What Money Is: Aggregates, Moneyness & Endogenous Money

Ring: Finance-Plus (institutional/CFA breadth) · Branch: Money & Monetary Systems (MS) · Node MS1.01 · ~16 hours · Mechanism exam ≥85%

You already know, from M7.01, the sentence that ends most people's macro education just as it should begin: banks do not lend out other people's deposits, they create new deposits the instant they lend. That sentence is correct, it is the single most important fact about modern money, and if you stopped there you would still be dangerously undertrained, because M7.01 told you the story at the level a macro narrative needs (one bank, one loan, one clean T-account) and left the plumbing underneath it entirely unopened. What happens the instant that new deposit is spent to somebody who banks somewhere else? What, precisely, stops a bank from creating money without limit? What does it mean for a rupee in your savings account to be "money" when the note in your wallet obviously is, but the mutual-fund unit in your other account obviously is not, and a large, uninsured balance at a bank you trust sits somewhere uncomfortably in between? This module is the plumbing. It takes the macro-narrative claims of M7.01 and M7.02 (loans create deposits, the aggregates run M0 through M3, reserves and broad money are different things) as read, and goes one level down: the actual balance-sheet mechanics of who owes what to whom, shown as T-accounts that must balance because arithmetic does not negotiate. This Ring is additive institutional/CFA breadth, it presents the toolkit and the plumbing alongside, not in place of, the program's process-over-P&L, concentration-over-diversification philosophy: nothing here argues that understanding a bank's reserve account changes how you should build conviction in a business; it argues that an analyst who covers banks, NBFCs, or macro-sensitive businesses should be able to read the ledger entries underneath the headlines instead of only the headlines.

Two ideas hold up everything that follows, and neither one is in M7.01 or M7.02 in the form you need it here. The first is moneyness: money is not a single, binary category, it is a hierarchy of promises, each layer only as good as its promise to convert into the layer above it, at par, on demand, and that promise can and does break under stress, which is precisely why a mutual-fund unit is not a bank deposit and a bank deposit is not a currency note, even though all three "feel like money" on an ordinary Tuesday. The second is settlement: the reason a single bank cannot create money without limit is not (mainly) a textbook reserve requirement, it is that the deposit it just created will, within hours, be spent to someone who banks somewhere else, draining the lending bank's reserves and forcing it into the interbank market to fund the gap. Get both ideas onto T-accounts that balance, in Python, and the rest of this Ring's monetary plumbing. MS1.02's payment rails, MS1.03's repo and shadow-banking chain, MS1.04's eurodollar system and stablecoins, stops being a list of acronyms and becomes an extension of a mechanism you already own.

One last framing point, stated once here because it governs the whole branch. Reasoning about "whose balance sheet expands, and by how much" is exactly the kind of clean, confident-sounding narrative a language model can produce fluently without any of it being correct, it can get the direction of an entry backwards, double-count a deposit that only moved, or invent a plausible-sounding multiplier that no actual ledger supports, and the prose will read exactly as authoritative either way. This region is very plausibly this program's single highest AI-over-trust risk, precisely because "money view" reasoning has no single number you can spot-check against a filing the way you would a reported EPS, the only real check is redoing the T-accounts yourself and confirming they balance. R3's law for this whole branch (every number recomputed by you, every balance sheet proven to balance) is not bureaucratic caution here. It is the entire defense.

Learning objectives

By the end you can:

1. Explain moneyness as a hierarchy, not a binary (Mehrling's "money view"): name the tiers from central-bank money down through bank deposits to money-market/shadow instruments, define what "trading at par, on demand" means, and explain why that promise can break under stress.
2. Read the M0/M1/M2/M3 aggregate ladder (owned by M7.02) and, given a system's raw components, compute each aggregate yourself and derive the base-money-to-broad-money ratio bottom-up from a balance sheet, rather than quoting it as a fact.
3. Extend the single-bank "loans create deposits" mechanism (owned by M7.01) to a multi-bank system: build full, balancing T-accounts showing what happens to both banks' balance sheets and reserve positions the instant a newly created deposit is spent to a customer at a different bank, and name this interbank settlement.
4. Distinguish the reserve constraint from the capital constraint on a bank's lending, compute each precisely from a bank's own numbers (CRR/NDTL for reserves; capital, risk-weighted assets, and CRAR for capital, referencing M5.01), and state which one actually binds day to day versus over the life of the bank.
5. Compute how much new lending capacity a given amount of fresh bank capital unlocks (the leverage-on-capital relationship), and explain, in balance-sheet terms, why banks raise equity to fund growth.
6. Quantify the moneyness of a retail bank deposit using India's DICGC and the US's FDIC insured limits, and, with one real historical case on each side, explain how a wholesale-money instrument and a retail deposit base can each lose moneyness under stress.
7. State why "money view" reasoning is this program's highest AI-over-trust-risk pattern (R10) and apply the by-hand T-account discipline that is the only real defense against a fluent, wrong balance sheet.

Prerequisites & connections

Builds on. M7.01 (The Economic Machine) is this module's macro-narrative foundation and is assumed, not re-taught: money's three functions (medium of exchange, unit of account, store of value), credit as a claim on future spending, and (critically) the single-bank "loans create deposits" mechanism with its Bank of England (2014) citation. If that sentence still feels slippery, stop and revisit M7.01 §4 before continuing; everything below extends it and will not re-derive it. M7.02 (Central Banks) owns the full M0–M4 aggregate ladder with India's and the US's exact definitions, the CRR/SLR toolkit, and the reserves-versus-broad-money mechanics of QE (its Case A/Case B T-accounts), this module takes that ladder and those cases as given inputs and computes with them, rather than restating them. M1.01 (double-entry bookkeeping) supplies the T-account grammar itself: every ledger in this module is a direct application of "assets = liabilities + equity, and every transaction hits at least two places," never re-derived here. M5.01 (bank analysis) owns CRAR, Basel III capital ratios, and India's PCA thresholds as bank-health and valuation metrics; this module borrows those exact ratios and definitions and repurposes them into a lending-capacity ceiling, a calculation M5.01 does not itself perform. M5.02 (NBFC analysis) supplies the IL&FS/DHFL commercial-paper-market-freeze narrative that this module references rather than re-tells when it discusses moneyness breaking down on the credit side.

Feeds into. MS1.02 (The Payment System & Plumbing) takes the bare fact this module establishes (that an interbank payment ultimately moves central-bank reserves between two accounts) and builds the full messaging and settlement rails on top of it: RTGS/Fedwire, ACH/NACH, UPI, card networks, and cross-border correspondent banking. This module deliberately stops at "reserves move between two accounts at the central bank" and leaves the rails themselves to MS1.02; do not expect (or attempt) that detail here. MS1.03 (Central-Bank Operations & the Repo/Money-Market Complex) takes this module's moneyness hierarchy and builds the full shadow-banking chain on top of its middle layers (repo, haircuts, collateral, money-market funds) that this module only introduces. MS1.04 (the branch's crown) situates the eurodollar system, CBDCs, and stablecoins explicitly within the hierarchy this module teaches; you cannot place a stablecoin correctly in that module without first knowing what "moneyness" means here. Elsewhere in the Ring, AL1.01 already named hedge-fund performance data as a comparably severe AI-over-trust risk region; this module's R10 statement is this branch's version of that same warning, applied to balance-sheet reasoning instead of self-reported returns.

The one-sentence version of this module. Money is a hierarchy of promises to convert into something more fundamental, at par, on demand (central-bank money at the top, bank deposits next, money-market instruments below that) and the entire architecture of modern banking (the aggregates, the way banks create money, the reserve requirement, the capital requirement, and the interbank market that settles payments between banks every single business day) is the machinery that keeps most of that hierarchy trading at par most of the time, and occasionally fails to.

Core teaching content

Money as a promise, briefly recapped

M7.01 already gave you money's three functions in full (medium of exchange, unit of account, store of value) and the deeper fact that most money is not cash but bank deposits. Hold both, but notice what they have in common: a bank deposit is not a physical thing sitting in a vault with your name on it. It is a promise, an entry in a bank's ledger stating that the bank owes you a certain number of rupees or dollars, redeemable in currency (or, more usually, transferable to someone else's account) on demand. Currency itself, once you strip away the paper, is also a promise, but a different kind: a central bank's own IOU, backed by nothing but the state's acceptance of it for taxes and its own balance sheet. Once you see that "money," in a modern economy, is almost entirely a layered set of promises rather than a pile of things, the natural next question is the one M7.01 and M7.02 never had to ask, because their story stayed inside a single layer: are all these promises equally good? They are not, and the difference between them is this module's actual subject.

The aggregates, computed rather than quoted

M7.02 §4.6 already gives you the full, exact ladder. India's official M0 through M4 as the RBI defines them, and the US Fed's M1/M2 (with M3 discontinued in 2006), and the real order-of-magnitude fact that broad money runs several times base money in both systems. Do not re-derive that table here; go reread it if the definitions have gone soft. What this module adds is the thing M7.02 never had you do: actually compute the aggregates yourself from raw components, on a snapshot you build, so that "M3/M0 ≈ 5" stops being a fact you were told and becomes a number you derived.

The mechanical recipe, using India's exact RBI definitions as the worked case (the US ladder works identically with its own definitions): M0 (reserve money) = currency in circulation + bankers' deposits with the central bank + other deposits with the central bank. M1 (narrow money) = currency with the public + demand deposits with banks + other deposits with the central bank. M2 = M1 + savings deposits with post-office savings banks. M3 (broad money) = M1 + time deposits with banks. Notice the structural point buried in the arithmetic: M0 counts what banks hold at the central bank; M1 counts what the public holds at banks and in hand. The two aggregates are built from almost entirely different balance sheets (M0 lives on the central bank's books, M1/M3 live on the commercial banking system's books) and the ratio between them is nothing more mysterious than the size of the banking system's own balance sheet relative to the base it was built on. Worked Example 6 builds a full synthetic system from six raw components and derives M0, M1, M2, and M3 directly, so you do this arithmetic with your own hands at least once.

Two clarifications worth carrying forward, because Practice Set items will test exactly this: first, required reserves and excess reserves are not separate aggregates, they are a split of the "bankers' deposits with the central bank" line inside M0, and the split is precisely §4's and §5's subject below. Second, a rupee moving from "bank deposit" (M1) into "currency in the wallet" (also M1, differently composed), a plain cash withdrawal, changes the composition of M1 without changing its total, because you have simply swapped one form of at-par money for another; Practice Set item P15 makes you prove this on a T-account so it stops being a verbal claim.

How banks create money: extended to a system that settles with itself

M7.01 built the single most important fact in this branch and built it correctly: a bank does not lend out a depositor's money; the instant it makes a loan, it creates a matching deposit, in one keystroke, on its own balance sheet. Hold that fact exactly as M7.01 gave it to you. What M7.01's telling necessarily compressed (because a macro narrative has no room for a second bank) is what happens in the next five minutes, and that is where the real constraint on a single bank's lending actually lives.

The instant of lending. Ganga Bank (synthetic) extends a ₹2 crore loan to Kaveri Textiles (synthetic), a creditworthy business customer. Exactly as M7.01 taught: Assets: +₹2 cr loan; Liabilities: +₹2 cr deposit (Kaveri's new account). Nothing has left the building yet. Ganga Bank's reserves (its own account at the RBI) have not moved by a single rupee.

The next five minutes. Kaveri Textiles did not borrow ₹2 crore to let it sit. It immediately pays a supplier, Narmada Traders (synthetic), the full amount, and Narmada Traders banks at Yamuna Bank (synthetic), a different institution entirely. This is the ordinary case, not an edge case: in a competitive banking system with thousands of institutions, the overwhelming majority of any newly created deposit is spent within days to someone who banks somewhere else. This is the fact M7.01's single-bank telling has no room to show you, and it is the actual mechanical answer to "why can't a bank just create infinite money?"

The moment Kaveri's payment clears, three things happen at once, and MS1.02 will show you exactly which rails carry the message; here you need only the settlement asset underneath whichever rail it is. Kaveri's deposit at Ganga Bank is extinguished. Narmada Traders' deposit at Yamuna Bank is created, for the identical amount. And, the step M7.01 never had to show. Ganga Bank's reserve account at the RBI falls by ₹2 crore, while Yamuna Bank's reserve account at the RBI rises by the same ₹2 crore. The central bank's own ledger is the settlement asset: it does nothing but move a balance from one account (Ganga's) to another (Yamuna's), and its own total liabilities to the banking system are unchanged. This is interbank settlement, and it is the mechanism that turns "a bank can create a deposit with a keystroke" into "a bank's reserves drain the moment that deposit is spent elsewhere."

Worked Example 1 carries this all the way through with real numbers, including the awkward moment when Ganga Bank, having lent freely, ends the day short of the reserves it is required to hold, and must borrow the shortfall overnight from another bank. That borrowing-to-cover-a-shortfall step is the entire content of "how banks settle with each other" in cash terms, and it is worth stating the punchline before you reach the arithmetic: system-wide, the new deposit is real and permanent (total deposits across Ganga and Yamuna together rise by exactly the loan amount), but the reserves that back it are not created anywhere in this story; they merely relocate from the lending bank to whichever bank the money was spent toward. A single bank is therefore reserve-constrained in the short run (it must fund whatever leaves the building) even though the system as a whole is not reserve-constrained by this transaction at all, because system-wide reserves are exactly unchanged (only a central-bank operation, or the public withdrawing currency, changes the system total). This single distinction (one bank versus the whole system) is the "one level down" correction this module exists to make to M7.01's necessarily simplified telling.

The true constraint on lending: capital, not reserves

M7.02 already told you, correctly, that reserve requirements are not the mechanism that actually limits bank lending in the modern system, that banks lend first and find reserves after, and that the US cut its reserve requirement to zero in March 2020 without monetary policy breaking. It named the real binding constraint in one sentence: bank capital. This module's job is to turn that sentence into arithmetic you can run yourself.

Why receiving reserves is capital-free. First, the fact that makes the two constraints genuinely separate rather than two names for the same thing: under the Basel standardized approach, central-bank reserves carry a 0% risk weight. A bank that receives ₹50 crore of fresh reserves, say, because the central bank bought a bond from one of the bank's customers, crediting the customer's account and the bank's reserve account in the same stroke (M7.02's own Case B), sees its risk-weighted assets (RWA) rise by precisely zero, because the new reserve asset carries no risk weight at all. Its capital ratio does not move. The ratio only moves once the bank converts some of that reserve asset into something risk-weighted, a loan. This is the precise mechanical reason reserves and capital are two separate axes, not one dominating the other in every situation: reserves govern short-run liquidity and settlement (§ above); capital governs how much risk-weighted lending the bank's own loss-absorbing cushion can support, full stop, regardless of how many "free" reserves happen to be sitting on its balance sheet.

The capital ceiling, in one formula. Referencing M5.01's own definitions and minimums without re-deriving them: a bank's total capital ratio is capital ÷ risk-weighted assets, and regulation sets a minimum. India's Basel III framework requires a total capital ratio of at least 11.5% including the capital conservation buffer (the same figure M5.01 cites as the PCA capital threshold, because breaching your own required minimum is exactly what triggers Prompt Corrective Action). Rearranging the ratio gives the ceiling directly: maximum RWA a bank can carry = capital ÷ minimum ratio. Subtract the RWA the bank already carries and you have its headroom, the new risk-weighted lending it can add without breaching the floor, with no reference to reserves anywhere in the calculation. Worked Examples 2 and 3 compute this precisely for named synthetic banks on both sides of R6; Practice Set items P6, P7, and P13 have you compute it independently.

The leverage-on-capital relationship. Because the ceiling is capital ÷ minimum ratio, every additional rupee (or dollar) of capital a bank raises mechanically unlocks 1 ÷ minimum ratio of new lending headroom, at India's 11.5% minimum, that is ≈8.70 rupees of new lending capacity per rupee of fresh capital (at a 100% risk weight; less if the new lending carries a lower risk weight, more if higher). This is not a rule of thumb; it falls directly out of the ceiling formula, and it is the balance-sheet reason a growing bank periodically raises equity, a QIP, a rights issue, a follow-on offering, even when it has no shortage of deposits or reserves to lend against. Worked Example 3 runs this exactly for a synthetic Indian bank raising fresh capital via a QIP; Practice Set item P7 has you reproduce it with different numbers.

Which one binds, when. Day to day, a bank's most pressing operational problem is usually the reserve/liquidity constraint, it must settle every payment leaving the building today, in central-bank money, today, and a shortfall means an overnight borrowing (§ above) or, failing that, a scramble that regulators watch closely. Over the life of the bank, the constraint that actually decides how large its loan book can grow is capital, a bank sitting at its regulatory floor cannot add a single rupee of net new risk-weighted lending, however much "free," zero-risk-weight liquidity is parked on its balance sheet, until it raises fresh capital, retains earnings, or shrinks risk-weighted assets elsewhere. Practice Set item P13 gives you a bank that is simultaneously short of reserves today and comfortably capitalized for growth over its horizon, the realistic case, and the one that separates learners who have merely memorized "capital is the real constraint" from those who can actually apply both constraints to the same balance sheet at once.

{
  "skill": "Extracting a real bank's actual capital ratio and lending headroom from its own regulatory disclosure, instead of a summarized commentary",
  "use": "Have a document-analysis AI read a bank's actual Basel III Pillar 3 disclosure or 10-K capital-adequacy note (a document you hold, pulled yourself from the bank's investor-relations site, the RBI's disclosures, or SEC EDGAR) and extract its reported capital, risk-weighted assets, and capital ratio — then compute the lending-capacity headroom yourself using this module's ceiling formula, rather than asking the AI to compute or interpret it.",
  "tools": ["Document/PDF AI grounded on the actual Pillar 3 disclosure or 10-K (not a bare query about the bank by name)", "A bank's own investor-relations site or SEC EDGAR / RBI disclosures to obtain the real filing", "A spreadsheet or short script for the headroom arithmetic itself"],
  "prompt": "Here is <BANK>'s latest Basel III Pillar 3 disclosure (or 10-K capital-adequacy note), attached. Extract, with the exact page/table reference: total capital, risk-weighted assets, and the reported capital ratio. Quote the exact figures; do not paraphrase a number you cannot point to a specific line for.",
  "verify": "Confirm the document is the bank's own current, dated regulatory disclosure (Guardrail #1 Trace, #4 Date-check — capital ratios are reported quarterly and go stale fast), then recompute the ratio yourself from the extracted capital and RWA figures and the headroom from this module's ceiling formula (Guardrail #3, Recompute) before trusting any interpretation the model offers on top of the raw numbers.",
  "diy": "You must already be able to derive the capital ceiling and headroom formula from a bank's capital ratio by hand (the section above) before delegating the extraction step to a tool — otherwise you have no way to notice the AI quietly using Tier 1 capital where the minimum applies to total capital, or RWA where the disclosure actually reports total assets.",
  "market": "US"
}

Common mistakes & how experts think differently

1. Treating "money" as one thing instead of a hierarchy. Novices ask "is this money or not?" as if the answer were binary. Experts ask "how much moneyness does this instrument carry, and what exactly stands behind its promise to convert at par?", because a bank deposit, a money-market fund share, and a currency note all "feel like cash" on an ordinary day and behave completely differently the one day it matters (Worked Example 4).
2. Confusing "the central bank created reserves" with "the money supply grew." M7.02 already dismantled this for QE specifically; the mistake recurs constantly elsewhere. Reserves are base money that only banks hold; they become broad money only if and when a bank converts them into a loan (creating a new deposit) or the seller of an asset to the central bank is a non-bank whose account gets credited. Experts always ask which of the two happened before concluding anything about broad money.
3. Imagining one bank re-lending the same money over and over. The textbook multiplier table is often pictured as a single bank somehow lending its own deposit back to itself repeatedly. In the real system, each successive "round" is a different bank, receiving a fresh deposit and a fresh reserve credit via interbank settlement, independently deciding whether it has a creditworthy borrower, which is exactly why the chain can and does stop at any link, for demand or capital reasons, rather than mechanically running to the textbook ceiling (Worked Example 1 and Worked Example 2).
4. Believing ample reserves mean a bank can always lend more. Reserves and capital are separate axes. A bank sitting exactly at its regulatory capital floor cannot add a rupee of net new risk-weighted lending no matter how many "free," zero-risk-weight reserves land on its balance sheet, it is capital-constrained, not reserve-constrained, and the two frictions require completely different fixes (raise capital versus source overnight funding).
5. Reading "the reserve requirement is low/zero" as "banks face no constraint." A near-zero or literally zero reserve requirement (the US since March 2020) removes one specific friction and leaves every other constraint (capital, loan demand, and the day-to-day need to fund whatever reserves leave the building via settlement) fully in place. "Ample reserves" describes a regime, not the absence of constraints.
6. Treating a deposit-insurance limit as covering "most of the bank's deposits." It reliably covers most accounts, by count, because most individual balances are small. It does not reliably cover most deposit value, because large balances concentrate disproportionately in a small number of accounts, exactly the gap Worked Example 5 quantifies, and exactly the distinction that made one bank's 2023 failure a systemic event rather than a routine resolution.
7. Assuming a fluent AI answer about "whose balance sheet expands" is itself verified. There is no primary-source filing to trace this reasoning to the way you would trace an EPS figure to a 10-K, the claim is the arithmetic itself. The only real check, every time, is rebuilding the T-accounts and confirming they balance (R10, and the AI-augment callouts throughout this module).
8. Believing the money multiplier is a lever the central bank pulls. It is a ratio that falls out of how much lending the banking system actually chooses to do against a given base, observed after the fact, not set in advance. A central bank can flood the base with reserves and the multiplier can simply fall, exactly as it did after 2008 (M7.02), because the missing ingredient was creditworthy demand and spare bank capital, not reserves.
9. Overstating what "capital raise unlocks lending" proves about a specific bank. The leverage-on-capital formula (1 ÷ minimum ratio) is the ceiling the new capital supports; whether the bank actually originates that much new lending still depends on loan demand and risk appetite, exactly as Worked Example 2 shows for a bank with ample capital headroom but no creditworthy borrowers in front of it.

Worked examples

WE1: India ₹: multi-bank settlement, a reserve deficit, and the overnight fix

Objective: extend M7.01's single-bank "loans create deposits" T-account into a real, multi-bank settlement, and show precisely why a lending bank ends up needing to borrow reserves.

Setup. Ganga Bank (synthetic) has ₹400 cr of deposits, a 4% cash reserve requirement, and ₹16.50 cr of actual reserves at the RBI (required reserves = 4% × 400 = ₹16.00 cr, so Ganga starts with a ₹0.50 cr surplus). Yamuna Bank (synthetic) has ₹250 cr of deposits and ₹10.20 cr of actual reserves (required = ₹10.00 cr, a ₹0.20 cr surplus).

Step 1, the loan. Ganga Bank extends a ₹2 cr loan to Kaveri Textiles (synthetic). Assets: +₹2 cr loan; Liabilities: +₹2 cr deposit. Reserves have not moved.

Step 2, the payment. Kaveri Textiles pays the full ₹2 cr to Narmada Traders (synthetic), which banks at Yamuna Bank. Kaveri's deposit at Ganga Bank drains to zero (Ganga's deposits fall back to exactly ₹400 cr, the round trip nets out) while Narmada's new deposit appears at Yamuna Bank (Yamuna's deposits rise to ₹252 cr). System-wide deposits rise from ₹650 cr to ₹652 cr, up by exactly the ₹2 cr loan, even though the money is no longer at the bank that created it.

Step 3: settlement. The ₹2 cr moves as central-bank reserves: Ganga Bank's RBI account falls to ₹14.50 cr; Yamuna Bank's RBI account rises to ₹12.20 cr. Total system reserves are unchanged at ₹26.70 cr, nothing was created or destroyed; it only relocated.

Step 4, read the two banks' positions. Ganga Bank's deposits are back to exactly ₹400 cr, so its required reserves are unchanged at ₹16.00 cr, but its actual reserves have fallen to ₹14.50 cr. Ganga Bank now has a ₹1.50 cr reserve deficit. Yamuna Bank's deposits have risen to ₹252 cr (required reserves rise only slightly, to ₹10.08 cr) while its actual reserves rose the full ₹2 cr, to ₹12.20 cr: a ₹2.12 cr surplus, up from ₹0.20 cr. Check the system: Yamuna's surplus (₹2.12 cr) minus Ganga's deficit (₹1.50 cr) equals ₹0.62 cr, exactly the system-wide surplus (₹26.70 cr actual − ₹26.08 cr required), confirming nothing leaked.

Step 5, the overnight fix. Ganga Bank borrows exactly its ₹1.50 cr shortfall from Yamuna Bank overnight in the interbank call-money market (the corridor M7.02 builds in full). After the loan: Ganga's reserves rise to ₹16.00 cr, exactly meeting its requirement, zero buffer, and Yamuna's fall to ₹10.70 cr against its ₹10.08 cr requirement, a comfortable ₹0.62 cr surplus: the entire system's surplus, now sitting at Yamuna, having travelled there by the identical mechanism (interbank settlement) that first delivered it. At an illustrative call rate of 6.5% p.a., that overnight borrowing costs Ganga Bank ₹2,671.23 for the single night (₹1.50 cr × 6.5% × 1/365), a real, small, daily line item behind every bank's treasury desk, and the concrete cash cost of having lent more freely than its reserve position could support without help.

Read it. Nobody created money out of nothing that a central bank did not ultimately allow: system reserves never changed. What changed is distribution. Ganga Bank could create the deposit with a keystroke, but the instant its customer spent it elsewhere, Ganga became reserve-short and Yamuna became reserve-long, and the interbank market exists precisely to move the surplus back to wherever the deficit landed, every single business day, for exactly this reason.

WE2: US $: ample reserves, scarce capital, why fresh reserves don't automatically become loans

Objective: show, bank by bank, that a zero (or near-zero) reserve requirement does not remove the capital constraint, and decompose "why didn't the reserves turn into lending" into its two genuinely distinct causes.

Setup. Three US banks (Brightwater Bank, Fenwick Bank, and Northgate Bank (all synthetic)) each receive $50mn of fresh reserves and a matching $50mn deposit, via the Fed buying $50mn of securities from a non-bank customer of each bank (M7.02's own Case B, referenced, not re-derived). The US reserve requirement is 0% (the ample-reserves regime M7.02 documents), so none of this $50mn is "required" reserves at any of the three banks, every rupee of it is, in reserve terms, free to lend against. Assume, illustratively, a 12% total-capital-ratio minimum at each bank (a round figure between the layers it must not be confused with: the ~12–12.5% M5.01 cites for a large, systemically important US bank is a CET1 requirement, and that bank's total-capital requirement runs about 15–16% once the stress buffer and G-SIB surcharge sit on the 8% minimum; a smaller bank's total-capital floor is the 10.5% Basel III baseline of 8% plus the 2.5% buffer, whose CET1 counterpart is 7%; pull each bank's real current requirement from its own disclosure) and a 100% risk weight on new unsecured corporate loans (the Basel standardized-approach default).

Brightwater Bank starts with $660mn of capital against $4,500mn of RWA, a ratio of 14.667%, comfortably above the 12% minimum. Maximum RWA at 12% = 660 ÷ 0.12 = $5,500mn; headroom = $1,000mn, vastly more than the $50mn it just received. Receiving the $50mn of reserves and deposit moves its ratio not at all (0% risk weight), so Brightwater lends $45mn to a creditworthy borrower (holding back $5mn as a precaution), its post-loan RWA is $4,545mn and its ratio is still 14.522%, nowhere near the floor. Brightwater lent because it both could (ample capital headroom) and wanted to (a willing, creditworthy borrower).

Fenwick Bank starts with $504mn of capital against $4,200mn of RWA, a ratio of exactly 12.000%, precisely at the floor. Maximum RWA at 12% = 504 ÷ 0.12 = $4,200mn: headroom = $0. Receiving the $50mn of reserves and deposit again moves nothing (still 12.000%, since reserves carry no risk weight), but Fenwick cannot extend a single dollar of net new risk-weighted lending without either raising fresh capital or shrinking RWA elsewhere, even with its loan officers holding willing, creditworthy applicants. Fenwick's $50mn simply sits as excess reserves. This is the pure capital-constrained case.

Northgate Bank starts with $800mn of capital against $5,500mn of RWA, a ratio of 14.545%, headroom of $1,166.67mn, just as ample as Brightwater's. Yet Northgate also lends $0 of its new $50mn, not because it can't, but because, this quarter, its market has no creditworthy demand for new credit. This is the pure demand-constrained case, and it looks identical from the outside (an unchanged loan book) to Fenwick's capital-constrained case, even though the underlying reason is entirely different.

System-wide. $150mn of fresh reserves entered the system; only $45mn of it became new lending, an effective near-term "multiplier" of 45 ÷ 150 = 0.30, far below one, let alone the old textbook's 1/r blow-up. This is a small, fully decomposed, mechanically exact replica of the real 2008–2015 US experience M7.02 narrates: reserves ballooned, lending did not follow proportionally, and the reason, as here, was never a shortage of reserves. It was scarce capital at some banks and scarce willing demand at others, two separate frictions that a single word like "constrained" too easily blurs together.

WE3: India ₹: a capital raise unlocks lending headroom

Objective: run the leverage-on-capital formula forward (from a capital raise to the new lending capacity it creates) for a named synthetic bank.

Krishna Bank (synthetic) carries ₹3,000 cr of total capital against ₹24,000 cr of risk-weighted assets, a CRAR of 3,000 ÷ 24,000 = 12.50%, comfortably above India's 11.5% minimum (M5.01). Maximum RWA at 11.5% = 3,000 ÷ 0.115 = ₹26,086.96 cr; current headroom = 26,086.96 − 24,000 = ₹2,086.96 cr.

Krishna Bank raises ₹500 cr of fresh equity via a Qualified Institutional Placement (QIP). New capital = ₹3,500 cr. New maximum RWA at 11.5% = 3,500 ÷ 0.115 = ₹30,434.78 cr; new headroom over the same ₹24,000 cr of current RWA = 30,434.78 − 24,000 = ₹6,434.78 cr.

The raise unlocked 6,434.78 − 2,086.96 = ₹4,347.82 cr of incremental new lending capacity, from a ₹500 cr capital raise. Check the leverage relationship directly: ₹4,347.82 cr ÷ ₹500 cr = 8.6957, exactly 1 ÷ 0.115. Every rupee of fresh capital Krishna Bank raised bought it approximately ₹8.70 of new risk-weighted lending capacity (at a 100% risk weight; a lower average risk weight on the incremental book, common for a retail-heavy lender, would stretch that further).

Read it. This is the balance-sheet answer to "why do banks periodically issue equity even when deposits are plentiful?" A bank that wants to grow its loan book beyond what its current capital supports has exactly two structural levers: retain more earnings over time, or raise fresh capital now, and the QIP arithmetic above is precisely what a bank's CFO is solving when sizing how much capital a growth plan actually requires.

WE4: US $ (real, historical): the Reserve Primary Fund breaks the buck

Objective: show a wholesale-money instrument's moneyness failing in practice, with a real, dated, and cited event, and a clean illustrative reconstruction of the mechanism.

On September 16, 2008 (the day after Lehman Brothers filed for bankruptcy) the Reserve Primary Fund, a large, long-established US money-market mutual fund, announced that its net asset value (NAV) per share had fallen to $0.97, down from the $1.00 every prime money-market fund is engineered to hold constant. It became just the second money-market fund in US history to "break the buck" (check against the fund's own SEC filings and the SEC's subsequent money-market-fund reform releases, 2010 and 2014, both adopted substantially in response to this event). The cause was mechanical, not exotic: the fund held Lehman Brothers commercial paper and other Lehman-related debt that lost most of its value the instant Lehman failed, and that loss, spread across every share outstanding, dragged the fund's NAV below $1.00.

The mechanism, reconstructed with clean illustrative numbers (a simplification of the fund's real, far larger and more diversified portfolio, built only to isolate the arithmetic): a money-market fund's NAV per share is simply total assets ÷ shares outstanding. Take a fund with $100 of assets and 100 shares outstanding, engineered to sit at NAV = 100 ÷ 100 = $1.00. To reach the real fund's reported $0.97, its assets need only fall to 0.97 × 100 = $97, a markdown of just $3, or 3% of total assets, concentrated in one holding (the Lehman paper) that most of the portfolio never touched. A 3% markdown on one position, spread across every shareholder, was enough to break a promise that millions of dollars of corporate treasury cash had been resting on as if it were as safe as a bank deposit.

What happened next, briefly (real, and well documented): the break triggered a run on prime money-market funds industry-wide, as investors, fearing other funds might also be holding impaired paper, redeemed en masse; that run froze parts of the commercial-paper market many corporations rely on for routine short-term funding; and the US Treasury announced an emergency temporary guarantee program for money-market funds on September 19, 2008, to stop the panic (check against the Treasury's own historical announcement).

Read it. This is the moneyness hierarchy failing in the most literal sense possible: an instrument sold and held as if it were interchangeable with a dollar in the bank turned out to be a claim on a portfolio, priced the way portfolios are priced, and the promise broke the moment the portfolio's value moved. No government guarantee stood behind it (unlike the insured bank deposit of Worked Example 5 below), which is exactly why MS1.03's shadow-banking chain, and the post-2010/2014 SEC reforms (floating NAV for institutional prime funds, liquidity fees and redemption gates), exist at all.

WE5: India ₹ and US $: quantifying the moneyness of a retail deposit, DICGC and FDIC

Objective: compute, on both sides of R6, the gap between "percent of accounts insured" and "percent of deposit value insured," and connect it to a real 2023 event.

Malwa Bank (synthetic, India). Population A: 15,00,000 retail accounts, average balance ₹80,000 (₹0.008 cr), comfortably under the ₹5,00,000 DICGC limit, so fully insured: total = 15,00,000 × ₹0.008 cr = ₹12,000 cr, all of it insured. Population B: 2,000 large corporate/HNI accounts, average balance ₹3 cr, each account is insured only for its first ₹5,00,000 (₹0.05 cr): total = 2,000 × ₹3 cr = ₹6,000 cr, of which only 2,000 × ₹0.05 cr = ₹100 cr is insured. Total deposits: ₹18,000 cr. Total insured: ₹12,100 cr: 67.22% of deposit value. But 1,500,000 of 1,502,000 accounts (99.87%) are fully covered.

Ridgeline Bank (synthetic, US), deliberately skewed toward large operating balances, mirroring a startup/venture-concentrated depositor base. Population A: 8,000 retail accounts, average $40,000 (fully insured): total = $320,000,000. Population B: 900 large business/operating accounts, average $3,000,000, each insured only for its first $250,000: total = $2,700,000,000, of which only 900 × $250,000 = $225,000,000 is insured. Total deposits: $3,020,000,000. Total insured: $545,000,000: 18.05% of deposit value, meaning 81.95% of this deposit base is uninsured.

The real event this mirrors. Silicon Valley Bank failed on March 10, 2023, with an unusually high concentration of large, uninsured operating deposits from technology and venture-backed clients, commonly reported in the high-80s-to-97% range across contemporaneous disclosures and the two federal post-mortems (the Federal Reserve's April 2023 review of its own supervision, and the FDIC's own material loss review; check the precise figure against those two named reports rather than a secondary summary). On March 12, 2023, the Treasury, the Federal Reserve, and the FDIC jointly invoked the systemic risk exception to guarantee all of the bank's depositors in full, insured and uninsured alike, a discretionary, extraordinary intervention, not a standing legal guarantee.

Read it. Ridgeline Bank's illustrative 81.95% uninsured share is not an extreme, invented outlier, it is the same structural pattern Malwa Bank shows at a gentler skew, and the same pattern SVB showed at a far sharper one: the vast majority of accounts, almost everywhere, are small and fully insured; a much smaller share of the underlying money is. A bank whose deposit base concentrates unusually heavily in large operating or institutional balances is, structurally, a bank whose moneyness rests less on a legal guarantee and more on confidence and, should that confidence fail, on a government's discretionary willingness to intervene, which is precisely the least reliable kind of "par" in the entire hierarchy.

WE6: India ₹: building the aggregate ladder from a consolidated system snapshot

Objective: compute M0, M1, M2, and M3 yourself from raw components, and derive the base-to-broad multiplier bottom-up, rather than reading it off M7.02's table.

Take a synthetic banking-system snapshot (illustrative, for live figures, pull the RBI's Weekly Statistical Supplement, as M7.02's own Reading & Resources points you to): currency with the public ₹34,00,000 cr; bankers' deposits with the RBI ₹6,00,000 cr; other deposits with the RBI ₹20,000 cr; demand deposits with banks ₹28,00,000 cr; time deposits with banks ₹1,80,00,000 cr; savings deposits with post-office savings banks ₹1,00,000 cr. (Simplifying assumption, stated explicitly: banks' own vault cash is treated as negligible here, so currency in circulation is taken as equal to currency held by the public, a real system nets this out precisely, and is not material to the mechanism this exercise teaches.)

Applying M7.02's own RBI definitions directly: M0 = 34,00,000 + 6,00,000 + 20,000 = ₹40,20,000 cr. M1 = 34,00,000 + 28,00,000 + 20,000 = ₹62,20,000 cr. M2 = M1 + 1,00,000 = ₹63,20,000 cr. M3 = M1 + 1,80,00,000 = ₹2,42,20,000 cr.

The multiplier falls straight out: M3 ÷ M0 = 2,42,20,000 ÷ 40,20,000 = 6.02. That number is not a fact you memorized from M7.02, it is a direct consequence of the six raw components above, and if you change any one of them (say, time deposits grow while the base does not), the ratio moves mechanically, with no need for a central bank to "set" it anywhere.

Read it. The base (M0) sits almost entirely on the central bank's own balance sheet; the broad aggregates (M1 through M3) sit on the commercial banking system's, and the ratio between them is just the size of one balance sheet relative to the other, built up, transaction by transaction, by exactly the lending-and-settlement mechanism Worked Examples 1 and 2 walked through by hand.

Practice set

Work each problem before reading its solution. P1–P5 are guided, P6–P12 independent, P13–P16 comprehensive/timed. All entities are synthetic unless a real, dated event is named explicitly.

P1 (guided: classify by moneyness). Rank, from most to least moneyness, and give a one-line reason for each: (a) a ₹500 note; (b) a savings-account balance; (c) a 91-day Treasury bill; (d) a AAA-rated commercial-paper holding inside a liquid mutual fund; (e) a 10-year corporate bond.

Solution. (a) Highest: currency is central-bank money; no promise, no conversion needed. (b) Second, a bank's promise to convert to currency at par on demand, backed (up to the DICGC/FDIC limit) by deposit insurance. (c) Third, a very short-dated, extremely safe government security; close to par but technically priced, not itself directly spendable. (d) Fourth, a corporate IOU marketed and normally traded as "near cash," but with real, uninsured credit and liquidity risk that can break its par-like behavior under stress (Worked Example 4). (e) Lowest (not money at all) a long-dated risk claim priced on discounted cash flows and a credit spread, with no par-on-demand promise whatsoever.

P2 (guided, compute M0). A system reports: currency in circulation ₹18,50,000 cr; bankers' deposits with the central bank ₹3,20,000 cr; other deposits with the central bank ₹15,000 cr. Compute M0.

Solution. M0 = 18,50,000 + 3,20,000 + 15,000 = ₹21,85,000 cr.

P3 (guided: recap the single-bank T-account). A bank extends a ₹25,00,000 loan to a creditworthy borrower. Write the bank's T-account at the instant of lending, and state in one sentence why this is not "lending out somebody else's deposit."

Solution. Assets: +₹25,00,000 loan; Liabilities: +₹25,00,000 deposit (the borrower's new account). The deposit is created by the act of lending, in the same instant, on the bank's own balance sheet (M7.01 §4b), it is not transferred out of any other depositor's existing balance, which is unaffected.

P4 (guided: multi-bank settlement). Beas Bank (synthetic) has ₹600 cr of deposits, a 4% reserve requirement, and ₹25 cr of actual reserves. It lends ₹5 cr, which is fully spent to a customer at Chenab Bank (synthetic), which starts with ₹350 cr of deposits and ₹15 cr of actual reserves. Compute each bank's reserve position (surplus or deficit) after settlement.

Solution. Beas Bank: deposits return to ₹600 cr (required reserves unchanged at ₹24 cr); reserves fall to 25 − 5 = ₹20 cr → deficit of ₹4 cr. Chenab Bank: deposits rise to ₹355 cr (required reserves rise to 4% × 355 = ₹14.20 cr); reserves rise to 15 + 5 = ₹20 cr → surplus of ₹5.80 cr. (Beas can borrow its ₹4 cr shortfall overnight from Chenab, which has more than enough spare.)

P5 (independent: CRR compliance). A bank's net demand and time liabilities (NDTL) are ₹12,000 cr, CRR is 4.5%, and its actual reserves are ₹510 cr. Compute the required reserves and the surplus or deficit.

Solution. Required reserves = 4.5% × 12,000 = ₹540 cr. Actual reserves ₹510 cr → deficit of ₹30 cr, which the bank must fund via the interbank/repo market or a standing facility.

P6 (independent, the capital ceiling). Fairmont Bank (synthetic, India) carries ₹1,800 cr of total capital against ₹15,000 cr of risk-weighted assets, against an 11.5% minimum total capital ratio. Compute its current CRAR, the maximum RWA it can carry, and its headroom.

Solution. CRAR = 1,800 ÷ 15,000 = 12.00%. Maximum RWA = 1,800 ÷ 0.115 = ₹15,652.17 cr. Headroom = 15,652.17 − 15,000 = ₹652.17 cr.

P7 (independent, a capital raise, continuing P6). Fairmont Bank raises ₹200 cr of fresh capital via a QIP. Compute the new headroom and the incremental lending capacity the raise unlocked.

Solution. New capital = 2,000 cr; new maximum RWA = 2,000 ÷ 0.115 = ₹17,391.30 cr; new headroom over the same ₹15,000 cr of current RWA = 17,391.30 − 15,000 = ₹2,391.30 cr. Incremental unlocked = 2,391.30 − 652.17 = ₹1,739.13 cr, check: ₹200 cr × (1 ÷ 0.115) = ₹1,739.13 cr. ✓

P8 (independent, the aggregate ladder, different numbers). A system reports: currency with the public ₹40,00,000 cr; bankers' deposits with the central bank ₹7,50,000 cr; other deposits with the central bank ₹25,000 cr; demand deposits ₹32,00,000 cr; time deposits ₹2,10,00,000 cr; post-office savings deposits ₹1,20,000 cr. Compute M0, M1, M2, M3, and the M3/M0 multiplier.

Solution. M0 = 40,00,000 + 7,50,000 + 25,000 = ₹47,75,000 cr. M1 = 40,00,000 + 32,00,000 + 25,000 = ₹72,25,000 cr. M2 = 72,25,000 + 1,20,000 = ₹73,45,000 cr. M3 = 72,25,000 + 2,10,00,000 = ₹2,82,25,000 cr. Multiplier = 2,82,25,000 ÷ 47,75,000 = 5.91.

P9 (independent: deposit insurance fraction, US $). Fernbrook Bank (synthetic, US) has 20,000 retail accounts averaging $25,000 (fully insured) and 150 large accounts averaging $1,800,000 (each insured only to the $250,000 FDIC limit). Compute total deposits, total insured, and the insured percentage by value.

Solution. Retail total = 20,000 × 25,000 = $500,000,000, fully insured. Large-account total = 150 × 1,800,000 = $270,000,000; insured portion = 150 × 250,000 = $37,500,000. Total deposits = $770,000,000. Total insured = 500,000,000 + 37,500,000 = $537,500,000 → 69.81% insured by value (30.19% uninsured), even though 20,000 of 20,150 accounts (99.26%) are fully covered.

P10 (independent: overnight borrowing cost). A bank ends the day with a ₹3.2 cr reserve deficit and borrows it overnight at 6.75% p.a. Compute the interest cost for the one night.

Solution. Cost = ₹3,20,00,000 × 6.75% × (1/365) = ₹5,917.81.

P11 (independent: moneyness under stress, reasoned ranking). Rank, from most to least likely to hold its par value in a genuine systemic panic, and justify: a fully insured bank deposit; a prime money-market fund share; an uninsured large deposit balance at a solvent bank; physical currency.

Solution. Currency ranks first, it is central-bank money, with no promise involved at all. The fully insured deposit ranks second, a state-backed, structural guarantee up to the limit. The uninsured large deposit balance and the prime money-market fund share both rank below these two, and reasonably arguing either one ahead of the other is acceptable, provided the reasoning is sound: an uninsured deposit is a direct claim on a specific bank that history shows often gets an ad hoc, discretionary full guarantee in a genuine systemic event (Worked Example 5), while a money-market fund share is a claim on a diversified portfolio with no legal guarantee at all and a real, if rare, history of breaking par outright (Worked Example 4). The graded point is not the exact order of the bottom two, it is correctly identifying that the top two carry a structural, legal guarantee and the bottom two do not, resting instead on trust, portfolio quality, or a government's discretionary choice to intervene.

P12 (independent: excess reserves). A bank has NDTL of ₹19,500 cr, a CRR of 4%, and actual reserves of ₹840 cr. Compute required reserves and excess reserves.

Solution. Required reserves = 4% × 19,500 = ₹780 cr. Excess reserves = 840 − 780 = ₹60 cr, available to lend in the interbank market or held earning interest on reserve balances (IORB, M7.02).

P13 (timed, ~8 min: reserves and capital, together). Oakridge Bank (synthetic, India) reports: NDTL ₹8,400 cr, CRR 4%, actual reserves ₹300 cr; total capital ₹950 cr, RWA ₹7,200 cr, minimum total capital ratio 11.5%. Compute (a) its reserve position, (b) its CRAR and capital headroom, and (c) state, in one sentence, which constraint binds today versus over its growth horizon.

Solution. (a) Required reserves = 4% × 8,400 = ₹336 cr; actual ₹300 cr → deficit of ₹36 cr (must be funded overnight). (b) CRAR = 950 ÷ 7,200 = 13.19%, comfortably above 11.5%; maximum RWA = 950 ÷ 0.115 = ₹8,260.87 cr; headroom = ₹1,060.87 cr. (c) Today, the reserve/liquidity constraint binds. Oakridge must borrow ₹36 cr overnight regardless of its healthy capital position; over its growth horizon, capital is not the constraint, it has over a thousand crore of headroom to expand lending well before it would need to raise fresh capital.

P14 (independent, the multiplier as a result, not a lever). A system injects ₹800 cr of new reserves via a central-bank operation, with a 4% reserve requirement. The theoretical multiplier ceiling implies a maximum of new deposits; the actually observed new lending across the system is only ₹950 cr. Compute the theoretical ceiling, the unused gap, and name the two possible causes of the shortfall.

Solution. Theoretical ceiling = 800 ÷ 0.04 = ₹20,000 cr. Actual new lending = ₹950 cr. Unused gap = 20,000 − 950 = ₹19,050 cr (95.25% of the theoretical ceiling unrealized). The two possible causes, exactly as Worked Example 2 decomposed them: capital-constrained banks that would lend if they could but are already near their regulatory floor, and demand-constrained banks that have ample capital headroom but no creditworthy borrowers presenting themselves, the multiplier is whatever the system's actual lending behavior produces, not a number the central bank sets by injecting reserves.

P15 (guided: T-account direction-matching). For each transaction at a single bank, state which account(s) change, on which side, and by how much: (a) the bank extends a ₹10,00,000 loan; (b) the borrower withdraws ₹2,00,000 of the new deposit in cash; (c) the borrower wires ₹5,00,000 to a person at a different bank; (d) the bank repays a ₹3,00,000 overnight interbank borrowing; (e) the central bank buys ₹8,00,000 of securities from this bank.

Solution. (a) Assets: +₹10,00,000 loan; Liabilities: +₹10,00,000 deposit. (b) Assets: −₹2,00,000 reserves (cash leaves the bank into circulation); Liabilities: −₹2,00,000 deposit. M1's total is unchanged, only its composition shifts from "bank deposit" to "currency with the public," both of which are already inside M1. (c) Assets: −₹5,00,000 reserves; Liabilities: −₹5,00,000 deposit (at this bank), mirrored by +₹5,00,000 reserves and +₹5,00,000 deposit at the receiving bank, exactly Worked Example 1's mechanism. (d) Assets: −₹3,00,000 reserves; Liabilities: −₹3,00,000 (interbank borrowing extinguished). (e) Assets: −₹8,00,000 securities, +₹8,00,000 reserves, an asset swap with no new deposit created (M7.02's Case A), because the seller was the bank itself, not one of its customers.

P16 (comprehensive, timed ~10 min). Millbrook Bank (synthetic, India): NDTL ₹5,000 cr, CRR 4%, actual reserves ₹185 cr (before any new lending); total capital ₹640 cr, RWA ₹5,100 cr, minimum CRAR 11.5%. Millbrook extends a ₹120 cr loan, fully spent to a customer at Redwood Bank (synthetic). Compute (i) Millbrook's reserve position after the loan is spent away, (ii) whether its capital position permits the loan (assume a 100% risk weight), and (iii) the overnight interest cost of covering any reserve shortfall at 6.5% p.a.

Solution. (i) Reserves fall to 185 − 120 = ₹65 cr; required reserves, once deposits net back to ₹5,000 cr, are unchanged at 4% × 5,000 = ₹200 cr → deficit of ₹135 cr. (ii) New RWA = 5,100 + 120 = 5,220 cr; new CRAR = 640 ÷ 5,220 = 12.26%, still above the 11.5% minimum, capital does not bind this loan (pre-loan headroom was 640 ÷ 0.115 − 5,100 = ₹465.22 cr, more than enough for a ₹120 cr addition). (iii) Overnight cost on the ₹135 cr shortfall = 1,35,00,00,000 × 6.5% × (1/365) = ₹2,40,410.96. Read it: the loan was entirely fine on capital grounds and entirely necessitated an overnight reserve borrowing, the two constraints, examined side by side on the same transaction, genuinely disagree about how "constrained" Millbrook Bank is today.

Applied mini-project: build the ledger, then verify one bank's real numbers

The task (≈3 hours, two parts).

Part A, build your own multi-bank settlement and capital-ceiling model. In a spreadsheet or in Python, build the balance sheets of at least three synthetic banks (fresh names of your own choosing) and simulate a short sequence of events: at least two loans, each spent away to a customer at a different bank in your set; at least one instance of a bank ending up in reserve deficit and borrowing overnight from a surplus bank; and at least one capital-ceiling calculation (given each bank's capital, RWA, and a stated minimum ratio) showing its current headroom. After every single step, assert, in code, not by eyeballing, that each bank's balance sheet balances (assets = liabilities + equity) and that system-wide reserves are unchanged by any interbank transaction (only your central-bank operation, if you include one, may change the system total). Report the final state of all three banks' reserve and capital positions in one table.

Part B, pull one real primary source and reconcile it against this module's formulas. Choose either (i) the RBI's Weekly Statistical Supplement (or DBIE) and reconcile the actual current M0 (reserve money) and its components against the definitions this module used in Worked Example 6, noting any component you could not cleanly map, or (ii) one real, currently listed bank's (India or US) latest Basel III Pillar 3 disclosure or 10-K capital-adequacy note, from which you extract its actual reported capital, RWA, and capital ratio, and compute its real lending-capacity headroom using this module's ceiling formula, comparing that theoretical headroom against the bank's actual reported loan-book growth over its last year or two. Write two paragraphs: what you could confirm directly from the primary source, and what you could not (a bank's actual internal management capital buffer above the regulatory floor, and its real, forward loan-demand pipeline, are almost never disclosed at the granularity this module's formulas assume), naming explicitly, per R10, why that gap exists.

Scoring rubric (max 20; ≥16 to pass).

Reading & resources

Everything below is an anchor to verify against, never a source to quote a live figure from without checking its date, aggregate and reserve data move weekly.

• Perry Mehrling, "The Economics of Money and Banking" (free, Coursera / YouTube), the anchor for the "money view" and the moneyness hierarchy this module builds on; start here. [Free] [Beginner]
• Perry Mehrling, The New Lombard Street: How the Fed Became the Dealer of Last Resort (Princeton, 2011), the deeper, fuller treatment of the hierarchy and the central bank's role as dealer of last resort. [Paid] [Advanced]
• Bank of England, "Money creation in the modern economy" (Quarterly Bulletin, 2014), the primary source behind "loans create deposits," already the authority for M7.01/M7.02; reread it here with the multi-bank settlement lens this module adds. [Free] [Intermediate]
• RBI, Weekly Statistical Supplement / DBIE (Database on Indian Economy), the free, primary source for live reserve-money and broad-money data; use it for the Applied Mini-Project's Part B. [Free] [Beginner–Intermediate]
• Federal Reserve, H.3 ("Aggregate Reserves of Depository Institutions") and H.6 ("Money Stock Measures") statistical releases, the free, primary US equivalents. [Free] [Beginner–Intermediate]
• Basel Committee on Banking Supervision, Basel III: A Global Regulatory Framework (bis.org), the primary text behind the capital-ratio minimums this module uses. [Free] [Advanced]
• RBI Master Circular/Direction on Basel III Capital Regulations, and the DICGC Act, 1961 (as amended), the primary Indian sources for CRAR minimums and the ₹5,00,000 insured-deposit limit. [Free] [Intermediate]
• FDIC, deposit insurance rules (fdic.gov), the primary US source for the $250,000 insured limit and its ownership-category rules. [Free] [Beginner]
• Federal Reserve, "Review of the Federal Reserve's Supervision and Regulation of Silicon Valley Bank" (April 2023) and the FDIC's own material loss review of SVB, the two named federal post-mortems behind Worked Example 5's real-world case. [Free] [Advanced]
• SEC historical record on money-market fund reform (2010 and 2014 amendments to Rule 2a-7), the primary regulatory response to Worked Example 4's Reserve Primary Fund event. [Free] [Advanced]
• Zoltan Pozsar's shadow-banking and money-market notes: free, practitioner-level "plumbing" writing that previews MS1.03's territory. [Free] [Advanced]
• Stigum & Crescenzi, Stigum's Money Market: the standard, exhaustive reference on money-market mechanics; a shelf book to consult, not read cover to cover. [Paid] [Reference/Advanced]
• CFA Institute curriculum readings on the monetary system and central-bank operations, the standard institutional-exam-level calibration this module is benchmarked against. [Paid] [Intermediate–Advanced]

Teach it back & journal

Feynman prompt. Explain to someone who has only ever thought of "money" as "the number in my banking app" why a bank deposit, a money-market fund unit, and a ₹500 note are not really the same kind of thing, even though all three let you buy a cup of coffee today. Use the hierarchy (currency → bank deposits → money-market/shadow money → everything else) and then walk them through the Reserve Primary Fund's $0.97 NAV specifically: a fund built to always be worth exactly $1.00 per share, marketed and used as if it were cash, that broke that promise because of a 3% markdown on one holding. If your listener can, unprompted, explain why a bank deposit didn't (and structurally couldn't, below the insured limit) break the same way, you have installed the actual mechanism, not just a list of vocabulary. Then add the settlement piece: why can't a single bank just keep creating deposits forever? Walk them through Ganga Bank lending to Kaveri Textiles, the money leaving for Yamuna Bank, and Ganga Bank ending the day short of reserves, in your own numbers, not this module's.

Journal prompt. Think of an instance (your own accounts, a family member's, something you've read about a company's treasury policy) where money was held in a form other than plain cash or a small, fully insured bank deposit: a large account balance, a money-market fund, a fixed deposit above the insured limit, anything with even slightly less than perfect moneyness. Write for fifteen minutes: did you (or they) actually know, at the time, exactly what stood behind that instrument's promise to convert back to cash at par? What would you need to check, and against which primary source (a bank's own Basel III disclosure, a fund's SEC filings, the DICGC/FDIC rules themselves), before trusting a large balance to an institution or instrument you had not examined this closely? Close with this, in writing: the next time an AI tool walks you fluently through "whose balance sheet expands and by how much" in a monetary-policy or banking question, what is the one habit from this module (rebuilding the T-accounts yourself) that stops you from repeating its conclusion before you have confirmed, with your own arithmetic, that every balance sheet involved actually balances?

This module's flashcards and mastery quiz are wired into the app: see the node's Quiz and Reviews.