Money & banking,
from zero to fluent.

Imagine an economy with no money. To eat, the baker who wants shoes must find a shoemaker who happens to want bread, right now, in the right amounts. Economists call this the double coincidence of wants, and it is the wall every barter economy runs into. Trade only happens when two people's needs line up exactly, which is rare, so most gains from trade simply never occur.

There is a second, quieter cost, and it is countable. In a barter economy, every pair of goods needs its own exchange rate. With n goods there are n(n-1)/2 pairwise prices to know and post. Introduce one money that everything trades against, and the count collapses to n prices, one per good. That drop is the whole reason money exists: it is a coordination device that shrinks the information every trader must carry.

Watch the two structures diverge. Add goods to the market and the barter web (every good linked to every other) explodes, while the money version (every good linked to one hub) grows in a straight line.

An artist offers a painting to fix her car, and the mechanic takes it: for that one trade, the painting worked as a medium of exchange. But try pricing a week of groceries in fractions of a painting (a useless unit of account), or parking your retirement savings in a canvas that can be stolen, damaged, or simply fall out of fashion (a poor store of value). A good money has to clear all three bars, not just one.

Anything can be money if it performs three jobs at once. Economists test candidates against these functions, not against what they are made of.

• Medium of exchange: people accept it in trade, so you never need a double coincidence of wants. This is the job barter fails.
• Unit of account: prices, debts, and books are all quoted in it, so values are comparable at a glance.
• Store of value: it holds purchasing power over time, so you can sell today and buy later.

The store-of-value job is the fragile one, because it depends on inflation. Cash under a mattress is a perfect medium of exchange and a fine unit of account, but as a store of value it quietly leaks. If prices rise at inflation rate π per year, then a dollar buys 1/(1+π) as much next year, and its purchasing power decays geometrically.

That decay compounds, so the damage is far larger over a lifetime than the annual number suggests. A useful shortcut: money loses half its value in about ln(2)/ln(1+π) years, close to 70/π for small rates. Set an inflation rate and sweep the years to watch $100 of stored cash erode, and read off its half-life.

Think about your own accounts. Cash in your wallet is spendable this instant. Money in checking is spendable almost as fast, a swipe or a transfer away. Money in savings takes an extra step, a transfer or a few days to settle. Economists sort an entire economy's money the same way, into layers by how fast it can be spent, and give each layer a name.

Modern money is fiat: it has value because a government declares it legal tender and, crucially, because everyone else accepts it. It is not backed by gold. That makes "how much money exists?" a real question with several answers, because money comes in layers of decreasing liquidity. Central banks report these as monetary aggregates.

Two subtleties trip up beginners. First, bank reserves are in M0 but not in M1 or M2, because you cannot spend reserves at a shop; only the bank and central bank touch them. Second, physical currency in your pocket counts in every aggregate. So the layers nest, M0 and M1 overlap only on currency, and most of what people call "money" (M2) is not issued by the central bank at all. It is created by commercial banks, which is the subject of the next part.

Build the aggregates from their components. Notice that adding reserves lifts M0 but leaves M1 and M2 untouched, and adding savings lifts only M2.

Say you buy a $1000 rental property using $900 borrowed and $100 of your own savings. Your $100 is the cushion: the first money to vanish if the property loses value. Only once losses pass $100 does your lender lose a dime. A bank runs the same trade at a much larger scale, and that cushion has a name.

To see how banks make money you first have to read one, and a bank is just a balance sheet. Everything it owns is an asset; everything it owes is a liability; the sliver left over belongs to the owners and is called capital or equity. The three always satisfy one identity.

For a simple bank the assets are the reserves it keeps liquid plus the loans it has made (loans are assets: borrowers owe the bank). The main liability is deposits, because a deposit is money the bank owes you on demand. Capital is whatever is left. Here is the thing that makes banking dangerous: capital is thin. A bank with $1000 of assets and $900 of deposits has only $100 of capital, a 10% cushion. If its loans lose more than 10% of their value, assets fall below deposits, capital goes negative, and the bank is insolvent: it cannot possibly repay everyone even if every borrower pays in full.

Set the bank up, then hit its loan book with losses. Watch the capital cushion (amber) shrink and, past the break point, flip red.

Here is the move that surprises almost everyone: banks create money when they lend. They do not lend out a pile of pre-existing cash. When a bank grants a loan, it credits the borrower's deposit account with new money that did not exist a moment earlier. Because deposits are money (they are in M1), the act of lending expands the money supply. Banks keep only a fraction of deposits as reserves and lend the rest; this is fractional-reserve banking.

Follow a single fresh $1000 through the system. Each round, a bank keeps rr and lends (1-rr); the loan is redeposited and the next bank does the same. New deposits in round n are 1000×(1-rr)n, a shrinking geometric series. Summed to the end, the total deposits created are exactly 1000 / rr, and the ratio of total money to the original injection is the money multiplier, 1/rr. Step the rounds and watch the cumulative deposits climb toward the amber ceiling, which is the honest limit of the series, not a guess.

The clean 1/rr multiplier is a ceiling, not a forecast. Two real-world leakages drain the process at every round, so actual money creation falls short of the textbook figure.

• Currency drain (c): people keep some cash in their pockets instead of redepositing it. Cash that leaves the banking system cannot be lent again.
• Excess reserves (e): banks may hold more reserves than required, especially when lending looks risky or when the central bank pays interest on reserves. Idle reserves are not lent.

Add both and the multiplier becomes the honest formula below, where c is the public's currency-to-deposit ratio and e is banks' excess-reserve ratio. Every extra leak enlarges the denominator and shrinks the multiplier.

Compare the textbook multiplier against the realistic one. Start with no leaks and they match; add a currency drain or excess reserves and the real multiplier drops well below 1/rr. After 2008, huge excess reserves pushed the effective multiplier close to 1, which is exactly why flooding banks with reserves did not ignite runaway inflation.

A central bank cannot force banks to lend, but it controls the raw material: reserves. Its main day-to-day tool is open-market operations, buying and selling government bonds. When the central bank buys a bond from a bank, it pays by crediting the bank's reserve account with newly created reserves. Bonds move to the central bank; reserves appear in the banking system. Selling does the reverse and drains reserves.

Because broad money is roughly the multiplier times the base, a reserve injection of ΔB can support up to m×ΔB of new money. Slide the operation from a sale (drain) to a purchase (inject) and watch reserves and the supported money supply move together. This is the pump; the next chapter is the thermostat.

Every night, banks square their books. Some end the day with spare cash sitting idle; others come up short. Rather than let it sit or scramble, they lend to each other overnight to close the gap, and the interest rate on those loans is the single number the central bank cares about most.

Modern central banks do not target a quantity of money; they target a price: the overnight interest rate at which banks lend reserves to each other. They hit it through the reserve market. Banks' demand for reserves slopes down: the cheaper reserves are, the more banks want to hold. Two policy rates fence that demand in.

• The discount rate is a ceiling: no bank borrows in the market above what the central bank charges directly.
• Interest on reserves (IOR) is a floor: no bank lends below what the central bank pays it to just hold reserves.

The overnight rate settles where reserve supply (a vertical line the central bank sets) crosses the fenced demand curve. This produces two regimes. In a corridor system, reserves are scarce and supply crosses the sloping middle, so the rate is sensitive: small reserve changes move it. In a floor system, reserves are abundant and supply crosses the flat IOR floor, so the rate sits at IOR no matter how many more reserves are added. The floor system is how most major central banks operate today: they set the rate by decree (IOR) and let the balance sheet be whatever it needs to be.

Slide the reserve supply and the two policy rates. Find the scarce region where the rate hangs on supply, then flood the market and watch it pin to the floor.

A dollar today is worth more than a dollar next year, because today's dollar can earn interest in the meantime. That single fact, the time value of money, is the engine under every interest rate, loan, and bond. To compare money across time you discount future amounts back to today.

Discounting is compounding run backward. Because (1+r)t grows fast, distant payments are worth strikingly little today, and a higher rate crushes them further. Set a future payment, a discount rate, and a horizon, and read off what it is worth now. This present-value machine is all you need to price a bond in the next chapter.

A bond is a loan you can trade. It promises fixed coupon payments each year and returns its face value at maturity. Its price is nothing more than the present value of all those future payments, discounted at the market's required return, the yield.

This formula forces the single most important fact in fixed income: price and yield move in opposite directions. The coupons are fixed, so if the market demands a higher yield, the only way to deliver it is for the price to fall. Raise the yield and every discounted term shrinks; the bond gets cheaper. Three cases fall out cleanly:

• Yield equals coupon rate → price = face value (the bond trades at par).
• Yield above coupon → price below face (a discount bond).
• Yield below coupon → price above face (a premium bond).

Drag the yield and watch the price slide down the curve, crossing face value exactly where yield meets the coupon rate. Longer maturities make the curve steeper: their price swings more for the same yield move, which is duration risk.

Would you rather lend a friend $100 for one month or ten years? For the ten-year loan you would want extra interest: your money is locked up far longer, and far more can go wrong before you get it back. That extra interest is the term premium.

Plot the yield of otherwise-identical bonds against their maturity and you get the yield curve, the single most-watched picture in finance. Its shape is a forecast. Under the expectations view, a long rate is roughly the average of the short rates markets expect over the bond's life, plus a term premium for tying money up longer.

Normally the curve slopes up: the term premium is positive and rates are expected to hold or rise. It inverts (slopes down) when markets expect short rates to fall sharply, which usually means they expect the central bank to cut rates to fight a downturn. That is why an inverted yield curve is a famous recession warning: it has preceded most modern recessions. Set today's short rate, the expected drift in short rates, and the term premium, then read the slope and the signal.

Banking has a structural flaw baked into its business model: it promises depositors their money on demand, but it lends that money out long term. This maturity mismatch is useful (it channels idle savings into productive loans) and dangerous (the bank cannot pay everyone at once). If enough depositors demand cash simultaneously, even a healthy bank collapses. This is a bank run.

The Diamond-Dybvig insight is that a run can be self-fulfilling. A bank holds a little liquid reserves and a lot of illiquid loans that fetch only a fraction κ if dumped early (a fire sale). Serve depositors one at a time, paying $1 each: the bank can satisfy at most reserves + κ×loans of them. If withdrawals stay below that threshold, everyone who waits is paid in full and the bank survives. But if depositors fear a run, rushing to withdraw is rational (get in line before the cash runs out), and the rush itself exhausts the bank. Two equilibria exist for the same solvent bank: calm, or panic.

The bank below is genuinely solvent (loans at full value exceed deposits). Set the fraction who withdraw, or hit Panic, and step the queue. The self-fulfilling failure is the whole lesson.

Why is a small loss enough to wipe out a bank? Because banks are leveraged: they fund a large pile of assets with a tiny sliver of their own capital and a mountain of borrowed money (deposits and debt). Leverage L is assets divided by equity. A bank with $100 of assets and $5 of equity has L = 20, meaning $95 is other people's money.

This is the mathematics of fragility. High leverage magnifies returns in good times and annihilates the firm in bad times, and the threshold is brutally simple. Set the leverage, then ramp an asset shock and watch the equity buffer vanish the instant the loss crosses 1/L.

A run and a fire sale can turn a bank that is merely illiquid (short on cash today) into one that is insolvent (unable to repay at all). The classic remedy, stated by Walter Bagehot in 1873, is a lender of last resort: in a panic the central bank should lend freely, against good collateral, at a penalty rate. By advancing cash against the bank's solid assets at close to full value, it removes the need to dump those assets at a loss.

The distinction is everything. A solvent bank facing withdrawals only needs a bridge. Force it to fire-sale, and the haircut h on each dollar of assets sold becomes a realized loss that eats into equity; a big enough forced sale converts a liquidity problem into insolvency. Route the same funding through the central bank at full value and the loss never happens. Below, an illiquid but solvent bank meets a withdrawal wave. Toggle the lender of last resort and watch fire-sale losses appear or vanish.

The 2008 crisis was every chapter of this part firing at once. Banks were extraordinarily leveraged against mortgage assets; when house prices fell, thin capital was wiped out; short-term funding ran like a bank run on the wholesale market; and forced fire sales spread losses across every balance sheet. With the overnight rate already cut to zero, the central bank could not cut further. So it reached for a new tool: quantitative easing.

Watch the central bank's balance sheet during QE. Each round it buys a chunk of bonds and creates an equal chunk of reserves, so assets and liabilities grow together and always balance. Crucially, those reserves piled up as excess reserves rather than multiplying into broad money, exactly the leakage from chapter 6, which is why the money supply and inflation stayed far more subdued than a naive multiplier predicted. QE is why we now live in a floor system: reserves are abundant, so the rate is set by IOR.

Step back and the whole system is one chain. Money exists to break the double coincidence of wants and to store value across time. Banks take that money as deposits and, by lending, create most of it, expanding a fresh injection by the multiplier 1/rr minus leakages. The central bank sits above them, setting the quantity of reserves through open-market operations and, in today's floor system, the price of reserves through interest on reserves.

That policy rate ripples outward through the time value of money: it discounts every future cash flow, prices every bond (inversely to yield), and shapes the yield curve whose slope forecasts the economy. And the same maturity mismatch that lets banks create money makes them fragile: thin capital, high leverage, and on-demand deposits mean a loss of confidence can become a run, a fire sale, and a solvency crisis, held back only by deposit insurance and a lender of last resort.

You can now read a bank's balance sheet, explain where money comes from, follow a rate decision from the reserve market to the mortgage market, price a bond, and diagnose a crisis. That is the working literacy of money and banking.

You have the mechanics. To go deeper, follow the threads that start here:

• Macroeconomics: how the policy rate feeds through to output, unemployment, and inflation (the sibling guide in this series).
• Monetary policy in practice: the Taylor rule, forward guidance, and how central banks communicate.
• Fixed income: duration, convexity, credit spreads, and how bond portfolios are actually managed.
• Financial regulation: Basel capital and liquidity rules, stress tests, and the design of deposit insurance.
• Crisis history: 1907, 1929, 2008, and 2023 read very differently once you can see the balance sheets.

Every model in this guide ran the real math in your browser. The best next step is to change the numbers until the mechanisms feel obvious, then go read the news with new eyes.