The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.

Authors for macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

« All Quiet On The Fed Funds Futures Front | Main | The CPI Report By The Numbers »

November 15, 2005

Much Ado About M3

A minor buzz has developed over last week's announcement from the Federal Reserve Board that it will discontinue the publication of the M3 monetary aggregate, effective March 23 2006.  You can find, not exactly positive, commentary from Mark Thoma (here), from Barry Ritholtz, from the Prudent Investor (here and here).

Let me be very clear -- I speak only for myself here.  The data collection function is a role that is delegated to the Board of Governors, and only they are in a position to comment on the whys and hows of the decision to discontinue collecting the data that makes up the M3 aggregate.  But, for my part, I think a good starting point is the opinion expressed at Institutional Economics:

Growth rates in broad money and credit aggregates tend to be dominated by trends in financial intermediation and thus have only a very tenuous relationship with monetary policy and even a somewhat loose relationship with economic activity.

From the point of view of a policy maker, why do we care about monetary measures in the first place? We care for the same reason everyone else does -- because most of us still believe that, ultimately inflation, inflation is everywhere and always a monetary phenomenon.  Here is one famous picture of this relationship, from George McCandless and Warren Weber:


That picture is based on the monetary aggregate M2.   McCandless and Weber explain this, and the other measures of money, they use in their study:

For each country with 10 or more years of data (110 countries), we calculate the geometric rate of growth for consumer prices... three definitions of money—M0, currency plus bank reserves...; M1, money easily used in transactions...; and M2, money easily used in or converted into use for transactions...

In short, monetary theories tell us that monetary assets are those that are held, at least in large part, for the purpose of supporting transactions.  M0, M1, and M2 arguably fit that definition.  Here is the definition of M3, from the Federal Reserve's education site (hat tip, Mark Thoma):

M3    Measure of the U.S. money stock that consists of M2, time deposits of $100,000 or more at all depository institutions, term repurchase agreements in amounts of $100,000 or more, certain term Eurodollars and balances in money market mutual funds restricted to institutional investors

Some of the items in M3, but not in M2, are more likely than others to fit the definition of transactions balances.  Institutional money market mutual funds, for example.  That series is in fact part of another popular monetary measure, MZM.  But that series is not being discontinued, presumably because it is used in constructing a monetary measure that is useful for thinking about the transactions balances of the public.

And there is this to think about.  The McCandless and Weber study is about the correlation between money growth and inflation, which is the connection that all of monetary theory leads us to.  The Big Picture repeats a picture, originally appearing at Economist's View, that highlights the levels of M1. M2. M3.  Here is my version...


... and sure enough there is an acceleration of M3 relative to both M2 and M1 that starts in the early 1980s (and accelerates with the asset-price booms the begin in the mid-1990s).  These trend differences are, with as close to certainty as we get, a result of financial market innovation and change that have very little to do with monetary policy. And, in this case, a look at the accumulated effects of these trend differentials can be quite deceiving.  Here is a look at the more relevant (for inflation) growth rates of these monetary measures:


The growth rate of M3 is indeed generally higher than the growth rate of M2 (as must be the case given the levels picture), but not dramatically so.  And there is absolutely no evidence that the difference between the two has been growing of late -- just the opposite, in fact.

For my money, the much ado about the pending disappearance of M3 is much ado about not much.

November 15, 2005 in Federal Reserve and Monetary Policy , Inflation | Permalink


TrackBack URL for this entry:

Listed below are links to blogs that reference Much Ado About M3 :


There seems to be an interesting schism between the Wall Street users of Fed data, and the more academic users.

Wall Street like a good clean data source, and the Fed provides that.

I suspect the Fed has a broader audience than they realize.

Posted by: Barry Ritholtz | November 15, 2005 at 11:37 AM

From the McCandless and Weber graph, couldn't you easily put Inflation on the horizontal axis and Money Growth on the vertical? In other words, wouldn't it be a possible interpretation of the data that money supply grows faster as a result of higher inflation?

Correlation, causality, etc.

Posted by: ErikR | November 15, 2005 at 12:35 PM

Ok, maybe the fed governors think that M3 is not useful as an indicator of future inflation. Why bother to stop collecting it? Why now when the new chairman has to show his determination to fight inflation, especially when he's supposed to be mr. transparency? The penalty for continuing to collect it is a few manhours from the fed staff. The cost is a potential credibility headache for the new chairman and worry for wonkish types on wall street.

Why take this on? Don't they have other things to think about?

Posted by: Mark Sullivan | November 15, 2005 at 12:40 PM

Curiously the ECB considers M3 (broad money) as one of its key indicators. Growth in M3 above targets is one of the arguments being used to justify the need for 'imminent rate rises' in the eurozone. You can use your own judgement to decide whether we are ahead of or behind the curve here:).

Indeed new candidate eurozone members like Poland have just started keeping track of M3 (in 2002) as part of the reform and standardisation programme:

"The scope and the presentation format are adjusted to standards used by the European Central Bank. Starting from March 2002, broad money M3 became basic analytical and publication category of money supply in Poland."

If you're really an M3 and ECB headbanger, you can try this working paper:


Posted by: Edward Hugh | November 15, 2005 at 01:13 PM

So if one was interested in looking at Eurodollars and repos, where would one look come April of 2006?

Posted by: Tim | November 15, 2005 at 08:16 PM

It would have been better had the first graph been a log-linear chart.

Posted by: jm | November 16, 2005 at 12:21 AM

Barry and Mark -- As I noted, this was a Board decision, so I don't know all of the factors that went into the decision. It will be interesting to see if there really is substantial push back on the announcement. As for the transparency issue, I'm not really sure how that comes into play. How does the discontinuing a series that plays no role at all in Fed policymaking harm the effort to be transparent?

Erik -- The causality/correlation identification problem is always an issue. M&W's decision to look at average cross-country experience is motivated in part by the hope that so doing mitigates any potential problem. All I can say is that I think most economists accept that the role of money is indeed causal.

Edward -- I've been meaning to welcom you back from the caves. As to the ECB decision to focus on M3, I've never quite understood it. The issue is not finding an aggregate that is stable, but one that actually measures transactions balances. To me, M3 is not the measure. Of course what we care about is the connection to inflation, nominal income, and the like, but the relationship with M3 has never (to my knowledge) been deemed useful in the US and the ECB experience is pretty short.

Tim -- don't know the answer to that one.

jm -- Money growth and inflation are already expressed in percentage terms, so I'm not sure how expressing the variables in logs would be that useful.

Posted by: Dave Altig | November 16, 2005 at 08:13 AM

"How does the discontinuing a series that plays no role at all in Fed policymaking harm the effort to be transparent?"

With all due respect, this sounds like, "We'll tell you what you need to know".

The "much ado". as far as I can tell, is about the unexplained disappearance of repo and Eurodollar statistics.

Posted by: Tim | November 16, 2005 at 10:50 AM

Tim --

Here's the point i was trying to make. To me, the transparency issue has to do with sharing information directly related to monetary policy decisions. Discontinuing collecting the M3 series would be directly counter to that if, in so doing, the Fed was withholding data from the public that were nontheless being used in policy deliberations or thinking. That is decidedly not the case here.

Here is the broader point -- the Federal Reserve is not itself a data collection agency. A lot of data is collected, of course, in the process of fulfilling the central bank's mandates under the Federal Reserve Act. That data is then, subject to confidentiality restrictions, shared with the public. Although I emphatically repeat that I am not speaking for the Board, the Board staff, or anyone else in the Federal Reserve System, my view is that, because the M3 statistic is not being used in the process of fulfilling the central bank's mandate, the case for continued efforts in collecting it is somewhat dicey.

Posted by: Dave Altig | November 16, 2005 at 03:54 PM


Thanks for taking the time to respond - I now understand the Fed perspective much better as a result.

Posted by: Tim | November 16, 2005 at 08:35 PM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

If you have a TypeKey or TypePad account, please Sign in

Google Search

Recent Posts



Powered by TypePad