Will the impulse response function and autocorrelation function of an AR(2) coincide? I say that they will not, because $IRF_{t}(t+k)$ (that is, the effect of a unit shock at time $t$, on the process at $t+k$) will be 1 when $k=0$. However, $corr(y_{t}, y_{t})$ is the variance of the $y_t$ process, and will not necessarily be 1. Thank you.

  • $\begingroup$ Hi: The autocorrelation at lag zero is, by definition, the scaled variance of $y_t$ which is 1.0. Recall that, in general, the auo-correlations at any lag are just scaled covariances; i.e: covariance divided by the variance and the covariance at lag zero is the variance. $\endgroup$
    – mark leeds
    May 22, 2021 at 13:31
  • $\begingroup$ assume that there is no constant, so that the autocovariance at lag 2 is $E[y_{t+2}y_t]$ and verify your claim $\endgroup$ May 22, 2021 at 13:42


Your Answer

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

Browse other questions tagged or ask your own question.