The M2 Money Supply Manual

Research note. Computed at render from our committed monthly M2 series (Fed H.6 via FRED M2SL, 1959+), CPI and the SPY/M2 ratio file. Prints are dated by initial release, not reference month. See How we checked it.

What is the M2 money supply?

M2 is the Federal Reserve’s broad count of spendable money in the United States: physical currency, checking-account deposits, savings deposits, small time deposits under $100,000, and retail money-market funds. It answers one question — how many dollars can households and businesses put to work quickly — and it currently answers it with $23.2T, about 81 times the $0.3T the series started at in 1959. What M2 excludes matters too: institutional money funds and large time deposits (once tracked as M3, discontinued in 2006), and anything already invested in stocks or bonds — money that bought an asset became someone else’s deposit, which is why “cash on the sidelines” framing needs care (our sideline cash ratio covers that trap).

The latest reading

The June 2026 report puts M2 at $23.2T, growing +5.5% year over year against a 6.6% median since 1960 — about +2.0% after inflation. Growth has normalized from both tails of the past six years: the 26.7% record of February 2021 and the -4.7% record contraction of April 2023. The next H.6 release lands around the fourth week of August 2026; the live chart is on our money supply page.

How the number is built

The Fed publishes M2 in its monthly H.6 “Money Stock Measures” release as a seasonally adjusted monthly average; our series is FRED’s M2SL mirror, and each row carries its initial publication date, so charts show when the market learned each number. Two definitional breaks are worth knowing. In May 2020 the Fed abolished the savings-account transfer limit, which moved savings deposits into M1 — M1 jumped severalfold overnight while M2 was unaffected (it already contained savings deposits). And in early 2021 the H.6 moved from weekly to monthly publication, which is why older commentary tracks weekly M2 wiggles that no longer exist.

The pandemic surge — 41% in twenty-six months

Between February 2020 and April 2022, M2 rose from $15.5T to $21.8T — $6.3T of new money, a 41% expansion — as pandemic fiscal transfers landed in deposit accounts while QE ran at full speed. Year-over-year growth peaked at 26.7% in February 2021, roughly double the fastest prints of the inflationary 1970s. Nothing else in the series’ 67 years resembles it, which is exactly why it carries most of the statistical relationship between M2 and inflation — the next section’s honest caveat.

The only contraction on record

From the April 2022 peak, M2 fell 4.9% to its October 2023 trough as quantitative tightening drained reserves and deposits chased money-market yields. Year-over-year growth printed negative for 15 consecutive months (December 2022 through February 2024) — the only negative prints in the entire 1959+ series. The scary charts of 2023 compared that decline to the 1930s; the honest frame is that it retraced a fraction of a 40% surge, and the level regained its old peak in April 2025. Depression analogies never fit a money stock that remained a third above its pre-pandemic trend line.

Does M2 growth predict inflation? The two-era answer

The textbook says more money chasing the same goods raises prices. Our data says the textbook works only when M2 goes to extremes. We correlated M2 growth with CPI inflation at leads of 0, 12 and 24 months, twice: over the full sample, and over 1990–2019 alone.

The full-sample correlation builds to r ≈ +0.43 at a two-year lead — and the 1990–2019 column shows where it comes from. For thirty years in the middle, faster M2 growth did not precede higher inflation at any horizon; the full-sample result is carried by the 1970s and by 2020–22, when the 27% surge preceded the worst inflation in four decades (CPI peaked around 9% roughly fifteen months after M2 growth peaked). The honest reading: M2 is a fire alarm, not a thermostat — informative at extremes it visits a few times a century, nearly silent in between. Velocity is the missing variable, and it is not stable.

Does M2 drive stocks?

“Liquidity drives markets” is the standard bull-case shorthand, and M2 is usually the liquidity series behind it. The long-run numbers are less obliging: since 1993, SPY has grown roughly 2.5× faster than M2 — the SPY-to-M2 ratio has risen from 13 to 32 — so the market is emphatically not just floating on the money stock. Where M2 earns its place is as a denominator: stocks priced in M2 units is one of the Bubble Tracker’s valuation reads, and margin debt as a share of M2 scales leverage to the economy’s cash base. As a timing signal, M2 growth told you little about forward equity returns outside the same extremes that made it useful for inflation.

Where it will mislead you

Four traps, in the order people fall into them. The level always goes up — an 81-fold rise since 1959 mostly restates that the economy grew and prices rose; only growth rates and ratios carry signal. Nominal growth is not purchasing power — the current 5.5% nominal print is +2.0% in real terms. Base effects manufacture drama — the 2023 “fastest M2 collapse since the Depression” charts were true and misleading at once, a 4.9% dip after a 41% surge. And definitional breaks bite: M1 comparisons across May 2020 are meaningless, weekly-M2 commentary died with the 2021 H.6 change, and anyone quoting M3 is using a series the Fed retired in 2006.

The last 12 prints

How we checked it

Every figure on this page is computed at render from three committed files: the monthly M2 series (FRED M2SL with initial release dates), CPI, and the SPY/M2 ratio. Real growth deflates the M2 year-over-year change by CPI over the matching window at the latest shared month. The inflation test is a plain Pearson correlation of M2 YoY against CPI YoY led by k months; overlapping monthly windows are not independent observations, and the two-era split is reported precisely because the pooled number alone overstates the everyday relationship. Surge and contraction episodes are level extremes located mechanically inside fixed windows, not hand-picked dates.

Frequently asked questions