Milton Friedman, Hack & Fraud

by Keynesian of sorts

Milton Friedman was not above using sleight of hand and sophistry to ensure his work reached the preferred conclusion. As Serban V.C Enache has shown before in this article, Friedman exaggerated the stability of the velocity of money to support his theory of inflation derived from the quantity theory of money. In light of this, I will highlight another example of Friedman’s use of sophistry to hide the empirical issues the monetarist theory runs into.

The work of Milton Friedman and Monetarism more widely is heavily based on the quantity theory of money which can be expressed by:


Where M is the money supply, V is the velocity of money, P is the general price level and Q is real GDP. Using some basic algebra this equation can be rewritten so it takes the form of:

1/V = M/PQ

Real GDP is given by the formula:

Q = NQ/P

Where NQ is nominal GDP. If you substitute equation (Q=NQ/P) into (1/V=M/PQ), it cancels out and the ratio can be expressed as:

1/V = M/NQ

One of Friedman’s major claims was that the money supply relative to output is a very good approximation of a long run price trend. However, if we graph the M2 measurement of the money supply relative to nominal GDP, we find that the graph does not resemble a price trend at all.

This was a real issue for Friedman’s theory because if the monetary system and the economy operated in the way the quantity theory of money claimed, then a price trend that aims upwards should be explained by increases in the money supply that outpace increases in nominal output.

To overcome this setback, Friedman simply graphed the money supply relative to real output as opposed to nominal output. When graphed, it looks like this:

This does look like a long run price trend, so Friedman and monetarism were saved from any critique that claims the money supply and general price level weren’t related in a proportionate positive way? Well, not quite. Friedman’s claim was that equation (1/V=M/NQ) should resemble a long run price trend, but the second chart doesn’t graph this equation. What Friedman is actually graphing is:

1/V = M/Q

Substituting in equation (Q=NQ/P) and simplifying we get:

1/V = MP/NQ

This equation features the general price level term P. So to make the ratio of the money supply relative to output look like a long run price trend, Friedman had to include a price level term in the equation so the trend looks the way it does simply because a price level tautologically appears in equation (1/V=MP/NQ). Of course, Friedman didn’t explicitly say that a price level term was included, but simply said that he was graphing the money supply relative to real GDP and therefore no one questioned his work.

The lesson here is that it can be very easy to manipulate data so that the researcher gets the conclusion he wants. Don’t take a graph as final proof of an argument until the data underlying that graph has been properly looked at.