Stationarity and unit roots
What stationarity means, why a random-walk unit root breaks regression, and how differencing or detrending restores it.
A 3 minute 19 second animated lesson on stationarity and unit roots. Built for ECON3006 Economic & Financial Modelling at Western Sydney University.
Weak stationarity means three things stay stable over time: a constant mean, a constant variance, and an autocovariance that depends only on the gap between observations, not on where you sit in the sample. The video contrasts a stationary, mean-reverting series with a random walk, the canonical unit root, where shocks are permanent and the variance grows without bound. That growing variance is exactly what the augmented Dickey-Fuller test is built to detect.
It closes on the practical rule: test for stationarity before you trust a regression, and restore it by differencing a unit-root series or detrending a deterministic trend. Pair it with the Atlas concept page for the formulas, a worked Stata example, and a quick quiz.