On Granger causality

On Granger causality

Many years ago, I learned the concept of Granger causality. How can a statistical test determine causality? I wondered.

Recently, I learned that Granger causality is not a causality. In the words of Granger himself, 

It is doubtful that philosophers would completely accept this definition, and possibly cause is too strong a term, or one too emotionally laden, to be used. A better term might be temporally related, but since cause is such a simple term we shall continue to use it. (Granger and Newbold, 1977, p 225)

Granger causality is really temporally related, not causality. But the term causality makes the research findings more prominent and makes the researchers themselves more prominent. Causality implies many policy implications littered in research papers. This term becomes extremely popular over time.

In general, an idea becomes popular not because it promotes truth, but because it promotes the advocates of the idea. This is the case in any human activities, including science.

Reference

Granger, C. W. J.; Newbold, Paul (1977). Forecasting Economic Time Series. New York: Academic Press. p. 225