On this website: https://people.duke.edu/~rnau/arimrule.htm I have found information that if the lag-1 autocorrelation is zero or negative, or the autocorrelations are all small and patternless, then the series does not need a higher order of differencing. If the lag-1 autocorrelation is -0.5 or more negative, the series may be overdifferenced. Does lag-1 here refers to the first lag or some "lag minus one"?


1 Answer 1


Let the time series be $x_t$, then "lag-1 autocorrelation" is $$\rho(1)=\frac{\operatorname{cov}(x_t,x_{t-1})}{\operatorname{var}(x_t)}$$

  • $\begingroup$ Could you please tell me whether these time series with the 1st order of difference are already overdifferenced? imgur.com/jsYTpIL With 2nd order difference I get the same results except that the first lag is positive instead of negative. Could you explain what this means? $\endgroup$
    – domke89
    Commented Feb 14, 2021 at 17:13
  • $\begingroup$ Hi @jnan12 it’s better be asked as a separate q. for others to comment as well. $\endgroup$
    – gunes
    Commented Feb 14, 2021 at 21:08

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.