The Two-Tier Banking System: T-Accounts and Loans Create Deposits

The two-tier banking system refers to the layered structure of the modern monetary order — the monetary system divides into three tiers: central bank → commercial banks → households (where “households” in economics means enterprises plus individuals); the central bank issues base money to banks, and banks create broad money for households — along with the analytical method of tracing this structure through three-party balance sheets (T-accounts). Its core conclusion is loans create deposits: when a bank makes a loan, it simultaneously records “loan / deposit” as a matched pair of entries, rather than needing “deposits first before lending.” Base money is the money held by banks (a liability of the central bank); broad money is the money households keep at banks (a liability of the banks). Both are, in essence, layered claims — they can be created, and they can be destroyed.

What It Is

The cleanest way to understand this system is a thought experiment: imagine a country moving in one step from barter to a modern monetary system, with only two balance sheets — a central bank and a commercial bank — and using double-entry bookkeeping (left must equal right) to trace every monetary transaction. In this experiment there is no gold or silver, no historical baggage, and the true face of money is thus laid bare: the entire monetary system is a nested structure of claims and obligations. Base money is the banks’ money — and simultaneously the central bank’s debt. Broad money is the households’ money — and simultaneously the banks’ debt. Money’s essence is an IOU/claim rather than a physical thing; it can be created (signing an IOU) and it can be destroyed (retiring an IOU).

The relationship among the three tiers is nested, not parallel: households generally do not deal directly with the central bank; households face banks, and banks face the central bank. Base money = the money held by banks (an asset of banks, a liability of the central bank); broad money = the money households keep at banks (an asset of households, a liability of banks) — the same word “money,” yet it sits in entirely different tiers on the balance sheet. This layering is the foundation for the analysis in Defining Liquidity and the Layering of Money.

Why It Is So

Walk through three T-accounts and the full sequence — from money’s creation to its change of form — requires only four steps.

Step one: the central bank signs an IOU, and base money enters the ledger. The central bank has a dual identity that, if left unclarified, makes money creation impossible to understand: as the issuer of currency it is a debtor — banknotes are IOUs it has signed; but when it lends money to commercial banks, it becomes a creditor. Such lending is collectively called re-lending; MLF / SLF / reverse repos are essentially the same, differing only in conditions. (Compiler’s note: in strict classification, reverse repos fall under open-market operations while MLF/SLF are standing lending facilities; the course groups them under the functional heading of “central bank liquidity injection.“) The bank deposits the borrowed money back at the central bank — this amount is base money: an asset of the bank, a liability of the central bank.

Step two: four cells recorded simultaneously, loans create deposits. An enterprise applies to borrow 200; the bank records “loan 200” on the asset side and “enterprise deposit 200” on the liability side; the enterprise records “deposit 200” on the asset side and “loan payable 200” on the liability side — four cells each showing 200, double-entry bookkeeping balances, cash need never physically appear, and the bank’s true holdings remain only those 100 units deposited at the central bank. Lending is not moving existing deposits to a borrower; it is writing a number into the enterprise’s account from thin air. “Deposits must come first before lending” is the logic of ancient commodity money (copper coins and the like), passed down through generations — but it does not apply to modern money (digital / IOU / accounting entries).

Step three: the constraint surfaces — fractional reserves. The enterprise will eventually withdraw; once withdrawals exceed the bank’s true holdings, the bank fails — so the most important premise for this business model to hold is “depositors will not all withdraw simultaneously in a single rush.” Under this statistical premise, if a bank holds 100 in real money (reserves) and the reserve ratio is 10%, the maximum loan volume is 1,000 (corresponding to 1,000 in deposits): with only 100 in base money, the bank can extend far more on the ledger. The reserve ratio as a regulatory tool is elaborated in The Reserve System: Required Reserve Ratio and Excess Reserves. (Compiler’s note: the “10% reserve ratio → 1,000 ceiling” in the text is a simplified closed-system pedagogical multiplier example; in practice, actual constraints on credit expansion come primarily from capital adequacy requirements, liquidity regulation, settlement funding, credit demand, and risk appetite — the reserve ratio is only one among them. See BoE 2014, “Money Creation in the Modern Economy.“)

Step four: cash withdrawal linkage — banknotes bypass banks. A household withdraws 50 in cash; all three balance sheets shift simultaneously: on the household side, deposits fall from 500 to 450, with 50 in banknotes appearing; on the bank side, deposits fall from 500 to 450, and balances held at the central bank fall from 100 to 50; on the central bank side, bank deposits fall from 100 to 50, with 50 in banknotes appearing. Withdrawal converts base money into banknotes — and banknotes are a special case: they are a direct claim between the central bank and households, bypassing banks. They are simultaneously broad money and base money; as banknote use declines, the importance of this special case diminishes.

flowchart TD
  A["Central bank issues IOU (issues currency)\nCentral bank = debtor"] --> B["Re-lending MLF/SLF/reverse repo\nCentral bank lends to commercial banks → central bank = creditor"]
  B --> C["Bank deposits money back at central bank\n= Base money (bank asset / central bank liability)"]
  C --> D["Enterprise applies for loan, bank lends\nAsset records loan · liability records deposit (four cells)"]
  D --> E["Loans create deposits\nNumbers entered from thin air · cash need not appear"]
  E --> F["Household deposits = broad money (household asset / bank liability)"]
  F --> G["Constraint: fractional reserves\nDepositors do not all withdraw simultaneously"]
  G --> H["Cash withdrawal = base money converts to banknotes\nBanknotes bypass banks (central bank ↔ household directly)"]

How to Judge

For any monetary transaction — lending, cash withdrawal, reserve posting, loan repayment — the judgment path is fixed:

• Draw three balance sheets first: central bank / commercial bank / household; each divided into asset and liability sides, maintaining left-right equality throughout.
• Ask about claims and obligations: for any monetary transaction, first ask “whose claim on whom is changing,” then record simultaneously in the corresponding cells of two balance sheets — looking at only one balance sheet’s ups and downs will inevitably produce a wrong reading.
• Distinguish tiers: identify whether the money in question is base money (held by banks / a central bank liability) or broad money (held by households / a bank liability); the two tiers do not interchange, and mixing tiers is the most common source of error in monetary analysis.
• See through “deposits first”: when you encounter the claim that “a bank must have deposits before it can lend,” recognize it as the logic of ancient commodity money and correct it with “loans create deposits.”
• Cash / withdrawal transactions: remember that banknotes represent a direct claim between the central bank and households, and will simultaneously alter all three balance sheets.

This judgment framework reaches only the “mechanism layer” — how accounts are recorded, how money is derived. The reserve ratio as a policy instrument and excess reserves as tradable liquidity are covered in The Reserve System: Required Reserve Ratio and Excess Reserves; the drainage and injection of base money by the interbank market and fiscal operations are covered in The Interbank Market and Fiscal Drain-and-Release.

Its Place in the Whole

This entry is the foundational entry in the monetary-system-and-circulation lineage of this site’s Fa layer: all higher-level liquidity analysis — total social financing, M2, central bank tools, the dollar circulation system — ultimately traces back to specific cells in these three balance sheets.

“Deposits first, then lending” is not merely an intuitive error; it is a category error — the physical causal chain of ancient commodity money (copper coins) being pasted onto a modern accounting system. Copper coins genuinely had to be present before they could be lent out; but a modern bank’s “deposit” and “loan” are both accounting numbers that can be generated simultaneously in a single transaction — the asset side records loan 200, the liability side records deposit 200, the books balance, and not a single physical coin has moved. Someone analyzing banks with the old logic will be systematically wrong on one specific judgment: “the scale of bank lending is constrained by the current balance of deposits it has attracted.” Within the course’s simplified model, the binding constraint is not the deposit balance but the 1,000-unit ceiling under a 10% reserve ratio (with 100 in base money), plus the behavioral premise that “depositors do not all withdraw simultaneously.” These two constraints are entirely different in nature — one is a proportional constraint, the other a statistical-regularity assumption — and conflating them leads to incorrect predictions about the amplification effects of monetary easing. (Compiler’s note: this is the course’s simplified closed-system formulation; real-world constraints are dominated by capital adequacy, liquidity regulation, settlement, and credit demand — see the earlier note.)

This yields the proprietary assertion of this entry: “Loans create deposits” is a conclusion from double-entry bookkeeping as an accounting identity, not a metaphor. “Deposits are a prerequisite for loans” and “the reserve ratio constrains loan volume” are two entirely different propositions — the former is false in the modern monetary system, the latter is true. Conflating them gets both the nature of the constraint mechanism and the direction of causation wrong at the same time.

Looking upward, this entry connects to the Dao-layer ordering of “why matters far more than how”: the four-cell entry method is the “how”; The Essence of Money Is an IOU: The Creation and Destruction of Credit provides the “why” — “money = an IOU that can be signed and torn up.” Reading the former without the latter, the T-account is merely an accounting trick; reading with it, the T-account is the visible trace of credit being created and destroyed from nothing on a balance sheet. Why money can be created — the answer is in that entry; the specific operational mechanism of creation — the three-party synchronized T-account method — is in this entry. Even reading both together is insufficient for judging central bank policy; Central Bank Policy Tools and Monetary Policy Transmission is still needed for the policy-tool layer.

See Also

• The Essence of Money Is an IOU: The Creation and Destruction of Credit

• The Reserve System: Required Reserve Ratio and Excess Reserves

• The Interbank Market and Fiscal Drain-and-Release

• Central Bank Policy Tools and Monetary Policy Transmission

• Total Social Financing and Broad Money M2: Accounting Definitions and Divergence

• The US Two-Tier Banking System: 2020 QE as Deficit Monetization and MMT in Practice

Sources

• Compiled draft z-0192 · collected 2026-07
• External course (collected anonymously) lectures 11-16 (primary), 11-17 (primary), 11-15 (supporting)