Question

Causality between two variables:

Describe how we test for causality and why this might be important in conducting empirical analysis.

Answer 1

We have,

NO regression technique, NO statistical analysis at all can test a causal relationship. Causality is no property contained in the data. The ONLY way to adress causality is to perform a controlled experiment, where you know, a priory, that only the arbitrarily changed condition (and nothing else) can be responsible for a possible change in the response. If you have only observational data (what is often the case in econometrics), you can only speculate or hypothesize about causal relations.

In an empirical analysis to check causality between response variable and one or more independent varaible is addressed by using causality test.