Metrics Day: Granger Causality

I kinda liked the way I explained Omitted Variable Bias in the previous post, so let's call today metrics day: A reader recently asked me to explain Granger Causality in a way anyone on the street would understand it. Here is what I had to say:

"A Granger causes B if A always comes before B". I am aware that it is quite a bit of simplification, and it is not entirely accurate. But it does capture, I believe, the essence, and it also lets you to understand the main problem with Granger Causality: If an alien came to Earth and wanted to find out what caused Christmas, and all the data he had was the date of Christmas cards, she would conclude that Christmas is caused by Christmas cards:)

Anyway, if you want to teach yourself metrics, I would strongly recommend the Peter Kennedy classic as well as the new Harmless Econometrics book.

As a side point, I should note that metrics is anything but harmless, and although I have taught time series to undergraduates, I still do not consider myself even near mastering Econometrics. BTW, the reader's response to this last comment was "thanks for the motivation":):):)