For part of a project, I wanted to see if electricity consumption causes GDP in Colorado. I initially intended to follow the approach of Mozumder and Marathe (2007), who use a VECM approach, but that requires cointegration.

I'm not sure what to do now. I tried the Toda-Yamamoto method but my errors seem to be serially correlated...


1 Answer 1


Causality between time-series variables does not require the two to be cointegrated.

First, cointegration requires that each series be $I(1)$. It is certainly possible for two $I(0)$ series to follow a causal relationship (or two $I(d)$ variables for that matter).

Second, cointegration implies a long-run equilibrium among the series, which is not required for a causal relationship. A long-run equilibrium between two time series variables $x$ and $y$ implies that $$y_t-a-bx_t$$ is a stationary process. In other words, if shocks are 'turned off', the values of $x$ and $y$ will converge such that $y_t=a+bx_t$.

  • 2
    $\begingroup$ I would suggest to rethink the part on Granger Causality. "Granger Causality" has nothing really to do with a causal relationship, it i just about predictive power, see economics.stackexchange.com/a/3056/61 $\endgroup$ Mar 22, 2019 at 18:41
  • $\begingroup$ You are right. I edited the post. Interesting discussion on the thread you posted. The statement 'weathermen Granger-cause the weather' is very illuminating about what Granger-causality really tells us. $\endgroup$
    – dlnB
    Mar 22, 2019 at 18:42
  • $\begingroup$ Yes, that was an excellent remark by John Cochrane. $\endgroup$ Mar 22, 2019 at 19:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.