Free To Choose Network | Free To Choose
Free To Choose
Money and Economic Development
The Horowitz Lectures of 1972
Milton Friedman

FIRST LECTURE: A Survey for the Evidence of Monetarism

A Sample of the Evidence

I can’t possibly present to you all the evidence that I have referred to, so I want to turn to a slide show now and give you a little flavour of the evidence. I understand that there is an old Yiddish proverb that one picture is worth a thousand equations. And that is what these pictures are for.

The first four slides come from a study of monetary trends that Anna Schwartz and I have been making. These four charts are for a period of almost a century, from about 1880 to 1970, for two countries, the United States and the United Kingdom. They are intended to illuminate longer period movements and so we have used as our basic unit of observation a business cycle phase, that is, either an expansion or a contraction. The individual observation plotted in Chart 1 is the average value during a half cycle, that is during the period from a business cycle trough to a business cycle peak, or from a peak to a trough. The purpose is to eliminate the cyclical movement and concentrate on the longer term movement.

In each part of the chart (the top is for the U.S., the bottom for the U.K.), there is plotted the level of nominal income (it is called “nominal” because it is in dollars or in pounds sterling, and not corrected for price changes) and the quantity of money, also in dollars or in pounds sterling.1 These are levels and because of the strong upward trend it is hard to see what is happening over shorter periods. However, it is clear that there is a very close relationship between the two, that the two move much in the same direction

In order to eliminate the very long period trend, the next chart (Chart 2) shows exactly the same data but in the form of rates of change computed from the earlier levels.2 For both the United States and the United Kingdom, there is a remarkable degree of agreement between the rate of change of nominal income and the rate of change of money over a century. The only really big discrepancy occurs for both the United States and the United Kingdom in World War II. You will note that in World War II the money stock rises more rapidly than income in the period of the war itself, and then it goes below it in the post-war period. What you had in both cases was that velocity fell sharply during the war and then rose back after the war. In the United States, the pre-war relation is restored by about 1960; in Great Britain, where the wartime disturbance was much greater, the pre-war relation was not yet restored in the period shown in that chart.

I find that particular deviation fascinating and very instructive because precisely the opposite happened in both countries in World War I: velocity rose during the war, that is, nominal income rose more rapidly than money, and after the war velocity fell. We shall come back to this in a later chart

The natural next question to ask is how the change in money which was reflected in the change in nominal income was reflected separately in prices and output?

Chart 3 parallels Chart 2 except that the rates of change this time are in the quantity of money per unit of output and in prices. That is, if money was growing at 10 per cent per year and output at 3 per cent per year, then money per unit of output would be growing at 7 per cent per year and that is the amount of monetary growth that would be related to prices, which is the other variable in the chart. Again, as you can see, for both countries for nearly a century, there is an extremely close relation between the rate of change of money per unit of output and the rate of change of prices.

The next chart shows velocity in the United States and the United Kingdom. This chart was the biggest surprise to me in the work we have done in this area, and has caused us to turn our study in a very different direction than I initially thought it was going to go. If you look at the levels of velocity, the striking thing is that here you have two different countries—the United States on the one hand and the United Kingdom on the other—yet, except for the initial discrepancy in the early period which I’ll come back to in a moment, the velocities are almost identical; they are practically the same series, one on top of the other. If I showed you one of those series by itself, I would challenge anybody, even if he knew the monetary history of the two countries very well, to know if it was for the United Kingdom or the United States.

The initial discrepancy reflects the much more rapid fall in velocity in the United States than in the United Kingdom until 1905. I believe that this more rapid fall is attributable to the increasing financial sophistication of the United States in that period. We have observed this phenomenon for many countries. In the early stages of economic development, people want to increase the amount of money they hold relative to their income; they want to have a larger stock of money. And this means that in early stages of development, for every 1 per cent increase in real income, you generally have about a 2 per cent increase in the real quantity of money people want to hold. That is so for the United States until about 1905; exactly the same thing holds for Japan in the past 20 years; I suspect that the same thing holds for Israel, if you can eliminate the effect of accelerating inflation, which has tended to work in the other direction.

The bottom part of Chart 4 gives rates of change and so abstracts from the very long-term trend. That too is remarkable.

Very often in economics, good statistical relationships are spurious. For example, if we correlate consumption with income, we are bound to get a good relation because consumption is 70 or 80 per cent or 90 per cent of income, so that we are in large measure correlating a series with itself. In the present case, no spurious statistical element whatsoever is present—and this applies also to the earlier graphs. The income series in each country is statistically independent of the money series, and certainly the series for the U.S. and the U.K. are statistically independent. Yet, for almost 90 years, velocity in the two countries moves together

What does that mean? It means that whatever forces have been producing the changes in velocity have been forces that have been common to the United States and the United Kingdom. In this respect, we have been doing some more studies which show that what happens to the quantity of money in the United States has an effect on the United Kingdom in ways other than through affecting the U.K. quantity of money, and vice versa. There really was one world.

I would like to see such comparisons for other countries, but as yet we haven’t made them. However, some have been made for Canada and the United States and they show almost exactly the same thing. If you want to know what happens to Canadian income, you do better to know what happens to the U.S. money stock than to know what happens to the Canadian money stock.

However, my present point is somewhat different. I think you will agree with me that this set of four charts, if it does nothing else, makes it clear that money and income are not independent magnitudes that can go their own ways, that what happens to the quantity of money is closely related to what happens to income and vice versa

These studies, as I say, are for two countries for a period of nearly a century. Chart 5 also deals with longer period relations but it deals with relations among countries. What you have here are 40 countries. Each dot plots the rate of change in prices over a 17-year period versus the rate of change in the quantity of money per unit of output over the same period. There are 40 dots for 40 countries. The diagonal line across the chart is the line on which 12 per cent change in prices is associated with a 12 per cent change in the quantity of money per unit of output, and so on. You can see that the points are very closely scattered along such a diagonal line. At the top of the chart are the countries that have had about a 30 per cent per year increase in the quantity of money per unit of output, and they have had about a 30 per cent per year change in prices. At the bottom are the countries that have had a small rate of change in money and a small rate of change in prices. Again, there is no escaping the conclusion that is illustrated by that chart—that inflation is primarily a monetary phenomenon and reflects what happens to the quantity of money per unit of output.

The next set of charts shifts from this very long-term view to a much shorter term view. Here we look at the opposite side of the picture—the amount of noise in the relationships. So far, the relationships I have been showing you have been very tight, because they have averaged out the irregularities from day to day, quarter to quarter. Chart 6 is for the United States for quarter-to-quarter changes over the past 17 years. The rate of change in gross national product from quarter-to-quarter is plotted against the rate of change of money (currency plus all commercial bank deposits) two quarters earlier, to allow for the characteristic tendency of GNP to lag money. For example, one of these points shows that GNP grew by 11⁄2 per cent from one quarter to the next quarter while the quantity of money declined at the rate of 11⁄2 per cent two quarters earlier.

There is broadly a positive relationship. However, from quarter-to-quarter there is an enormous amount of noise, an enormous amount of irregularity, which shows up very sharply in Chart 6. I believe that it is extremely important to recognize the limitations as well as the closeness of the relation between money and income. There is a very loose relation between changes in money and income over very short periods, a very close relation over longer periods of time.

In order to eliminate some of that noise, the next chart averages over four quarters. It plots the change from one quarter to a quarter a year later. For example, instead of plotting the rate of change in GNP from the first three months of 1971 to the second three months, it plots the rate of change from the first three months of 1970 to the first three months of 1971. As you can see, averaging over four quarters brings the scatter a lot closer together but it still leaves a very considerable amount of noise.

The next chart shows that even that residual noise is very deceptive, that there is in one way less to it than one might gather from Chart 7 alone. The line in this chart labelled “actual” shows the rate of change of personal income in the United States averaged over a six-month period to reduce random perturbations. The dashed line labelled “predicted” shows the rate of change in personal income predicted from the movement of money nine months earlier. Put differently, that predicted series is the movement of money itself nine months earlier adjusted in scale to correspond to personal income on the basis of a correlation computed for a 17-year period.

Now I have labelled some of those points: 1, 2, 3, 4--because those are points that are explained by exceptional circumstances. Point No. 1 reflects a retroactive pay increase which was all included in one month; point No. 2, the GM strike; point No. 3, the threat of a steel strike, and point No. 4, which is most interesting in another connection that I am tempted to dwell on at length but shall only pass over here, reflects the effect of the price freeze of August 15 which chilled a nicely developing expansion for some months by introducing great uncertainty into the economy.

As you can see, the predicted values go on longer than the actual values because there is another nine months for prediction. As you can see, the series is coming back up to the predicted path after the shock introduced by the wage and price freeze.

If you eliminate those exceptional events, I think you will agree that the chart shows extraordinarily good predictions for a very simple-minded relationship. There is nothing in the chart except the movement of money nine months earlier and the movement of personal income.

Chart 9 shows both the strength and weakness of this relation-ship. It is identical with Chart 8 except that it covers the preceding 15 years, 1953 to 1968. Again I won’t discuss the exceptional numbered points. I want to stress something very different. From about 1958 on, the relationship is just as close as in Chart 8. On the other hand, for the first five years, the relationship is much looser. The pattern of the two series is the same but the actual personal income series fluctuates much more widely than the predicted series.

I have no good explanation for this divergence. I conjecture that it has something to do with the system settling down after the shock of the Korean War. But in any event, including both periods gives an indication of both the strength and the limitations of this relationship.

The next chart shows the same relationship separately for output and for prices. The upper panel shows the relationship of money to industrial production six months later. One of the things that we found and that I shall discuss below, is that a change in the quantity of money has an effect on output much earlier than it does on prices. While I used a nine-month lead for nominal income, a six-month lead is best for output alone. The panel shows a very close relation although a slight difference in level. For prices for the United States for this period, it turned out that a 23 month lead of money over prices gives the best results. As you can see, that relation is not nearly as close over short periods as the relation for output and the delay is much longer.

I didn’t want to give a lecture in Israel without having at least a few numbers for Israel, so Anna Schwartz managed to scrounge up some numbers for me in New York on Israel. The next and final chart shows the same kind of a relationship for the last few years in Israel. The solid series, based on rough estimates of your current GNP on a quarterly basis, shows the actual rate of change from quarter to quarter (the U.S. charts use monthly data for a six-month period). The dashed line shows the rate of change of GNP predicted from the change in high-powered money two quarters earlier. I am not sure what the best monetary aggregate is for Israel. When I studied the Israeli situation in 1962, I decided that currency was most reliable because your authorities were changing the meaning of deposits so rapidly that the deposit series had very little economic significance. I suspect that may still be the case. In any case, that is why we used high-powered money. The sharp decline and then rise in the rate of growth of GNP corresponds to the Six Day War. Note that that first period is rather like the early years in Chart 9 for the U.S.—the general pattern is the same for the actual and predicted, but the fluctuations in actual income are much wider. The second period is like the later period for the U.S. data—the two series are much closer. Towards the end of the period, the actual GNP is starting to diverge from the predicted. I suspect that this reflects what is happening to your rate of inflation. Such a divergence is precisely what accelerating inflation could be expected to produce.

At any rate, the relationships I have described are not relationships that prevail exclusively for the United States. They are relevant to Israel as well as to other countries.

That is the end of the slide show.


1. Nominal income is net national product; money is currency outside commercial banks plus all deposits (time and demand) of the public at commercial banks.
2. Each rate of change is computed as the slope of a least squares straight line between the natural logarithm of the variable (income or money) and time fitted to three successive phase averages.
3. This summary is adapted from Milton Friedman, The Counter-Revolution in Monetary Theory, Occasional Paper 33 (London: Institute of Economic Affairs, 1970), pp. 22-26.
4. See my "The Optimum Quantity of Money" in The Optimum Quantity of Money and Other Essays (Chicago: Aldine Publishing Co. 1969).
5. "Government Revenue from Inflation, "Journal of Political Economy, Vol. 79 (July/August, 1971), 846-856.

Reprinted from Money and Economic Development The Horowitz Lectures 1972 by Milton Friedman, copyright © 1973

Back to 'Money and Economic Development' Home