Money Statistics as Inflation Predictors

M-1, M-2, M-3, MZM: What does it all mean? If the Federal Reserve System drops M-3 after decades, what does that say for the other three?

Why did the FED gather the statistics for M-3 in the first place? What was it that the central planners of the monetary system thought they gained from having reams of data defined as M-3? What theory did the central planners have? How did they decide which data to monitor? How did the statisticians decide which samples were the most reliable? Finally, why did they give up on the entire project in 2006?

Obviously, I am not actually interested enough in getting answers to these questions as I am in judging the usefulness of the whole enterprise. It was a waste of resources. The FED’s managers finally concluded this in 2006.

In this report, I will cover some of the basics of Federal Reserve reporting. I will offer some guidelines on which monetary statistics are relevant for forecasting, and which are not.

In doing this, let me say from the outset that the world would be far better off if the FED confined itself to gathering statistics and got out of the money-management business.

In 1961, Murray Rothbard wrote an essay for The Freeman, “Statistics: Achilles’ Heel of Government.” He argued that the State needs to create the illusion of being able to plan an economy rationally. To do this, it needs statistics. Without statistics, it would visibly be flying blind. So, the State gathers statistics of all kinds.

The central planners believe that with statistics, they can make rational, effective, beneficial decisions with other people’s money. Now, consider the Federal Reserve System. It actually creates other people’s money. It buys T-bills and other government debt certificates. It buys them from a group of about 20 companies, who make money on every transaction. The companies then send money to the U.S. Treasury. The government then writes checks to pay for whatever the government has decided to spend.

Rothbard, following Mises, argued that statistics are records of the past. To make accurate forecasts suitable for rational economic planning, the government’s salaried specialists must be able to forecast the future in terms of a specific theory of economic causation. Then they must select the appropriate statistics that will provide insight regarding the future. “If we do this, then that will be the result.” Why? “Because the immediate economic past was this.” How do you know? “Because we have these statistics.” How do you know they still apply. They’re old. “We can safely assume it.” How do you know that?

“Because it is safer to assume it than to assume it’s wrong.” Safer for whom? “Us, dummy. That’s what we’re paid to do.” They are paid very, very well.

The monetarists, more than any other economic school of thought, have bet their careers — and ours — on the assumption that monetary policy is the crucial factor in preventing economic recessions. Keynesians bet their careers — and ours — on the assumption that fiscal policy is more important than monetary policy.

Both sides agree: the government must tax and spend to overcome recessions. They disagree on the most effective way for government to get its hands on the money to spend. It’s “tax, borrow, and print predictably” vs. “tax, borrow, and print whatever taxing and borrowing doesn’t pull in.”

Then there are the supply-siders. Their position is “tax less in the higher marginal income tax brackets, spend more, borrow more, and print whatever is needed to keep interest rates low.” A few of them say they are gold standard people, but their gold standard is never a gold coin standard with complete redeemability at a fixed price at every bank. Their gold standard is some sort of mathematical formula that only they understand. It’s not “Here’s my paper money. Give me my gold coins.”

The Austrians disagree with all three schools. Their position is “Tax less in all tax brackets, spend less, borrow nothing, print nothing. If the budget goes into deficit, spend even less.” Nobody in government or academia listens.

The Role of Central Banking

The monetarists say that the most important justification for the Federal Reserve System to control money is to keep monetary policy away from Congress. They don’t trust democracy. Basically, this is the #1 justification of central banking. It is maintained by all schools of economics except the Austrians, who are ready to trust Congress and the commercial banks before they trust central bankers. Given the Austrian’s low opinion of the Federal government, this reveals a great deal about their opinion of central banking.

The monetarists don’t want to give discretion to the FED. They want the FED to increase the money supply by [?]% per annum. No, they don’t agree on the percentage. But this doesn’t really matter, because central bankers never bind themselves to a fixed rule of anything other than the usual rule of getting ahead at someone else’s expense.

So, the Keynesians worry about interest rates, since that affects the government’s ability to borrow, and the monetarists worry about prices, since that affects the performance of the economy. The monetarists worry about anything that slows down economic growth.

The Austrians worry about the ability of the free market to survive in the face of planning by Keynesians, monetarists, and supply-siders.

What we have heard from Federal Reserve Board chairman since 1945 is that inflation is the problem. Inflation remains a problem, year after year.

The Federal Reserve is in control of monetary policy. What does this tell us about the cause of inflation? “Shhhh. Nobody is supposed to ask that question.”

For a Federal Reserve official to say that inflation remains a problem is like Willie Sutton warning us against the unacceptably high rate of bank robberies.

If you want to understand the role of central banking, read all 800 pages of Jesús Huerta de Soto’s book, Money, Bank Credit, and Economic Cycles (Mises Institute, 2006). I have never read anything remotely as clear on banking theory and policy. It is flawless, except for pages 200—260, which were written for his academic peers, and who won’t read it anyway. You can buy it for $45. It will cost you a great deal more than $45 in forfeited time. But if you want to understand what is wrong with central banking — legally, economically, and morally — this is the book to read, more than any book ever written.

The bureaucrats who work for the Federal Reserve System have a theory of economic causation that rests on this principle: “The free market cannot supply an appropriate quantity of money. It cannot produce conditions that are favorable to boom-free, bust-free economic growth. Only a well-run central bank can do this.” Everything else follows.

Predicting Prices

Which prices? Retail prices? Wholesale prices? Discount prices? Going out of business prices? Three a.m. infomercial prices? Used goods prices? Stolen goods prices? Black market prices? Raw materials prices?

The statistician’s answer is a price index. But whose price index? According to what theory of economic causation? Reported by whom? Reported to whom? How recently? Under what degree of legal compulsion? Assessed by which sampling technique? And, most important of all, weighed sectorally for its economic relevance by what formula?

Only the most sophisticated graduate student in statistics and economics can understand these issues, let alone devise correct procedures that will actually be followed. In short, the science of economic statistics functions in much the same way that a priesthood functions. The statistician is not the high priest, but he is surely the equivalent of a Levite. He recites jargon that sounds suspiciously like prayers recited in the outer court of the temple by the Levites just before the high priest slips behind the veil of the holy of holies and stands before the ark of the covenant.

The high priest is the Chairman of the Board of Governors. He is less like a high priest and more like the Wizard of Oz, surrounded by flashing lights and smoke. Congress is like the scarecrow and the lion combined: neither brains nor courage. (They all say they have a heart, so I’ll leave out the tin man. But whenever they seek to prove their heart-felt love, they invariably spend my money.)

Milton Friedman became famous among other economists for insisting that the only meaningful test of an economic theory is its ability to enable its believers to make accurate predictions. He almost convinced me of this in the 1970’s, because I remembered his prediction in the late 1960’s that gold’s price might fall if the government ever stopped buying it for $35. I saw this as a good reason to dismiss monetarism as an anti-gold, pro-inflation defense of fiat money. But I knew that long before Nixon closed the gold window in 1971. I didn’t need gold at $700 to prove to me that monetarism is statist to the core. I understood this a priori!

Nevertheless, when it comes to making predictions regarding future price indexes, we can make valid assessments of the effectiveness of the various M’s. It is not a question of which M provides a good forecasting tool. It is a question of which one performs least poorly.

Note: I refer to price indexes. As to how much good they are for predicting prices that I will be facing — that will matter to me in my personal situation — is a matter of considerable guesswork on my part. What about you? How many prices are really significant for you? That depends on your age, health, tastes, spouse’s tastes, and a lot of other factors you don’t recall right now but will recall the next time some item costs 30% more than today.

I use the median CPI, published by the Cleveland Federal Reserve Bank. I like it because I think its assumptions are more reasonable than the CPI’s assumptions. It seems to fluctuate less. I can track it quarter by quarter back to 1980. I can see its general direction.

But the CPI gets more coverage. It’s not significantly different from the Median CPI. Also, because the Federal Reserve Bank of St. Louis publishes more statistics than the other regional FEDs, and because it makes the CPI available, I will use it in this report.

The Consumer Price Index

The statistics for the CPI go back to 1913. There are real conceptual problems with any price index this old. The enormous productivity of modern capitalism has made obsolete just about everything that was counted in the CPI in 1913, other than basic foodstuffs and the most basic industrial commodities: steel, wood, paper, cement, glass. Is there anything made in 1914 that you would want today instead of what you’ve got? I would like a three-piece, custom-made wool suit. Make that six. You probably wouldn’t. A one-story home near the library at Berkeley, where you don’t need home air conditioning, which hadn’t been invented yet. What about services? A haircut. A shoe shine. A gourmet dinner in New Orleans, with a Dixieland band afterwards. That’s about it.

So, I am interested in the CPI since about 1965, when the Vietnam War started blowing apart Federal budgets. I can remember 1965. We had color TV shows, and better ones to watch. We had movie theaters with better films to watch — and larger screens. We had books. We had air conditioning. The main difference for me is word processing, and that has not changed radically for me since 1981. The difference between 1980 and 1981 was gigantic . . . even life-changing. It hasn’t changed at all since 1991 on the composing side, just the typesetting side. It hasn’t changed at all since 1995.

Let’s take a look at the CPI. This chart goes back to 1947, but not much changed between 1947 and 1967. As you can see, the real change begins in 1971, when Nixon took the country off the Bretton Woods pseudo-gold standard.

I got my first full-time job one month after he did that. It’s odd. I can remember the day he did it. My friend Bob Warford called. It was a Sunday. He had just heard it. I was wrapping up my graduate school days in California. A month later, I arrived in Irvington, New York, to work for Leonard E. Read at the Foundation for Education. I had come very close to working with the Nixon White House. My contact there later went to jail. It was a good thing I got an offer from Read.

I can recall how upset Read was because of the price controls. He told our lunch group that he had called some old buddy at the Chamber of Commerce to vent his spleen. “But Leonard,” the man had said, “the Board just came out in favor of the controls.” Read told that story with a sense of incredulity. Were the leaders of American business opinion this blind? They were indeed; the National Association of Manufacturers had also publicly supported the controls. It was as if Read had spent a quarter century shouting into the wind. But I’m sure he was happy he had not accepted the job offer as head of the International Chamber in 1946. He started FEE instead.

Because we were followers of Mises, we all knew what was coming next. The controls would cause shortages. They would distort production. They would not hold down the consumer price index very much, but they would surely hamper the flow of accurate information. They would distort relative prices, which does the most harm to an economy. This is exactly what happened over the next two years.

So, the most that we can safely say about the following chart is that it displays the effects of government when the President can call the Chairman of the Federal Reserve Board and beg him to Do Something about recession. Nixon had a bad one in 1971. The government ran back-to-back deficits of $23 billion. Horrible! We get some sense of the world we have lost when we recall that a $23 billion Federal deficit was considered politically unacceptable in 1971.

To overcome the recession, Nixon wanted a growing economy. That meant lower interest rates. He told Arthur Burns to provide them. Burns complied by buying U.S. T-bills with newly created money. The result was what we see below.

If we look at 1971, it looks as though the upward sloping line is somewhere around 40. If you look at it today, it’s around 200. If we divide 200 by 40, we get 5. Prices are five times higher. We can verify this with the inflation calculator. It is posted on the site of the Bureau of Labor Statistics. Entering 1000 in the 1971 box, we get 5024 for 2007. This is close to 5 to 1.

The exercise I propose is to take the standard money measures — M’s — to see which one would have provided the best indicator of what happened, 1971 to 2007. Put another way, if you were to use one of them to predict the general movement of prices, which one would you trust?

Adjusted Monetary Base

This is the indicator I use most often. The reason for this is simple: it is the one aggregate that the Federal Open Market Committee (FOMC) can control directly. It records the FED’s holdings of assets that serve as legal reserves for the money supply. The monetarists call this high-powered money.

Because the FOMC controls this aggregate, it provides the best indicator of FOMC policy. We can see what the bureaucrats who make the decision about the direction of consumer prices thought they had to do to achieve this result.

The point is this: we can compare FOMC policy with the statistical results in the most commonly used and quoted price index. To the extent that we are affected by the overall movement of prices, the adjusted monetary base is what lets us assess the effectiveness of the FED with respect to price inflation. That’s what FED Chairmen keep telling us is most important. Let’s take them at their word.


This series has been collected only since 1984, so it doesn’t give me what I really want. But I can still use this. With the index a little under 200 in 1984 and 845 today, we have a way to derive a figure to use. Divide 845 by 190. We get 4.4.

Using the BLS inflation calculator, I find that the CPI increased by 1.95.

So, the price level increased by less than 2 times. The AMB overestimated price increases by a factor of 2.2 (4.4 divided by 2). Prices increased much less than this figure would have led us to believe.


This is the good old boy of the monetary statistics. It is currency in circulation plus checking accounts, meaning demand deposits. Currency is instant money. A checking account is close to it.

M-1 1984

If 1984 is about 525, and 2007 is about 1390, the increase is 1390 divided by 525, or 2.6. In other words, M-1 in 2007 is about 2.6 times larger than in 1984. With the CPI increasing by just under 2, this indicator is very close: 2.6 is only 30% above 2. (2.6 – 2.0 = .6 divided by 2.0 = .3 = 30%).


The M-2 figure used to be the preferred figure for the monetarists. I never understood why. I always used M-1 because it is more liquid. M-2 is M-1 plus time deposits, or savings accounts.

M2 84

If 1984 is about 2100, and 2007 is about 7200, then the increase is 3.4. This compares with a 2-fold increase in the CPI. If we subtract 2 from 3.4, we get 1.4. Thus, the M-2 figure is 70% higher than the increase in the CPI (3.4 – 2.0 = 1.4 divided by 2 = .7 = 70%).


M-3 is the series that ended in 2006. The FED says it stopped collecting the figures. Let’s see why.


If 1984 is about 2,400, and 2006 was about 10,300, then the increase was 4.3. We are told by site after site that M-3 is up by 10% since the figure stopped being reported. So, add 10% to 10,300. That’s 11,330. So, the increase is by a factor of 4.7 (470%). Yet the actual increase in the price level was a little under 2. The overestimation of prices was considerable. Prices increased by less than half of what M-3 would have led us to believe.


This figure is not cited very often. It supposedly is highly liquid.

The 1985 reference point looks to be about 1,500. Today, it’s at 7,400. The increased by a factor of 4.9. This is by far the least accurate statistic. It overestimates the rate of price inflation by almost two and a half times.

The Best Indicator

It turns out that the simplest definition of money, M-1, would have provided the best indicator of price level increases of all the competing definitions. M-1 is the closest definition of instant money that the St. Louis FED publishes. This definition was the most effective monetary forecasting tool available to the public.

What this tells us is that the FED implemented its monetary policy by way of the adjusted monetary base. Yet if we had used this indicator the whole time, we would have significantly misforecasted the performance of the consumer price index.

The CPI increased from 1971 to 2007 by a factor of five. Using the four M’s — not the adjusted monetary base — from 1971 to 2007, the figures are as follows:

M-1 = 5.75x

M-2 = 12x

M-3 = 17x

MZM = 15x

There is no question which monetary aggregate would have supplied the most accurate assessment: M-1. It was quite close. The others are so inaccurate as to be useless. But beyond any doubt, M-3 is the least accurate. It vastly overstated the rate of price inflation to be expected.

M-3: Seller’s Remorse

Ever since the FED discontinued publishing M-3, the hard-money e-letter world has gone ballistic. It’s M-3 this and M-3 that. It’s as if the FED had this secret tool for forecasting and is keeping it all to itself. The FED is plotting mass price inflation. Gold will go to the moon! Keep you eye on M-3. Of course, it’s gone. But there are ways of tracking M-3. Nobody can say exactly how, but there are sites that report M-3, and e-letter editors refer to them.

Yet M-3 was hardly reported by anyone, including the FED, prior to 2006. It reminds me of seller’s remorse. He finds this old hi-fi in his attic that he has not turned on for 30 years. He sells it at a garage sale for $10. Then someone tells him it has a gazorninplex dual fazooler. The sound is just perfect. Why, it’s worth a fortune to insiders who really know the world of classic stereo.

The seller is depressed. He wants it back. He finds out that someone sells a piece of software that converts an ordinary scratch-free CD-ROM into a perfect imitation of a stereo system with a gazorninplex dual fazooler.

The guy installs it. He’s not sure if it really works as claimed. He thinks it might. The only trouble is, the software reinserts the original clicks and pops of his vinyl records.

My conclusion: let M-3 die the death of a crazy aunt who lived in the basement. She meant well, but she was an embarrassment.

M-1 After 1994

If we look at M-1 from 2001 to the present, it reveals a flattening since 2003. It has gone nowhere.

Prices are up about 15% since 2001. M-1 reveals an increase of 25% since 2001. All of the other monetary aggregates reveal an increase of over 40% since 2001. But M-3 indicates an increase of 60% since 2001. Again, M-3 was utterly useless. Of all the indicators, it deserved most to be discarded.

All of the monetary aggregates overshot the target, if the goal of a monetary aggregate is to forecast the consumer price index. But M-1 has been by far the most accurate from 1971 to the present.

If this is correct, then what is M-1 telling us about the rate of price inflation? It is not headed higher. Look at M-1 since the beginning of 2006. It is falling.

In contrast, the adjusted monetary base is up less than 2%. It is now running closer than any other aggregate to the 3% per annum increase that has been reported.

What can I say other than this? The rate of price inflation is unlikely to rise rapidly in the United States over the next three years. The 10% figure that we keep hearing about regarding the “true” increase of the “true” money — M-3 — should not be taken seriously by anybody. The statistic was useless from day one as a means of forecasting price inflation.

August7, 2007

Gary North [send him mail] is the author of Mises on Money. Visit He is also the author of a free 19-volume series, An Economic Commentary on the Bible.

Copyright © 2007