Thursday, September 29, 2011

But who's counting?


The third graph from my four o'clock, repeated here as Graph #1:

Graph #1: Corporate Debt 1949-2011

Quarterly data from FRED, fed into OpenOffice, plus an exponential trend...

Graph #2: Same debt, with an exponential trend line

... with an R-squared greater than 0.9899, for the record.

3 comments:

Clonal said...

Art,

Just for the record, R sq are good where you have many y's per x, and many x's per y. They tend not to give much information if you have only one y per x

In the graph you show on the business debt, the trend is definitely not exponential. It is growing, but not exponentially - the divergence is too great since 2000.

Plotting on a log scale will tell you about trend divergence.

The Arthurian said...

Hi, Clonal. I majored in math, but never took statistics. I guess you can tell, huh.

Thanks for setting me straight.

LiminalHack said...

"It is growing, but not exponentially - the divergence is too great since 2000."

The exponential prior to 2000 is still remarkable though. The post 2000 period might be considered a settling period following a fast transition such as the ringing seen on a wire driven from 0V to 5V.

The deviations prior to 2000 are clearly demographically driven.

The post 2000 behavior is much more about mature economies reaching peak debt and then hunting for some kind of equilibrium.