Please turn on JavaScript

Brooks Wilson's Economics Blog: Mankiw, Krugman, Barro and Causation

Thursday, September 15, 2011

Mankiw, Krugman, Barro and Causation

Economists develop models that imply causation and test them with statistics that measure correlation.  As everybody with statistical training knows, including economists, correlation does not imply causation.  It is an easy mistake to make and in a blog post Greg Mankiw argues that Paul Krugman falls victim to this error in his critique of an article by Robert Barro.  

The problem that Paul glosses over is that correlation does not imply causation.  Paul appears to jump to the conclusion that this correlation establishes that the business cycle is the driving force behind investment spending.  But it could just as easily be the opposite (or a third factor driving both).  I am completely confused as to why Paul thinks this graph establishes much of anything at all.

The points of this post are that causation is difficult to establish and that economists expend a great deal of energy at it.

1 comment:

  1. Calin O. Baban said...

    Just because two events happen to happen together does not necessarily mean that one caused the other to happen. Therefore, correlation does not always imply causation. It would be too hard to establish if it did. However, certain phenomena like unemployment and inflation would be a good example where one caused the other. In other words, changes in one variable may or may not influence the operation of the other variable.