Mehrling's hierarchy of money

M7.02 and M7.06 each pointed you, in their Reading & Resources, toward Perry Mehrling's "money view" as a going-deeper reading. This is the module that cashes in that pointer, because moneyness is this module's mandate, not a footnote to it.

What are the layers of the hierarchy of money?

The idea. Every instrument that "acts like money" is really somebody's IOU, and IOUs differ in exactly one dimension that matters here: how reliably, how quickly, and at what price they convert into the layer of money above them. Mehrling's term for this is moneyness, not a yes/no property but a degree, and the instruments that carry the most of it sit at the top of a hierarchy:

1. Central-bank money: currency in circulation and commercial banks' reserve balances at the central bank. This is the top of the pyramid because it carries no counterparty risk within its own currency system: a rupee of RBI reserve money is not a promise to pay a rupee, it is the rupee, full stop, and every other layer below is ultimately a promise to deliver this.
2. Bank deposits: a commercial bank's promise to convert your balance into central-bank money, at par, on demand. In ordinary times this promise is so reliable that people forget it is a promise at all: you never think of your savings-account balance as "a claim on a private company" the way you think of a share of stock, even though structurally that is exactly what it is. Two things make it feel like money anyway: deposit insurance up to a limit (see the DICGC limit), and the central bank standing behind the banking system as lender of last resort, a backstop that exists precisely because this promise, unlike currency, can fail.
3. Money-market / "shadow" money: commercial paper (CP), repo, and money-market-fund (MMF) shares. These are IOUs of non-bank or lightly regulated entities, engineered to be short-dated and highly liquid enough that, in normal times, they circulate almost as if they were deposits, a corporate treasurer or a fund manager holds CP or an MMF share instead of a bank deposit because it pays a little more, while still expecting to get "a dollar back for a dollar in" essentially on demand. Nothing government-guaranteed backs this layer: no deposit insurance, and (outside a discretionary, ad hoc rescue) no lender-of-last-resort promise. MS1.03 builds this entire layer (repo, haircuts, the whole shadow-banking chain) in full; here you need only its place in the hierarchy and the fact that its par-like behavior is a market convention, not a legal guarantee.
4. Everything else: bonds, equities, and other financial claims. These are not money at all, by construction: a ten-year bond does not promise to convert into cash on demand at a fixed value; it promises a schedule of cash flows discounted at whatever rate the market currently demands, and its price moves accordingly. Nobody expects to "always get a dollar back" from a share of stock, which is exactly why nobody calls it money.

What does it mean for money to be at par?

What "par" actually means, precisely. An instrument is at "par" when one unit of it converts into one unit of the layer above it, instantly, predictably, with no haircut and no delay. A ₹100 note is always worth exactly ₹100 of central-bank money, because it is central-bank money. A bank deposit is at par with currency as long as the bank is solvent, liquid, and (for retail-sized balances) covered by deposit insurance. A money-market instrument is usually close to par, a CP note or an MMF share is priced to be worth almost exactly what it says on the label, but "usually" and "close to" are doing real work in that sentence, and the entire point of this section is that those two words can fail at the worst possible moment.

What happens to the hierarchy under stress?

The hierarchy is not fixed, it is tested, and it can telescope. In calm markets, a remarkably wide range of instruments trade as if they were interchangeable with money: a corporate treasurer holds CP instead of a deposit and does not think twice; a household holds an MMF share instead of a savings account and calls it "as good as cash." The entire apparatus of modern short-term finance. MS1.03's repo and shadow-banking chain, depends on that "as if" holding up, day after day, for instruments with no legal par guarantee at all. Under stress, the hierarchy telescopes: participants suddenly stop trusting the lower layers to convert at par, rush toward the layers above them (a "flight to quality" or "flight to liquidity"), and an instrument that was priced at $1.00 a week ago can reprice sharply, discontinuously, and all at once. Worked Example 4 walks through the single most famous instance of exactly this (a money-market fund's shares failing to hold par) with real, cited numbers.

Hold the hierarchy loosely on one point: it is a teaching device for relative trust, not a claim that any layer below the top is worthless or that the ranking never shifts. What it buys you, permanently, is the reflex to ask, of any "money-like" instrument someone hands you, a bank deposit, a fund unit, a stablecoin you will meet properly in MS1.04, exactly what does this promise to convert into, and what stands behind that promise if it is tested?

{
  "skill": "Using AI to reason through a 'money view' balance-sheet question — who owes what to whom, whose balance sheet expands",
  "use": "Have a chat assistant draft a first-pass answer to a balance-sheet question (e.g., 'what happens to bank reserves and to broad money if the central bank buys a bond from a commercial bank versus from a mutual fund?'), then rebuild the T-accounts yourself from scratch before trusting any part of the answer.",
  "tools": ["Chat assistants — Claude, ChatGPT, Gemini", "A spreadsheet or a short Python script to hold the actual T-accounts, not the model's prose"],
  "prompt": "Walk me through the balance-sheet entries, step by step, for <SCENARIO>. For each step, state explicitly which account on which entity's balance sheet changes, by how much, and whether it is an asset or a liability. Do not skip any entity involved in the transaction.",
  "verify": "This is exactly the reasoning pattern this module names as the program's highest AI-over-trust risk: the prose can sound completely authoritative while getting a single entry's direction backwards or silently double-counting a deposit that only moved between banks. There is no filing to trace this to (Guardrail #1 has no primary source to point at) — the only real check is Guardrail #3, Recompute: rebuild every T-account yourself, by hand or in code, and confirm assets equal liabilities plus equity on every single balance sheet, at every step, exactly as this module's worked examples do.",
  "diy": "You must already be able to build a single bank's loan-creates-a-deposit T-account cold (M7.01, and Worked Example 1 below) before you can catch an AI-generated version quietly getting the multi-bank settlement step wrong — which is precisely where these answers most often go astray.",
  "market": "IN"
}