For no particular reason other than posting something on it not too long ago, and then again more recently, I've been revisiting the issue of using monetary aggregates as targets for monetary policy. First recall, in very general terms, one reason why there is an issue. The Fed has one policy tool. With a single tool it is not possible to control two variables, a monetary aggregate and an interest rate simultaneously. So the Fed must choose one or the other (or some combination strategy where both are kept within some tolerable range, but I'll set that aside for this discussion to keep it simple). In recent years, the Fed has chosen to target the overnight borrowing rate between banks, the federal funds rate, but there was a time when the Fed relied much more on monetary aggregates, first M1, then M2. In this speech, Alan Greenspan explains why the Fed has deemphasized monetary aggregates. The first event that undermined aggregates was the introduction of NOW accounts. The appearance of NOW accounts made M1 assets much more interest sensitive and hence much more volatile, and that volatility made the relationship between M1 and other variables of interest such as output and inflation more difficult to discern:
Rules vs. discretionary monetary policy, by Chairman Alan Greenspan, Stanford University, 1997: ...Although the ultimate goals of policy have remained the same over these past fifteen years, the techniques used in formulating and implementing policy have changed considerably as a consequence of vast changes in technology and regulation. Focusing on M1 ... was extraordinarily useful in the early Volcker years. But after nationwide NOW accounts were introduced, the demand for M1 in the judgment of the Federal Open Market Committee became too interest sensitive for that aggregate to be useful in implementing policy. Because the velocity of such an aggregate varies substantially in response to small changes in interest rates, target ranges for M1 growth in its judgment no longer were reliable guides for outcomes in nominal spending and inflation. ... As a consequence, by late 1982, M1 was de-emphasized... However, in recognition of the longer-run relationship of prices and M2, especially its stable long-term velocity, this broader aggregate was accorded more weight, along with a variety of other indicators, in setting our policy stance.
By turning to M2, the Fed stabilized the target monetary aggregate since much of the asset movement was between M1 and M2. But as he notes, this did not last long as financial innovation brought about highly liquid assets outside of the definition of M2 which began to attract financial investment:
As an indicator, M2 served us well for a number of years. But by the early 1990s, its usefulness was undercut by the increased attractiveness and availability of alternative outlets for saving, such as bond and stock mutual funds, and by mounting financial difficulties for depositories and depositors... The apparent result was a significant rise in the velocity of M2, which was especially unusual given continuing declines in short-term market interest rates. By 1993, this extraordinary velocity behavior had become so pronounced that the Federal Reserve was forced to begin disregarding the signals M2 was sending... Data since mid-1994 do seem to show the reemergence of a relationship of M2 with nominal income and short-term interest rates similar to that experienced during the three decades of the 1960s through the 1980s. ...however, the period of predictable velocity is too brief to justify restoring M2 to its role of earlier years... The absence of a monetary aggregate anchor ... has not left policy completely adrift. From a longer-term perspective we have been guided by a firm commitment to ... the ultimate goal of achieving price stability. ...
The period he talks about, 1990-1994, is evident in this graph from a recent post repeated for convenience:
What explains the behavior in the early 1990s? Here's one explanation from the Cleveland Fed (this is long already, so maybe I can get to the recent MZM results, e.g. here, some other time):
Results of a Study of the Stability of Cointegrating Relations Comprised of Broad Monetary Aggregates, by John B. Carlson, Dennis L. Hoffmanb, Benjamin D. Keenc, Robert H. Rasche, Federal Reserve Bank of Cleveland (1999): Abstract We find strong evidence of a stable “money demand” relationship for ... M2M through the 1990s. Though the M2 relation breaks down somewhere around 1990, evidence has been accumulating that the disturbance is well characterized as a permanent upward shift in M2 velocity, which began around 1990 and was largely over by 1994. Taken together, our results support the hypothesis that households permanently reallocated a portion of their wealth from time deposits to mutual funds...
But this is certainly not the only explanation. E.g., from the Dallas Fed:
What was Behind the M2 Breakdown?, Dallas Fed, 1999: A deterioration in the link between the M2 monetary aggregate and GDP, along with large errors in predicting M2 growth, led the Board of Governors to downgrade the M2 aggregate as a reliable indicator of monetary policy in 1993. In this paper, we argue that the financial condition of depository institutions was a major factor behind the unusual pattern of M2 growth in the early 1990s...
The unusual M2 growth pattern can be seen here in the graph from the post linked above. By targeting an interest rate rather than a monetary aggregate, the Fed has avoided the problems associated with targeting unstable aggregates, and interest rate targeting is also supported by the theoretical literature. Because of this, the focus on monetary aggregates has waned in recent years.