2
$\begingroup$

I am trying to understand how to identify lag length to use for a Granger Causality test. The process as I understand it is:

  1. Use an information criterion such as AIC or BIC to calculate the number of lags to use for each time series.
  2. Choose the larger of the two lags.

I am having trouble understanding how you do step 1 of this process. I program in Python, and am trying to implement this process into code.

$\endgroup$
  • 1
    $\begingroup$ Have you read Dave Gile's blog posts on testing Granger causality? (They will not help you with coding but with lag order selection.) $\endgroup$ – Richard Hardy Nov 17 '16 at 6:23
  • $\begingroup$ You can have a look at this document: stat.wisc.edu/courses/st333-larget/aic.pdf It shows how to do the implementation in R, but the idea should be easily portable to Python. $\endgroup$ – Matthias Schmidtblaicher Nov 17 '16 at 10:18
0
$\begingroup$

Partially answered in comments:

Have you read Dave Gile's blog posts on testing Granger causality? (They will not help you with coding but with lag order selection.) – Richard Hardy
https://davegiles.blogspot.com/2011/04/testing-for-granger-causality.html

You can have a look at this document: http://www.stat.wisc.edu/courses/st333-larget/aic.pdf It shows how to do the implementation in R, but the idea should be easily portable to Python. – Matthias Schmidtblaicher

$\endgroup$
  • $\begingroup$ The second link you provided doesn't work anymore. Would you happen to know what it has changed to? $\endgroup$ – altabq Oct 17 at 18:36

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.