8.2 Spurious Regression

This is about having nonstationary data, where the nonstationarity is able to prove a relationship between a dependent variable and independent variable(s).

Running a regression on this, will not make any sense, as it is an effect that is not included in the model, that is actually proving the relationship. Meaning that you don’t know if the relationship is actually true or not, but you are apparently with a statistical test able to prove it. Hence we get a spurious regression if:

1. The variables have another variable in common, but it is not included in the model.
2. It is a coincidence