I am a bit confused about why would I use the Ljung-Box test in order to determine whether a series is a white noise ?

I know that the Ljung-Box test studies whether we have serial correlation. I also read that weak stationarity implies no serial correlation.

So can anyone explain to me the relation between these three concepts: serial correlation, white noise and weak stationarity?

Thank you!

  • $\begingroup$ White noise, autocorrelation, and stationarity are quite disjoint entities. Read first if all what I.I.D. means that clarify your question at least. $\endgroup$ Sep 3, 2019 at 17:18
  • 2
    $\begingroup$ "weak stationarity implies no serial correlation" is an incorrect statement and probably the source of your doubt. Just Google. You'll get enough explanation. $\endgroup$
    – Dayne
    Sep 4, 2019 at 1:09

1 Answer 1


The Ljung_Box test is a global/cumulative test for all estimated autocorrelations. It assumes that the error process is free of latent deterministic structure (pulses,level shifts,local time trends and seasonal pulses. If these assumptions are ok with you then use the test.


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.