I am using E-views and estimating GJR-GARCH (i.e. GARCH(1,1) with a Threshold Order of 1). The data in question is Daily returns for the ASX200 index. I am only using a constant in my my mean equation with no ARMA terms.

The formula E-views uses is as below.

GARCH = C(2) + C(3)*RESID(-1)^2 + C(4)RESID(-1)^2(RESID(-1)<0) +

The output yields a very high p-value for RESID(-1)^2. It yields 0.8900 to be exact. What could be the cause for this and is there a reason to worry, or a way to correct it?

EDIT: Have included the equation output, and the estimation command given in E-views enter image description here enter image description here

  • $\begingroup$ Perhaps you could add a picture with the output? (It is not strictly necessary, but maybe there is something interesting to find.) $\endgroup$ – Richard Hardy Apr 6 '17 at 13:09
  • $\begingroup$ @RichardHardy Have edited with the pictures added. $\endgroup$ – Albe Apr 6 '17 at 14:04
  • $\begingroup$ @RichardHardy Any thoughts Richard? I've read an interpretation elsewhere, unsure how correct. Basically the asymmetry term (3rd term in the variance equation) is positve and highly significant indicating an asymmetry effect. However, Resid(-1)^2 (or Alpha) is insignificant because positive shocks have no effect on the performance of the index. And this positive effect is measured by Alpha, hence why the coefficient is negative and so low. Basically, the coefficient and p-value are working together. Conclusion: bad news has a greater effect on volatility than good news (Alpha + Asymmetry term) $\endgroup$ – Albe Apr 7 '17 at 18:59
  • $\begingroup$ That sounds more or less right. Note that you have $u_{t-1}^2$ and $u_{t-1}^2\times I(u_{t-1}<0)$ rather than $u_{t-1}^2\times I(u_{t-1}>0)$ and $u_{t-1}^2\times I(u_{t-1}<0)$, and the two are not the same. $\endgroup$ – Richard Hardy Apr 7 '17 at 19:18
  • $\begingroup$ Could you kindly elbaorate on your last comment, please. What are the implications? I have simply computed the output as per the software and Brooks Introductory Econometrics. $\endgroup$ – Albe Apr 7 '17 at 19:27

Your Answer

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

Browse other questions tagged or ask your own question.