How banks create money

Here is the crux: the single most misunderstood mechanism in all of macroeconomics, and the place where a very famous textbook model is quietly wrong about causation. You will learn both models honestly, because the wrong one is still taught everywhere and you must be able to speak it. Then you will learn why the profession has moved.

4a. The textbook model: the money multiplier

The story in most introductory textbooks (and in a lot of RBI/Fed "how banking works" explainers) goes like this. Banks operate on fractional reserves: they must hold only a fraction of their deposits as reserves and may lend out the rest. Call the required reserve fraction r. Now inject fresh reserves and watch them multiply.

Suppose the central bank creates ₹1,000 of new reserves (it buys a government bond, paying with newly created reserve money) and this lands as a ₹1,000 deposit in Bank A. With r = 10%, Bank A must keep ₹100 and may lend ₹900. It lends ₹900 to a borrower, who spends it, and the ₹900 is deposited in Bank B. Bank B keeps ₹90 and lends ₹810, which is deposited in Bank C, which keeps ₹81 and lends ₹729, and so on, each round 90% of the last:

Money multiplier formula

The deposits form a geometric series 1,000 × (1 + 0.9 + 0.9² + …) = 1,000 ÷ (1 − 0.9) = 1,000 ÷ 0.1 = ₹10,000. The original ₹1,000 of reserves has supported ₹10,000 of deposits. The money supply expanded tenfold. The multiplier is:

Money multiplier = 1 / r        (here 1 / 0.10 = 10)

The tidy implications: the central bank controls the money supply by controlling reserves; a lower reserve requirement r means a bigger multiplier and more money; reserves are the binding constraint on lending, since banks lend only because they received reserves and deposits first. This is clean, memorable, and it is how most people picture banking. Learn the arithmetic cold, because it appears on exams and in the interpretation of India's CRR. But hold the causation loosely. It is the part that is wrong.

4b. The modern correction: loans create deposits (endogenous money)

In 2014 the Bank of England published a short, plain-English paper by McLeay, Radia and Thomas, Money creation in the modern economy (Q1 Bulletin). It stated, with the full authority of a central bank, what monetary specialists had long known and textbooks had long garbled. Its thesis, in its own words: "Whenever a bank makes a loan, it simultaneously creates a matching deposit in the borrower's bank account, thereby creating new money." And, pointedly: "This is quite different from the description found in some economics textbooks" — banks do not wait to receive deposits and then multiply them out; the act of lending itself creates the deposit. Causation runs the opposite way to the multiplier story: loans → deposits → (reserves found afterward), not reserves → loans.

Watch it in the bank's own bookkeeping. A creditworthy customer walks into HDFC Bank and takes a ₹10 lakh home loan. The bank does not go to a vault, or wait for someone else to deposit ₹10 lakh. It makes two ledger entries at once:

HDFC Bank
  Assets                          Liabilities
  + Loan to customer  ₹10,00,000  + Customer's deposit  ₹10,00,000

That is the entire operation. On the asset side sits the loan (the customer's promise to repay); on the liability side sits a brand-new ₹10 lakh deposit in the customer's account, which the customer can now spend. The money supply just rose by ₹10 lakh, created out of a keystroke, backed by nothing but the loan contract. No prior deposit was lent out; no reserves were required to start. The bank created money by extending credit, which is why, as the functions of money showed, "whoever creates deposits creates most of the money, and that is the commercial banks."

What limits how much money banks can create?

If lending is not constrained by a pre-existing pile of reserves, what does constrain it? The BoE paper is explicit, and this is the analyst's real checklist for how much money an economy will create:

• Profitable lending opportunities and loan demand. Banks lend when there are creditworthy borrowers who want to borrow at a rate that covers the bank's costs and risk. In a slump, few want to borrow and few are creditworthy, so money creation stalls no matter how many reserves exist.
• Bank capital, not reserves. The binding regulatory constraint on how much a bank can lend is its capital (equity absorbing potential losses, the CRAR/CET1 you learned in M5.01), because each loan carries risk that must be capitalised. A bank runs out of capital headroom long before it runs out of the ability to create deposits.
• The price of reserves: monetary policy. After lending, banks do need reserves, both to settle payments between each other and to meet reserve requirements (India's CRR). But they obtain those reserves after the fact, borrowing them in the interbank market or from the central bank, at a price the central bank sets: the policy rate. The central bank does not ration the quantity of reserves to control lending; it sets their price and supplies whatever quantity the system needs at that price. So monetary policy is the ultimate limit on money creation, but it works through the interest rate, not through a reserve dam. (This is why the section on interest rates calls the interest rate the master price, and why M7.02 is about the rate, not the quantity.)
• Repayment destroys money. The mechanism runs in reverse too, and symmetrically. When the customer repays the ₹10 lakh loan, both entries are cancelled, the loan asset and the deposit both vanish, and that money is destroyed. In an economy where borrowers are collectively repaying more than they borrow (a deleveraging), the money supply shrinks. This is not a metaphor; it is arithmetic, and it is why deleveragings are deflationary unless offset.

Which model do economists use today?

Which does the profession favour, and why? The endogenous "loans create deposits" view is now the mainstream description of how modern banking actually works. It is the Bank of England's official account, it matches how the Federal Reserve and the RBI actually operate (setting a rate and supplying reserves elastically to hit it), and it is the consensus of monetary economists. The simple money multiplier survives only as a limiting-case teaching device. It is not useless, since reserves and deposits are linked, and under a genuinely binding reserve requirement the 1/r ratio marks a real ceiling. But it is misleading about causation, because it implies the central bank pushes money out through reserves when in fact banks pull money into existence through lending and the central bank accommodates.

The decisive real-world evidence is the great post-2008 experiment in quantitative easing. From 2008 to 2015 the Fed (and later the ECB, BoJ, and others) created trillions of new reserves, and the monetary base exploded. The multiplier model predicted a proportional explosion in the money supply and in lending. It did not happen: banks sat on the reserves, broad money and lending grew sluggishly, and inflation stayed low for years, because lending was constrained by weak demand and scarce bank capital, not by a shortage of reserves. Reserves ballooned; loans did not multiply out of them.

The multiplier's causation had it backwards, in public, at civilisation scale.

What should an analyst watch?

The analyst's takeaway. Watch credit growth, the flow of new bank lending, as the driver of money and spending, not the monetary base. When you read that "the RBI cut the CRR" or "the Fed expanded its balance sheet," do not assume money and spending must surge; ask instead whether banks want to lend and borrowers want to borrow, because that, plus the price of credit (the rate), is what actually moves the money supply. A bank stock (HDFC Bank, Bajaj Finance, JPMorgan) is, in the deepest sense, a money-creation machine whose fortunes track the credit cycle, which is why the cycles in Dalio's economic machine are a lender's entire world.

(India note: the RBI's Cash Reserve Ratio (~3% as of mid-2026) and the Statutory Liquidity Ratio (~18%) give reserve/liquidity requirements more day-to-day bite in India than in, say, the post-2020 US, where the Fed set the reserve requirement to zero and runs an "ample-reserves" regime. But even in India the causation is loan-led: banks lend to creditworthy demand against their capital, then meet CRR/SLR; the RBI manages the price of liquidity, not a lending dam.)