — you saw that — but there is one kind they cannot conjure: central-bank money. It comes in exactly two forms: the physical banknotes and coins in your wallet, and the reserve balances banks hold at the central bank. Add those two together and you get the [[monetary-base|monetary base]], also called M0 or 'high-powered money'. It is the bedrock the whole pyramid of bank-created money is built upon, and only the central bank can issue it.
How does the central bank create base money? In the modern world, simply by typing. When it wants more reserves in the system, it buys an asset — typically a government bond — from a bank and pays by crediting that bank's reserve account with newly created reserves. No printing press, no gold dug up: the reserves are willed into existence with a keystroke, in the same way a commercial bank's loan willed your deposit into existence. The reverse also holds: when the central bank sells bonds back, it deletes the reserves it receives. These purchases and sales are called [[open-market-operations|open-market operations]], and they are the main tap by which the central bank turns the monetary base up or down.
THE MONEY PYRAMID (illustrative scale) Broad money (M2) ~ $1,000 <- mostly bank deposits, /-------------------------\ created by lending / \ / Monetary base (M0) ~ $150 \ <- ONLY the central bank = cash in wallets can make this layer + reserves at central bank Banks build the wide top from the narrow base. Numbers are illustrative, not any one country's.
Why a fraction makes runs possible
Now we can see exactly why the very design that makes banks useful also makes them fragile. Under [[fractional-reserve-banking|fractional-reserve banking]], a bank promises every depositor instant access to cash, yet it holds only a fraction of those promises as actual reserves; the rest is tied up in loans that will not come back for years. On a normal day this is fine — withdrawals and deposits roughly cancel, and the thin reserve cushion covers the trickle. But the promise of *everyone* getting cash *at once* is one the bank simply cannot keep. The cash isn't there; it is out working as somebody's mortgage and somebody else's business loan.
That is the seed of a [[bank-run-and-deposit-insurance|bank run]], and its logic is a vicious coordination game. Suppose a rumor spreads that your bank is shaky. You might believe the rumor is nonsense — but you also know that if *enough other people* believe it and rush to withdraw, the thin reserves will run dry, and whoever arrives last gets nothing. So the rational move, even for a skeptic, is to get in line *first*. Everyone reasoning the same way produces the stampede. The cruel twist: a perfectly healthy bank, every loan sound, can be toppled purely because people *expected* it to fail and acted on the expectation. The fear manufactures the very collapse it feared — economists call this a self-fulfilling equilibrium.
Two firewalls: deposit insurance and the lender of last resort
If a run is driven by the fear that others will run, the cure is to remove the fear. The first firewall does exactly that: deposit insurance. A government agency guarantees each depositor's money up to a limit — say 250,000 dollars in the US, or 100,000 euros across the EU. Once you know your savings are safe whatever happens, you have no reason to join the stampede. And here is the beautiful part: because nobody runs, the insurance rarely has to pay out at all. The guarantee works mostly by being believed — a promise that, by removing the incentive to act on fear, stops the feared event from ever occurring.
The second firewall is the one only a central bank can man: the lender of last resort. When a fundamentally sound bank is hit by a sudden cash crunch, the central bank lends it reserves overnight — against good collateral — so it never has to dump its long-term loans at fire-sale prices just to find cash today. The classic rule, stated by Walter Bagehot back in 1873, still guides central bankers: in a panic, lend *freely*, to *solvent* firms, against *good collateral*, at a *penalty rate*. The penalty matters — it ensures banks treat this as emergency help, not a cheap subsidy, and reach for it only when truly cornered.
Be honest, though: these firewalls are not free. By promising to backstop banks, the system invites a danger called [[eco-moral-hazard|moral hazard]] — if a bank knows it will be rescued, it has a quiet incentive to take wilder risks, pocketing the gains in good times and handing the losses to the public in bad. Worse, banks judged 'too big to fail' may count on rescue with near-certainty. This is why backstops always come bundled with regulation — capital requirements, stress tests, supervision — meant to stop banks from abusing the safety net. Economists genuinely disagree about how to strike this balance, and every crisis reopens the argument; pretending the trade-off away would be dishonest.
From guardian to steering wheel: the bridge to monetary policy
Keeping the banking system from collapsing is only half of what a central bank does. The other half is steering the whole economy — and the same levers reappear, now used for a different purpose. A central bank typically operates under a [[central-bank-mandate|mandate]]: a legal job description, most often to keep prices stable (low, steady inflation) and, in some countries, to support full employment too. To pursue that mandate it leans on the very tools we just met as crisis cures.
Here is the bridge. By adjusting the supply of reserves through open-market operations, the central bank nudges the interest rate banks charge each other to borrow reserves overnight. That single rate then ripples outward — into the rates on your mortgage, a firm's expansion loan, a saver's deposit — gently encouraging borrowing and spending when set low, restraining them when set high. Steering the economy by working this lever is [[monetary-policy|monetary policy]], and it is exactly where the next rung of this ladder begins. The guardian of the banks, it turns out, is also the hand on the economy's throttle.