Velocity of Money

What velocity actually counts

Imagine a single dollar bill issued in January. Over the course of the next twelve months it might pay for a coffee, then a barista's bus fare, then a driver's lunch, then a cook's groceries, and so on. If by December it has been spent on final goods and services seven times, it has a velocity of seven for that year. Average that over every dollar in circulation and you have the velocity of money for the economy.

The fundamental measurement is from the equation of exchange. Total nominal spending on final output in a year equals price level times real output, PY — what national accounts call nominal GDP. That same flow of spending must also equal the money stock M times the average number of times each unit changes hands, V. So

M × V = P × Y          (equation of exchange)
V = PY / M             (definition of velocity)

The equation is an accounting identity. If you accept the definitions of M, P, and Y, then V is determined exactly. There is no measurement of velocity independent of this calculation — you don't go out and count transactions and divide by the money stock. You compute V as the residual that makes nominal GDP equal to MV.

Which is a feature, not a bug. The residual contains real economic information: it tells you how intensively the existing stock of money is being used to fund spending. A society where everyone runs through their paycheck quickly has high V. A society where everyone parks savings in checking accounts has low V.

The US M2 velocity series, 1960 – present

The Federal Reserve publishes M2 velocity as the ratio of quarterly nominal GDP to quarterly M2, annualized. The series tells a clean story in four chapters.

Period	Approximate M2 V	Regime
1960 – 1980	1.7 – 1.9	Bretton Woods, regulated banking, Phillips-curve era
1980 – 1997	1.7 → 2.2 (rising)	Volcker disinflation, financial deregulation, cards spread
1997 – 2008	2.2 → 1.7 (drifting down)	Falling rates, savings glut, dot-com to housing boom
2008 – 2019	1.7 → 1.4	QE, zero lower bound, banks hold excess reserves
2020 – 2024	1.1 – 1.3	COVID stimulus, deposit surge, then partial recovery

The 1997 peak is the highest reading in the postwar record. By Q2 2020 the series printed 1.10 — a level not seen since the 1940s. Every textbook diagram of monetary policy from the 1980s was built around an assumption — that M2 V hovered near 1.8 with small cyclical variation — that the next twenty years systematically broke.

What moves velocity

Velocity is not exogenous noise; it responds to identifiable forces. Four matter most.

1. Payment technology. Credit and debit cards, ACH transfers, mobile payments, and instant-settlement rails all reduce the average balance you need to hold to support a given level of spending. A consumer who carried $200 cash in 1975 might only need $40 in a checking account today, because she can borrow $300 against her card for the lag between paycheck and bill. Faster turnover of the money you do hold raises V. This is the dominant explanation for the 1980 – 1997 rise.

2. Interest rates. The opportunity cost of holding non-interest-bearing money is the rate you forgo on a near-substitute (Treasury bills, money market funds). When short rates are 8 percent — as in the early 1980s — households and firms work hard to economize on idle cash, which raises V. When rates are at the zero lower bound — as from 2008 to 2022 — the opportunity cost is essentially zero, and money piles up in deposits. This is the dominant explanation for the post-2008 fall.

3. Confidence and expected inflation. In a hyperinflation, V can rise tenfold or more as people refuse to hold rapidly depreciating cash. In a deflationary panic, the reverse: holding cash earns a positive real return relative to falling-price assets, and V falls. Confidence in the banking system also matters — bank runs sharply lower V as people withdraw deposits and sit on currency rather than spending it.

4. Demographics and savings preferences. Aging populations save more relative to spending, raising the demand to hold money and lowering V. Japan's velocity has been low and falling for thirty years partly for this reason. Wealth concentration matters too: high-income households have lower marginal propensities to spend out of liquid balances than lower-income households, so a shift of wealth upward reduces V at the aggregate level.

Which M? The choice matters

"Velocity" is not a single number. It depends on the money aggregate. The standard US definitions are layered:

Aggregate	Includes	2024 stock ($T)	2024 V
MB (base)	Currency + bank reserves at Fed	~5.5	~5.0
M1	Currency + checkable deposits + savings (post 2020)	~18	~1.6
M2	M1 + small time deposits + retail money funds	~21	~1.3
MZM (discontinued)	Zero-maturity money — broader than M2 minus time deposits	—	—

Notice that V is roughly inversely proportional to the breadth of the aggregate, which is mechanical: broader aggregates include more savings-like balances that turn over rarely. A 2020 redefinition moved most savings deposits from M2 into M1, which is why M1 V looks oddly low after that date — it is not a behavioral change but an accounting one.

The Fed has not maintained a single preferred "money" measure since the late 1980s. The Divisia monetary aggregates — index numbers that weight each component by how money-like it is in transactions — give a smoother and more economically meaningful V than the simple-sum M2, but they are not part of standard reporting.

Worked example: decomposing 2020

In Q4 2019, US nominal GDP was $21.8 trillion (annualized) and M2 was $15.4 trillion. So V was

V = PY / M = 21.8 / 15.4 = 1.42 turnovers/year

One year later, in Q4 2020, nominal GDP was $21.5 trillion (a slight decline as the COVID recession outweighed the late-year recovery) and M2 had jumped to $19.5 trillion. Velocity:

V = 21.5 / 19.5 = 1.10 turnovers/year

M2 grew about 27 percent in one year — the largest annual rise on record. Nominal GDP fell slightly. The arithmetic of V = PY/M forces velocity to drop by about 22 percent. The economic interpretation: stimulus payments and forced-savings deposits piled up in checking and savings accounts at a rate the spending side of the economy could not absorb. The dollars existed; they were just not moving.

From 2021 onward, as the economy reopened and inflation accelerated, V partly recovered (back to about 1.30 by 2023) but never returned to its 1990s level. The shift was structural, not just cyclical.

Why this killed the 1980s monetarist experiment

Milton Friedman's policy prescription was the k-percent rule: the central bank should grow the money supply at a fixed rate k each year, where k is chosen to deliver low and stable inflation given real growth. The argument relied on two empirical claims about velocity.

1. V is roughly constant over time, varying smoothly with technology and institutions.
2. V responds predictably to interest rates, so a money-targeting rule still works through standard money-demand functions.

In October 1979, the Volcker-led Federal Reserve switched its operating target from the federal funds rate to non-borrowed reserves, effectively making M1 growth the policy variable. The rationale was Friedmanite: kill inflation expectations by demonstrating quantitative discipline. The next three years produced two recessions, unemployment above 10 percent, and a fall in CPI inflation from 13.5 percent in 1980 to 3.2 percent in 1983. Monetarism's central claim — that disinflation is achievable through monetary discipline — was vindicated.

But the joint stability of M and V was not. Even during the Volcker disinflation, M1 V swung erratically as households shifted between checking deposits, NOW accounts, money market funds, and time deposits. By 1982 the Fed was already de-emphasizing M1 in favor of M2; by 1987 it was using neither as a hard target. In 1993 Alan Greenspan formally told Congress the Fed no longer believed monetary aggregates contained reliable signal. Inflation targeting using the federal funds rate — first New Zealand in 1990, then most developed-world central banks by 2000 — became the operational successor.

The deep reason was simple. Financial deregulation in the 1980s — money market funds, the elimination of Regulation Q deposit-rate caps, the proliferation of sweep accounts that automatically moved corporate balances from non-interest-bearing checking to interest-bearing instruments overnight — fundamentally altered the relationship between any specific money aggregate and aggregate spending. The aggregate measure became a moving target. Even if you controlled M perfectly, V drifted enough to break the link between M and PY.

Velocity in extremis: hyperinflation

The most spectacular fact about velocity is what it does in hyperinflations. When inflation runs 50 percent per month or more — Cagan's classical threshold for hyperinflation — the cost of holding currency for even a few days exceeds any plausible transaction benefit. V rises sharply. This is not a marginal effect: in Hungary 1946, V is estimated to have reached values several thousand times its peacetime norm before the pengő was finally retired and replaced with the forint.

The feedback is dangerous. Higher V means higher PY at unchanged M, which is realized as higher P. Higher P validates the inflation expectations that drove V up in the first place. Without an anchor — usually a credible commitment to stop monetizing fiscal deficits, often paired with a currency reform — the spiral does not self-arrest. The Weimar Germany hyperinflation of 1923 ended only when the Rentenmark was introduced with strict backing; the Zimbabwean hyperinflation of 2008 ended when the country effectively dollarized.

Episode	Peak monthly inflation	Velocity behavior
Weimar Germany, 1922–23	~29,500 % (Oct 1923)	V rises ~100× peacetime; workers paid twice daily
Hungary, 1945–46	~4.19 × 10¹⁶ % (Jul 1946)	Highest V on record; price-doubling time hours
Yugoslavia, 1992–94	~313,000,000 % (Jan 1994)	Currency replaced six times; instant pass-through
Zimbabwe, 2007–08	~7.96 × 10¹⁰ % (Nov 2008)	Dollarization absorbed transactional balances
Venezuela, 2017–19	~196 % monthly	Dollar substitution; bolivar V approaches infinity

The takeaway: velocity is a feature of money demand, and in extreme inflation, demand collapses faster than the supply can be reduced. That is why "fight inflation by tightening M" works in normal regimes but cannot, alone, end a hyperinflation once V has unmoored.

Velocity vs. the money multiplier

These are persistently confused because both involve money and both fell after 2008. They are not the same.

	Velocity (V)	Money multiplier (m)
Definition	V = PY / M	m = M / MB
Ratio type	Flow to stock	Stock to stock
What it measures	How fast money is spent	How much broad money the banking system creates per dollar of base
Units	1 / year (turnovers)	Dimensionless
2008 behavior	Fell from 1.9 → 1.5 (slowly)	Collapsed from ~9 → ~3 in months
What controls it	Money demand: rates, technology, confidence	Reserve requirements, lending appetite, IORB

The 2008 multiplier collapse is its own story: when the Fed introduced interest on reserve balances (IORB) in October 2008 and then injected trillions through quantitative easing, banks chose to hold the new base money as reserves at the Fed rather than lend it out. So the monetary base exploded while M2 grew much less; m fell from 9 to 3 essentially overnight. Velocity then fell more slowly over the following decade as the broad money created sat in deposits rather than circulating. Both ratios were doing real work, but on different parts of the chain MB → M → PY.

Modern central banking: V as residual, not target

No major central bank today targets money supply directly. The dominant frameworks are:

• Inflation targeting — set the policy rate to keep CPI or PCE inflation near a stated target (typically 2 percent in advanced economies). The Fed, ECB, Bank of England, Bank of Japan, and most emerging-market central banks operate variants.
• Dual mandate (Fed) — joint targets for inflation and maximum employment, implemented through the federal funds rate.
• Average inflation targeting — the Fed's 2020 framework refinement: tolerate inflation overshoots after periods of undershoot. Effectively abandoned in 2022 once inflation overshot.
• Forward guidance and balance-sheet policy — communicating expected future rates and adjusting reserve quantities. Velocity does not appear as a target variable.

In all of these, V is treated as endogenous: a residual that adjusts as households and firms respond to interest rates, expectations, and payment options. Policy works through those underlying drivers — primarily the policy rate — not through the velocity number itself. This is a cleaner framework precisely because controlling V directly is not possible.

It also means that in a deflationary trap — when the policy rate is at zero and inflation expectations are flat — falling V is symptom and constraint rather than something the central bank can engineer up. Japan after 1995 illustrates the regime: zero rates, falling V, and persistent deflation that ordinary monetary tools could not durably reverse.

Common pitfalls

• Treating MV = PY as a theory. It is an accounting identity. Theory enters only when you make claims about how M, V, P, and Y move jointly. Reading "M up therefore P up" out of the equation without justifying V and Y is the original sin of bad quantity-theory polemic.
• Confusing M growth with inflation. M can grow rapidly while inflation stays low if V falls just as fast. The decade after 2008 is the largest macroeconomic example: M2 doubled, inflation stayed near 2 percent, because V essentially halved.
• Conflating velocity with the multiplier. See the table above. Velocity is a flow-to-stock ratio of spending to money; the multiplier is a stock-to-stock ratio of broad money to base money. Both involve money — they are not interchangeable.
• Assuming V is exogenous. Velocity is determined by money demand, which depends on interest rates, technology, confidence, and demographics. A model that treats V as a fixed parameter will systematically misfire in any regime where one of those underlying variables shifts.
• Reading short-term V changes as policy. Central banks do not "lower V" or "raise V" directly. They move policy instruments (mostly interest rates), and V responds. Headlines that frame the Fed as "increasing velocity" by raising rates are slipping cause and effect.