Monday, September 12, 2011


Jazzbumpa left a comment linking to a PDF on "Money Growth and Inflation".

I read the PDF and responded in a handful of posts beginning here.

Clonal picked up on this one, and looked at the

correlation between the annual percent change in the CPI, and the annual percent change in M1 -- guess what!

A correlation coefficient of close to zero (0.07 to be exact Rsq of 0.006) In other words no explanatory power at all.

Clonal also provided a couple versions of a graph showing what he was looking at. I called them star clusters:

Graph #1: CPI v M1, 1960-2011

I never do scatter plots. Never saw a use for them, I guess. But these graphs were relevant, and I started paying attention. Then Clonal came back with more graphs.

The energy started to build.

Of the three graphs below, Clonal writes:

M2 shows a high negative correlation with CPI. This really kills the Monetarists.

Separate the years 1960-1970; 1971 to 1983 and 1984-2011. You get three very interesting correlations, all negative.

R = -0.38, R = -0.75, R = -0.20

Graph #2: CPI v M2, 1960-1970

The axis-values vary on these graphs, because the data points tend to be lower on one, higher on another. You can click the graphs to see them bigger and check the numbers.

Graph #3: CPI v M2, 1971-1983

For the record, the highest CPI values on Graph #3 are twice as high as those on Graphs #2 and #4. The highest money-growth values are higher also, by about half as much.

Graph #4: CPI v M2, 1984-2011

At that point Jazzbumpa came back into the conversation, adding his perspective and dialing up the energy:

If inflation is dollars chasing goods, then the relationship to the money measure type is exactly backwards. Base money and M1 are basically money in circulation, and therefore involved in the chase. The higher numbered aggregates include less and less active money. How can those dollars be involved in goods-chasing, and therefore in inflation?

I'm beginning to think that nobody has any understanding of inflation.

And while I was writing this, Clonal came back with correlation numbers for MZM:

1960-2011 r = -0.24
1960-1970 r = -0.78
1971-1983 r = -0.66
1984-2011 r = -0.23

So let me get in the game and finish up this post, so I can go watch the Jets lose win their season opener.

I took a look at Total Debt versus the CPI, using annual numbers and change-from-previous-year values from FRED. And using an OpenOffice spreadsheet.

If you drew a trend-line through these data points, the line would slope up to the right. This would be a positive correlation value, I think.

Graph #5: CPI v TCMDO, 1950-2010
In Graph #5 the data points are color-coded by decade. For Graph #6 below, I figured average values for each decade and just plotted the averages to clean up the picture a bit. The color codes are the same, except I mixed up the 1970s and the 1980s. Oops.

Graph #6: CPI v TCMDO, Averages by Decade
So, I do scatter plots now.

Excel spreadsheet (created in OpenOffice)
Google Docs spreadsheet (no graphs)

I'm out.

No comments: