Granger Causality
A statistical hypothesis test to determine whether one time series is useful in forecasting another.
What is Granger Causality?
A variable X is said to 'Granger-cause' Y if past values of X contain information that helps predict Y above and beyond the information contained in past values of Y alone.
Why Granger Causality Matters
In time-series econometrics, true causality is difficult to prove. Granger causality provides a rigorous empirical test of predictive causality—which variable temporally precedes and forecasts the other.
How to Interpret
A significant result means X has predictive value for Y. It does NOT prove strict philosophical causality.
Example
If changes in consumer sentiment indices consistently happen a month before changes in retail sales, and knowing the sentiment improves the forecast of retail sales, then consumer sentiment Granger-causes retail sales.
Common Mistakes
- Interpreting Granger causality as true structural causation. It is merely a test of temporal precedence and predictive ability (e.g., lightning 'Granger-causes' thunder, but Christmas card sales might 'Granger-cause' Christmas, which is structurally false).