Which Money Supply Measure Best Predicts Inflation

Which money supply measure best predicts inflation compares M1, M2, M3, and alternative Divisia aggregates as leading indicators—with the answer shifting across historical episodes and depending heavily on which phase of the economic cycle and financial innovation cycle the economy is in. No single aggregate reliably forecasts inflation across all regimes, but understanding which measures lead and lag in each era is essential for both central bankers and investors trying to get ahead of price pressure.

The quantity theory and inflation expectations

The classical relationship between money supply and inflation rests on the quantity theory of money: in steady state, the growth rate of money should roughly equal the growth rate of prices plus the growth rate of output. If money supply grows 8% and real GDP grows 2%, prices should rise roughly 6% (accounting for inflation expectations and velocity).

This theory suggests that the level of money supply relative to economic activity (not the current inflation rate itself) predicts future inflation. A sudden surge in money relative to goods and services available for purchase will eventually push up prices.

The challenge is defining “money” itself. The narrowest measure, M1, includes only currency and demand deposits—the most liquid assets that function as immediate payment. But as interest rates rise and financial products evolve, holders shift between M1 and M2 (savings accounts) and M2 and M3 (CDs and wholesale funding). A rise in M1 that reflects people moving money out of savings accounts (M2) is not the same as a rise in M1 that reflects new money creation by the central bank.

M1 vs. M2: Which leads inflation?

Empirical tests over the past 50 years show that M2 has been a more reliable leading indicator of inflation than M1, though the relationship weakens during and after major financial innovations.

In the 1970s and 1980s, M2 growth accelerated 12–18 months before inflation peaked. The Federal Reserve, under Paul Volcker’s disinflation campaign, watched M2 closely because its growth rate in 1978–1980 had presaged the double-digit inflation of 1980–1981. By slowing M2 growth, Volcker signaled a credible commitment to reducing inflation expectations, and the mechanism worked.

The relationship held reasonably well through the 1990s and into the 2000s. A surge in M2 growth often preceded an acceleration in consumer-price inflation by 12–24 months. This made M2 valuable for policy makers and investors forecasting future inflation risks.

However, M1 has proven more volatile and less predictable. It responds sharply to changes in interest rates (as deposits shift in and out) and to technological shifts (the rise of money-market funds and, later, digital payments). A spike in M1 might reflect money moving from low-interest savings accounts into M1, not new money creation. Similarly, a drop in M1 during the 2000s reflected the rise of interest-bearing checking accounts, which technically belong to M2, not a real contraction in money.

The M3 episode: Why the Fed gave up

The Federal Reserve published M3 (the broadest measure, including large CDs and wholesale funding) regularly through the 1980s and 1990s. M3 was thought to be an even more complete picture of money available for spending, since it included the funding sources that banks use to make loans.

By the early 2000s, however, the relationship between M3 and inflation had deteriorated sharply. M3 grew rapidly (10–15% annually in 2004–2006), but inflation remained tame (2–3%), contradicting the quantity theory’s prediction. In 2006, the Federal Reserve discontinued publishing M3 data, arguing that it had become less useful as a policy gauge.

The main reason: M3 had become dominated by wholesale funding and shadow banking flows that did not translate into immediate spending power. Banks were using short-term repurchase agreements (repos) and large CDs to fund mortgage lending, but the money was not “hot” in the way M1 or M2 was—it was locked in longer-duration assets. Additionally, the rise of money-market funds and other near-banks meant that credit and liquidity were flowing outside the traditional banking system, making M3 incomplete even as a measure of financial system liquidity.

The discontinuation of M3 signaled that no single monetary aggregate reliably predicts inflation across all regimes. The measurement problem—deciding what counts as “money”—is not merely technical; it reflects structural changes in how credit and payment systems actually function.

Divisia indices: A weighted alternative

Some economists and central banks have turned to Divisia indices, which weight each component of the money supply by its “economic liquidity”—how readily it can be used for spending.

In a simple-sum M2 (the standard measure), a $100,000 large CD is counted identically to a $100,000 in a checking account, even though the CD is illiquid for months. A Divisia index applies a lower weight to the large CD, reflecting its reduced moneyness.

The Federal Reserve and the Center for Financial Stability publish Divisia aggregates. They show a somewhat tighter historical relationship to inflation than standard M1 or M2 in some episodes, but the improvement is modest and has also weakened in recent decades.

The advantage of Divisia is conceptual clarity: it explicitly acknowledges that “money” is a matter of degree, not a binary. The disadvantage is that it requires more data and computation, and the weights themselves are arbitrary (how much less liquid is a CD than a checking account?). Moreover, the relationship to inflation remains loose enough that Divisia indices are rarely used for real-time policy decisions.

The post-2008 complication: Quantitative easing and excess reserves

The 2008 financial crisis created a new complication. The Federal Reserve engaged in quantitative easing, expanding the monetary base (the amount of cash and reserves in the system) from $800 billion to over $2 trillion by 2014.

Standard quantity theory would predict violent inflation. Instead, inflation remained subdued (1–2%). The reason: the Fed’s newly created reserves sat in the banking system as excess reserves—money that banks could not easily lend out because credit demand was weak and regulations (post-Dodd-Frank) made lending riskier. The velocity of money (the speed at which each dollar circulates through the economy) collapsed.

This episode revealed a limitation of using money supply measures to forecast inflation without also modeling credit demand, fiscal stimulus, and inflation expectations. A doubling of M2 does not always lead to inflation if the new money is hoarded in bank reserves or if monetary velocity falls sharply.

The post-2008 era made it clearer that leading indicators of inflation must include fiscal policy, commodity prices, and wage growth alongside monetary aggregates. The money supply is a necessary condition for inflation but no longer a sufficient one.

The 2021–2022 test: Did M2 predict the inflation spike?

The pandemic and 2021–2022 episode offers a crucial test. The Federal Reserve expanded M2 by roughly 40% from 2019 to 2021 (an unprecedented rate). Many observers warned that this would generate inflation, citing the historical relationship between M2 and prices.

Inflation did spike in 2021–2022, reaching 9% in mid-2022, the highest in 40 years. On the surface, the M2 thesis appeared vindicated.

However, the lag was unusual. Normally, M2 growth is followed by inflation 12–24 months later. In 2021–2022, the surge in inflation came within 12–18 months, and some commentators argued that the timing matched fiscal stimulus spending (the American Rescue Plan in March 2021) more closely than M2 growth rates. Additionally, global supply-chain disruptions and energy prices (exacerbated by Russia’s invasion of Ukraine) played outsized roles.

The episode suggests that in high-inflation regimes—when expectations are unanchored and supply shocks are large—the monetary-aggregate approach to forecasting inflation is less reliable than in low-inflation regimes where expectations are stable and shocks are primarily demand-driven.

Practical use: Combining measures and context

For investors and policy analysts, the practical lesson is that no single money supply measure reliably predicts inflation in all eras. Instead:

• M2 growth remains the most commonly watched aggregate for medium-term (12–24 month) inflation forecasting, especially in regimes where expectations are anchored and supply is not severely disrupted.
• M1 is more volatile and less predictive of inflation, but spikes in M1 relative to M2 can signal shift in risk appetite and may precede asset-price inflation (stock or real-estate bubbles) rather than consumer-price inflation.
• Divisia indices offer a more nuanced picture but require more data and are rarely used by central banks for real-time decisions.
• Context matters more than the measure itself: in low-inflation regimes, M2 is a decent leading indicator; in high-inflation or crisis regimes with fiscal shocks, it is less useful.

Central banks now track all three alongside velocity of money, credit spreads, commodity prices, and expectations surveys. The era of relying on a single aggregate is over.

See also