2
$\begingroup$

I want to estimate granger causality between two series. Visual inspection indicates it might be useful to normalize data first (i.e. (X-mean(x))/ (sample stdev(x)) )

Are there some caveats associated with that?

$\endgroup$
1
  • $\begingroup$ Why do you normalize your data? For example, are you trying to avoid computational issues due to very different scales of the original variables? If not for computational issues, normalization should not affect inference on Granger causality just as it would not affect inference on coefficient significance in a multiple regression. $\endgroup$ Mar 25 '15 at 14:57
2
$\begingroup$

Normalising your data does not affect statistical inference in a regression model (see here When conducting multiple regression, when should you center your predictor variables & when should you standardize them?). Hence, you can just carry out tests for Granger causality as per usual.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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