M1 vs. M2: What Money Supply Measures Count and the 2020 Jump

M1 and M2 group selected financial assets by how readily they can be used for payments or converted into payment balances. They are monetary aggregates: totals built from defined categories of cash and deposit-like holdings. They are not measures of a household’s net worth, the total value of financial assets, or all the dollars that could eventually be spent. A home, a Treasury bond, and a retirement account do not become part of M1 merely because their owner could sell them.

The labels describe related but different perimeters. M1 is the narrower transaction-oriented measure. M2 includes M1 plus additional liquid savings instruments, subject to exclusions in the Federal Reserve’s definitions. A statistic can change because people move balances between accounts, because institutions report differently, because seasonal adjustment or benchmark estimates are revised, or because the definition itself changes. To interpret a line on a chart, first ask what assets and dates the series covers.

These are U.S. measures maintained in the Federal Reserve’s H.6 release. Other countries use their own aggregate definitions, so an M1 or M2 label should not be assumed to mean exactly the same basket everywhere. The current H.6 release provides the latest U.S. components and data notes.

Under the current U.S. definition, M1 includes currency outside the U.S. Treasury, Federal Reserve Banks, and depository institutions; demand deposits at commercial banks, after specified exclusions and collection adjustments; and other liquid deposits. The last category includes other checkable deposits as well as savings deposits, including money-market deposit accounts. In everyday terms, M1 combines cash held outside banks with certain bank and credit-union balances that are readily accessible.

M2 is broader. It combines M1 with small-denomination time deposits (issued in amounts below $100,000) and retail money-market-fund balances, then subtracts IRA and Keogh balances held at depository institutions and money-market funds from the aggregate. Since July 28, 2026, the Federal Reserve has applied this retirement-balance adjustment to total M2 rather than netting it against each component. The change applies retroactively; nonseasonally adjusted M2 was unchanged, while seasonally adjusted M2 had minor revisions. The H.6 component definitions and H.6 technical Q&As explain the current calculation. A bank money-market deposit account is a deposit; a money-market mutual fund is a fund share.

The phrase “money supply” can make these categories sound like one pile of identical cash. They are not. A dollar of currency, a checking balance, a savings deposit, and a retail money-market fund share differ in legal form and payment access. The aggregates combine them using statistical rules; they do not claim that every component can be spent in exactly the same way at the same moment.

Before May 2020, savings deposits were included in M2 but excluded from M1. In April 2020, after reserve requirement ratios had been reduced to zero, the Federal Reserve removed Regulation D’s six-per-month transfer limit on savings deposits. The Board said this made their liquidity characteristics comparable to transaction accounts for H.6 reporting. It began grouping savings deposits with other checkable deposits as “other liquid deposits” and including that combined category in M1.

The accounting perimeter therefore changed. A very large pool of savings deposits that was already included in M2 moved inside M1. The Board explicitly said the reclassification would significantly increase measured M1 while leaving M2 unchanged from the classification change itself. That is why the M1 series has a pronounced step around May 2020: the full step should not be read as a same-sized cash transfer to households or as a sudden equivalent increase in money created that month.

This does not mean that no real balances changed in 2020. Households and businesses also changed deposits, cash holdings, and fund balances during the pandemic period. The useful distinction is between (1) an actual change in the amount of an asset and (2) the statistical decision about which aggregate counts that asset. The Federal Reserve’s 2020 H.6 announcement explains the regulation and the expected aggregate effect; its H.6 technical Q&As document the May 2020 treatment and later methodology updates.

Consider a simplified household moving $1,000 from a savings deposit to a checking account, with no other balance changes. Before May 2020, the checking balance counted in M1 while the savings balance did not. Under that old definition, the transfer would raise M1 by $1,000. Both balances were already inside M2, so M2 would not change from the transfer.

Under the definition used from May 2020 onward, both the checking balance and savings deposit are included in M1. Moving the same $1,000 between those two accounts therefore changes neither M1 nor M2 by itself. The household still owns the same total deposit balance. Only the location of the balance inside its own accounts has changed.

This example isolates classification from economic activity. A deposit can also rise because a bank extends credit, a payment arrives, or funds move from an asset outside the aggregate; it can fall when balances move the other way. Those are different transactions. When explaining a historical movement, say whether you mean a definitional reclassification, a transfer among included categories, or a net change in the total assets counted.

The Federal Reserve publishes money-stock components in H.6, and FRED distributes seasonally adjusted series such as M1 (M1SL) and M2 (M2SL). The current FRED series are monthly averages, seasonally adjusted, and expressed in billions of dollars. Those properties matter: a monthly average is not a point-in-time bank balance, and billions are not millions or percentages.

The series notes point readers to the 2020 change in the M1 definition. Later benchmark revisions can also alter historical observations as the Federal Reserve incorporates more complete reporting. For example, the August 2021 H.6 release revised deposit data starting in 2020 for banks and thrift institutions that were not weekly deposit reporters, retail money-market mutual-fund data starting in 1996, and foreign-currency exclusions using a new quarterly benchmark. Those revisions changed estimates of earlier periods; they were separate from the 2020 M1 reclassification. Therefore two screenshots or downloaded spreadsheets can show different historical values if they were retrieved at different times or use different seasonal-adjustment vintages. Record the series ID, retrieval date, unit, seasonal-adjustment status, and observation frequency when preserving a chart or calculation.

Seasonal adjustment is another distinction from the simplified transfer example. H.6 seasonally adjusts currency, demand deposits, and other liquid deposits separately when constructing M1. For seasonally adjusted M2, it adds seasonally adjusted small time deposits and retail money-market-fund balances to seasonally adjusted M1, then subtracts IRA and Keogh balances on a nonseasonally adjusted basis. The July 2026 change moved that deduction from the individual components to the aggregate; it did not change nonseasonally adjusted M2 and caused only minor revisions to seasonally adjusted M2. This statistical treatment does not change a depositor’s balance. The current H.6 release and its technical Q&As document the construction.

Thus, the $1,000 transfer shows which category counts under each definition for raw balances; it does not promise a $1,000 movement in the reported seasonally adjusted M1SL series. That series is a monthly average of adjusted components and may also reflect changes in other balances. To reproduce a growth rate, compare consecutive observations from the same series and retain its units, seasonal-adjustment status, and data vintage. FRED’s M1 series notes describe those details.

Do not compare a pre-May 2020 M1 value with a post-May value as though its perimeter were constant. The H.6 revision updated data under the new definition beginning in May 2020; it did not add savings deposits to pre-May M1. For a comparison that crosses the break, flag the definition change or construct comparable values from the component data, and identify the data vintage. For comparisons within a period that uses one definition, use the latest FRED/H.6 data. The H.6 technical Q&As document the May 2020 boundary and subsequent methodology changes, while FRED’s M1 series notes describe the series. M2 did not undergo the same savings-deposit reclassification because savings were already included in it, although later data revisions can still affect measured levels.

A stock is measured at a date or averaged over a period; a growth rate compares two observations. The simple percentage-change formula is:

Percentage change = (new value − old value) ÷ old value × 100

For a hypothetical example, if a monthly average rose from $20.0 trillion to $20.2 trillion, the change would be ($20.2 − $20.0) ÷ $20.0 × 100 = 1.0% over that comparison period. This arithmetic does not say whether the increase came from new deposits, cash demand, reclassification, or a revised estimate. The units and series definition still determine what was measured.

Also check whether a quoted growth rate is year over year, month over month, or annualized from a shorter interval. A 1% change in one month is not the same as 1% over twelve months. If that one-month pace were compounded for twelve months, the hypothetical annualized rate would be (1.01)^12 − 1, about 12.7%; it would not be a forecast that the pace will persist. Federal Reserve releases sometimes present changes at seasonally adjusted annual rates for specified shorter windows, so read the column label and dates before comparing figures.

No. An increase in a monetary aggregate alone does not establish that consumer prices will rise by a particular amount, or at a particular time. Inflation indexes track prices paid for baskets of goods and services; M1 and M2 count specified financial balances. Those are different measurements. People can hold more deposits without immediately spending them, and the relationship between measured money, spending, output, interest rates, and prices can change over time.

The identity often written as M × V = P × Y relates a money measure (M), its velocity (V), a price level (P), and real output (Y). It is useful as an accounting framework, but it does not by itself tell you that velocity is fixed or that a change in M mechanically causes a matching change in P. You need to define the money measure and time period and examine what happened to spending, output, and the demand to hold liquid balances. An identity is not a stand-alone forecast model.

For context, compare the CPI, PCE, and GDP deflator, which measure price changes in different baskets, and nominal versus real GDP, which separates current-dollar output from an inflation-adjusted measure. If your question is about falling prices rather than money balances, see disinflation versus deflation. None of those measures can be substituted for M1 or M2.

M1 and M2 are not bank reserves, the monetary base, or a direct reading of the Federal Reserve’s policy stance. The H.6 release lists reserve balances and the monetary base separately from M1 and M2; the base includes currency in circulation plus depository institutions’ reserve balances at Federal Reserve Banks. Bank reserves are balances banks hold at the Fed, not household deposits included in M1 or M2. Mixing these series can create a false comparison between the public’s money-like assets and banks’ settlement balances.

A rising aggregate also does not identify who holds the balances, why they changed, how quickly they will be spent, whether credit conditions are easing, or what will happen to prices. M1 and M2 help describe the scale and composition of selected liquid assets. To draw a broader conclusion, state the definition and period, inspect component series, account for the 2020 M1 break, and compare with spending, output, credit, and price data that answer separate questions.