Skip to content
Time-Series
Econometrics
Ian Helfrich PhD ↗

Lecture 6 summary · February 19, 2027

Trends, unit roots, and misleading level regressions

Complete chapter · Laboratory ch06 · Practice · All summaries

The economic question

A deterministic trend with temporary deviations differs from a stochastic trend with permanent innovations. Their variances and testing references differ.

Central relation

Yt=Yt−1+d+εt,Y_t=Y_{t-1}+d+\varepsilon_t,
Y^t+h∣t=Yt+hd,\widehat Y_{t+h|t}=Y_t+hd,
Vh=hσ2 V_h=h\sigma^2

Individual problem

A random walk has drift 0.3, current level 20, and innovation variance 1.5. Find its four-period forecast and variance. Interpret a unit-root nonrejection.

Worked reasoning

The forecast is 21.2 and variance is six. Nonrejection is not proof of a unit root; power and deterministic specification matter.

A consequential failure

An ordinary t reference is applied to a DF statistic, or a levels regression of one random walk on another is read with conventional t statistics.

Repair

Use a deterministic-case reference. Draw Monte Carlo replications from one stream, and report their uncertainty, initialization, sample size, and number; retain weak-correlation realizations too.