Cointegration
A statistical property where two or more non-stationary time series share a long-term equilibrium relationship.
What is Cointegration?
If two non-stationary (trending) variables move together over time such that a specific linear combination of them becomes stationary, they are cointegrated. They may drift apart in the short term, but fundamental economic forces bring them back together in the long term.
Why Cointegration Matters
It provides a mathematically valid way to run regressions on non-stationary variables without suffering from the spurious regression problem, allowing researchers to model true long-run economic relationships.
How to Interpret
If variables are cointegrated, researchers often use an Error Correction Model (ECM) to analyze both their short-term dynamics and long-term equilibrium.
Example
The price of oil and the price of gasoline are non-stationary. However, because gasoline is refined from oil, their prices cannot drift infinitely far apart. They are cointegrated.
Common Mistakes
- Confusing correlation with cointegration. Two completely unrelated variables (like US GDP and global temperature) can be highly correlated simply because both grow over time, but they are not cointegrated.