Differencing two related price levels removes the very equilibrium a relative-value trader needs. Cointegration identifies the exceptional I(1) regression whose residual is I(0): an equilibrium error with stable moments even though both legs wander. Engle--Granger is the quick bivariate route from tested levels to that residual.
Learning objectives. By the end of this note you should be able to:
- define cointegration and a cointegrating vector;
- execute and interpret both stages of Engle--Granger;
- use the correct residual-test critical values;
- explain superconsistency and the procedure’s asymmetry.
In this lesson
- A stationary relation between nonstationary levels
- The two steps
- From a tested relation to a usable model
Before you start
The full lesson, with worked examples and exercise solutions, is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Econometrics and Financial Time Series syllabus.