Causal Inference

Causal inference is the branch of statistics concerned with estimating the causal effect of one variable on another — not just their association. A regression coefficient is causal only if you have ruled out confounders (other things that vary with $X$ and affect $Y$), reverse causality ($Y$ causing $X$), and selection (non-random assignment of $X$).

When to use

This is the framing for every applied-econometrics problem in the course. The toolkit available to you grows with each lecture: DAGs to identify confounders, randomisation when you can run experiments, Instrumental Variables when an exogenous shifter exists, Fixed Effects for repeated observations of the same unit, Difference-in-Differences for staggered treatment, and Regression Discontinuity for sharp cutoffs. Which tool fits depends on the source of exogenous variation in your data.