The Counting Rule: Liquidity, Use, and the Four Measurement Traps
The subject of this chapter is the decision rule behind M1 and non-M1 M2: what an asset does for its holder, how quickly it becomes spendable cash, and whether its balance can be measured without distortion. That rule is more useful than asking whether an asset is “digital” or “on-chain.” Applying it to tokenized deposits, tokenized money-market funds, and payment stablecoins gives the reader a practical way to tell a genuinely new monetary question from a familiar claim using a new transfer rail.
M1 is about payment readiness, not transfer speed
The first distinction is between a medium of exchange and a store of value. In the Federal Reserve’s framework, a medium of exchange is highly liquid, immediately available for transactions, used for payments, and generally low-yielding. That profile points toward M1. A store-of-value asset can also be safe and liquid, but it is primarily held for short-term saving, earns a return closer to market rates, or must be redeemed before it becomes ordinary spending money. That profile points toward non-M1 M2.
The distinction is functional rather than absolute. An asset may be easy to sell and still not be ready to serve as payment. Suppose a person owns a share in a fund whose market is open and whose price is stable. The person can transfer the share quickly, but a merchant accepting dollars does not thereby receive the fund’s underlying cash. The share must be redeemed, and the redemption process introduces time, procedures, and possibly a different intermediary. Transferability describes how a claim moves. Payment readiness describes whether the recipient can use the claim as money without another conversion step.
This is why the word “instant” needs a subject. A blockchain can make the transfer of a token instant while leaving the conversion of that token into dollars unchanged. A bank can make a demand deposit available through a new settlement system without changing the fact that the deposit is a bank liability payable on demand. A money-market fund can make its shares transferable around the clock while remaining a savings product whose dollars arrive after redemption. The speed of the ledger is one property in the chain, not the classification rule itself.
Yield is another clue, though not a mechanical verdict. A checking-style balance is held mainly to spend, so its low or nonexistent return is consistent with M1. A fund share that passes through the interest earned on short-term securities is more naturally a store of value. Payment stablecoins covered by the Federal Reserve note cannot pay interest directly to holders, but the note still treats their actual use as decisive: use for crypto trading and short-term investment can resemble saving, while widespread use for ordinary payments could make them more like a medium of exchange.
The word “dominant” matters here. People can use the same asset for more than one purpose. A stablecoin may be held between speculative trades, sent across a border, used as collateral, or offered to a merchant. Classification asks which function best describes the asset’s role at the relevant time and scale. If that role changes, an aggregate definition may need monitoring or reconsideration. The point is not to assign a permanent label to a token, but to describe the economic behavior that the label is meant to summarize.
Legal status and backing still matter, but they answer different questions. A tokenized deposit can be a bank deposit because it remains on a bank’s balance sheet; that fact helps explain its legal relationship and the protection attached to it. It does not replace the functional test. A tokenized money-market fund is a fund share, not a demand deposit, even if the fund holds low-risk instruments. A payment stablecoin is a privately issued dollar-linked claim whose reserves and redemption promise affect its reliability, but its eventual M1 or M2 placement still depends on how holders use it and how quickly they can turn it into dollars.
A useful diagnostic therefore asks three questions in order. Can the holder use the asset directly to pay a third party? If not, can the holder convert it to cash immediately, without selling an investment or waiting for a fund process? And is the asset primarily held for transactions or for a return-bearing store of value? The answers do not settle every case, but they prevent the common mistake of treating a fast transfer as proof of M1 liquidity.
The distinction can be applied without knowing the brand of a token. A demand-style tokenized deposit is likely to answer yes to direct payment access and immediate cash conversion. A tokenized money-market share may answer yes to rapid transfer but no to immediate cash conversion. A payment stablecoin may offer direct on-chain payment, yet its classification can remain conditional if its principal use is still crypto settlement or saving and its practical redemption terms are unsettled.
The four tests that turn a classification into a statistic
Functional classification is only the first part of the Federal Reserve’s analytical approach. A statistic also needs a dependable way to observe the balance. The first practical test is data availability: is there reliable, timely information about how much of the asset exists? A theoretical claim that belongs in M1 is not enough if nobody can measure its outstanding amount at the frequency required for the aggregate.
The second test is reporting infrastructure. A balance can exist and even be visible on a ledger without arriving in a consistent statistical form. A reporting system needs common definitions, a regular reporting schedule, and a way to combine information from different institutions without changing the meaning of the total. Data availability asks whether the information exists; reporting infrastructure asks whether the information can be collected and aggregated systematically.
The contrast among the three asset types makes this difference concrete. Depository institutions already report tokenized deposits together with traditional deposits. That means the current aggregate can capture the balance through an existing bank-reporting channel, even though the data do not separately reveal how much activity uses blockchain rails. The measurement system is not blind to the deposit, but it is less informative about the adoption of tokenized banking.
Tokenized money-market funds have a different reporting path. The Investment Company Institute supplies net-asset data and distinguishes retail from institutional funds, but its routine information does not identify whether a fund has been issued in tokenized form. The aggregate can therefore include the fund category without showing the tokenized subset. This is a visibility gap, not necessarily a missing-money gap.
Payment stablecoins make the reporting question more explicit. The Federal Reserve note says that the GENIUS Act requires permitted issuers to report their reserve holdings monthly. Those reports could provide a foundation for estimating the amount in circulation because the stablecoins are supposed to be backed one-to-one by reserve assets. But a foundation is not a finished statistical system. Issuers would still need standardized reporting formats, and policymakers would need to decide whether information should flow directly to the Federal Reserve or through another regulator. A ledger may show individual transfers while still failing to provide a uniform, timely measure of the outstanding monetary claim.
The third test is double-counting risk. A token and its backing can look like two assets even when the backing is already represented in an existing aggregate. The Federal Reserve uses retail money-market funds to illustrate the issue: a fund’s net assets can include bank demand deposits, which are already part of M1. The risk is considered relatively small in that example because demand deposits yield little and funds have an incentive to minimize them, but “small” is not the same as “zero.” The composition of the backing determines how much adjustment is needed.
Payment stablecoins make the overlap harder to evaluate because their reserves can include bank deposits, government money funds, and Treasury securities. Deposits and money-fund assets are already represented in parts of M1 or M2. If an analyst simply adds stablecoin balances to the existing total, the result could count a private digital claim and the assets supporting that claim without deciding whether both represent distinct money available to the economy. The stablecoin is real; the problem is whether its reserve has already been counted for another reason.
The fourth test is geographic scope. A U.S. monetary aggregate is not automatically a census of every dollar-linked claim used anywhere in the world. The Federal Reserve note distinguishes globally circulating currency from deposit-related components defined around holdings at banks in the United States. Dollar deposits at banks outside the United States are excluded from those deposit components. The definition follows the kind of claim and the measurement purpose, not merely the currency in which the claim is denominated.
Public blockchains complicate that boundary. A payment stablecoin issued by a U.S.-regulated entity can be held in a U.S. wallet, sent by an overseas business, or transferred between pseudonymous addresses whose users cannot be identified from the transaction alone. Reserve reports may indicate how many tokens have been issued, but they do not automatically reveal where the economic activity occurs. Counting global circulation as U.S. money would answer a different question from measuring money available through the U.S. financial system.
The geographic issue also changes how the other tests should be interpreted. A reporting system may know the issuer and the reserve account while knowing little about the final holder. Conversely, a public ledger may reveal a large number of transfers while providing no reliable measure of whether those transfers represent residents, intermediaries, automated activity, or claims being held idle. The relevant denominator is not always “all tokens visible on-chain.” It is the population and type of claim that the aggregate is designed to describe.
The four tests also point to different remedies. If the problem is data availability, observers need a dependable balance measure. If the problem is reporting infrastructure, they need common fields and a regular reporting channel. If the problem is double-counting, they need information about reserve composition and a rule for handling overlap. If the problem is geography, they need better attribution of economic activity—or a clear decision not to treat a global balance as a U.S. deposit measure. Better blockchain analytics may help with the last question, but the Federal Reserve note does not present a completed method for turning pseudonymous activity into a domestic statistic.
This is why “include the token” is too blunt a policy instruction. Tokenized deposits may need an additional indicator that separates blockchain adoption from the existing deposit total. Tokenized money-market funds may need a way to identify the tokenized subset without moving the whole fund category into M1. Payment stablecoins may require issuer reports that connect circulation, reserve composition, and regulated status, followed by an explicit decision about which reserve assets are already represented in M1 or M2. The right response to a visibility gap can be better reporting rather than a new aggregate component.
The practical sequence is visible in the path from a holder to an aggregate:
The sequence also shows why the four tests should not be collapsed. An asset can pass the function test and fail the reporting test. It can have good reporting and still raise a reserve-overlap problem. It can be fully observable in aggregate and still leave the location of users unclear. These are different forms of uncertainty with different remedies.
What evidence would justify monitoring, inclusion, or reclassification? At minimum, observers would need evidence about the asset’s dominant use, its cash-conversion path, its outstanding balance over time, and the institutions responsible for reporting it. They would also need to inspect reserve composition, determine whether related assets already enter M1 or M2, and define the geographic population the statistic is intended to describe. For payment stablecoins, the Federal Reserve note supports monitoring and preparation, not a settled placement. It explicitly leaves open whether future payment use would justify M1 or whether current saving and trading use would favor non-M1 M2.
That is the central discipline of the counting rule. Classification begins with economic function, but a monetary statistic is credible only when the balance can be observed, reported, de-duplicated, and located for the purpose at hand. The next comparison applies those tests to the three asset types side by side, where the differences between a bank liability, a fund share, and a privately issued stablecoin become harder to mistake.