View all Articles
Commentary By Peter Ireland

The Right Measure of the Money Supply

Economics Finance

How useful have monetary aggregates been in signaling the stance of monetary policy in real time? These four panel graphs show business-cycle components of various measures of money supply. In particular, the Federal Reserve’s official M1 and M2 measures are shown, together with their Divisia counterparts constructed by William Barnett and his associates at the Center for Financial Stability (CFS) in New York City.

Rather than simply adding up the values of funds held in various types of bank accounts, as the Federal Reserve’s measures do, the Divisia aggregates draw on methods that explicitly recognize, for instance, that a dollar held in a NOW account is more liquid—and can therefore be said to provide a larger flow of “monetary services”—than the same dollar held in a three-month CD. Quite helpfully, too, the CFS aggregates also adjust the Federal Reserve’s official measures to remove the distortionary statistical effects of computerized problems that most banks now use to invisibly “sweep” funds on deposit in customers’ checking accounts into savings accounts for the purpose of minimizing statutory reserves requirements.

Despite some differences in the behavior of the four series, a number of general conclusions can be drawn from their movements:

1) The monetary aggregates do seem to anticipate subsequent movements in output and prices in a manner that is fully consistent with traditional accounts of monetary policy and its effects. Monetary policy, as reflected in the behavior of the Divisia M1 aggregate, moved unrelentingly from accommodation to tightness in the years from 2005 through 2008. The economy weakened and then fell into a deep, deflationary recession. Monetary constraint continued through mid-2011, foreshadowing the tepid recovery and slow inflation that we are seeing now.

2) None of the graphs in this picture support the popular view that monetary policy has been consistently and excessively accommodative since 2008. To the contrary, focusing on money growth as an indicator, policy seems to have been more often a source of deflationary impulses.

3) The enormous expansion in bank reserves since 2008 has, for the most part, not translated into rapid growth in the broader monetary aggregates. Indeed, the shaded periods in each graph, marking the three waves of large-scale asset purchases known popularly as “quantitative easing (QE),” seem at best only loosely connected with sustained movements in money growth. Divisia M1 rose, but then fell just as sharply, during QE1, increased substantially, but only in the final months of QE2, and has remained essentially flat during QE3. One reason for this may be that the Federal Reserve has been paying interest on reserves at a rate that, though small, nevertheless exceeds the rate that banks can earn on comparable risk-free assets like Treasury bills. Policies that pay interest on reserves are best viewed as acting on the demand for reserves, so QE1 through QE3 may have been necessary simply to accommodate this increase in the demand for reserves with a corresponding increase in supply.

Another reason might be that, by focusing so heavily on the behavior of long-term interest rates, the Federal Reserve has neglected to consider, and exploit, the traditional channels linking well-designed open market operations to changes in money growth, which can then give rise to changes in output and inflation.

4) Money growth has recovered since the middle of 2011, suggesting that monetary policy is finally lending support to the economic recovery. So long as this growth in the monetary aggregates continues, we should expect to start seeing a movement in inflation back towards the Fed’s two percent target, with further gains in employment, income, and spending as well.

5) From here any marked acceleration or deceleration of growth in the monetary aggregates ought to be taken seriously, as a warning sign that the Federal Reserve’s exit has gone off track.

One final question concerns measurement: Which of the four monetary aggregates shown is likely to be most useful as an indicator of monetary policy, going forward?

William Barnett showed more than thirty years ago that the standard, simple-sum approach to monetary aggregate makes no sense. Just as no one would take seriously a measure of real economic activity that assigned equal weight to each good produced—adding apples and oranges or, worse yet, adding apples, oranges, luxury automobiles, and aircraft carriers—so, too, no one should rely on a monetary aggregate that assigns equal weight to a dollar held as currency and the same dollar held in a three-month time deposit.

Despite their fancy and foreign-sounding name, the Divisia monetary aggregates—call them “appropriately-weighted measures of money,” if you prefer—proposed by Barnett simply apply the same, value-weighted scheme to monetary aggregation that national income accountants use to measure real GDP.

In the figure, one can see that in recent years, the Federal Reserve’s official, simple sum aggregates and Barnett’s preferred, Divisia counterparts, have behaved similarly. Yet that has not always been the case. In the early 1980s, for example, Nobel Prize winning economist Milton Friedman forecast a return to high inflation based on observations of very rapid growth in simple-sum M1, a rare misstep that, Barnett notes, he might have avoided if he had been looking at the Divisia measure instead.

Likewise, my colleague Michael Belongia and I have shown that when Divisia measures of the money supply are used in place of the Federal Reserve’s official simple-sum aggregates, strong statistical links between movements in Divisia measures of money and subsequent changes in output and prices can be found in data spanning the Great Inflation of the 1970s, the Great Moderation of the 1980s and 1990s, and the Great Recession of 2008.

It is clear that economists owe a debt of gratitude to Barnett, and to his colleagues Richard Anderson and Barry Jones, for their careful work in adjusting the official measures of the money supply. Less clear is why the Federal Reserve cannot or will not simply make those adjustments to the official statistics themselves.

Peter Ireland is a Professor of Economics at Boston College