Variance Ratio- homoscedasticity

[richa61416]

I am trying to do my dissertation, I notice most persons include the homoscedasticity results in their testing.
Based on the results from Eviews it looks like heteroscedasticity may be default. How can I perform my Lo-MacKinlay variance ratio testing assuming homoscedasticity?

Please let me know. Thanks.
Also I notice that no critical values are given for the Variance ratio test for the Z value, in the same way as the ADF test?
Any reason for this? Should I just use the p-value for my conclusion?

[trubador]

Uncheck the "Use heteroskedastic robust S.E." checkbox.
Yes, you can use p-values. I would also try Bootstrap to make sure that results are in line with one another.

[richa61416]

Ok. I did that. No comparing my Heteroskedasticity implies the series is a martingale, but my Homoscedasticity implies that the hypothesis be rejected since the p value comes to .0045 compared to heteroskedasticity with a value of .1619.

Which one is more conclusive to use. Does this mean the dataset on a whole cannot be termed a random walk? Since Homo and Heter do not agree?
Do I have to use boot strapped for testing the Homo concept in eviews? That would make them align, based on what I saw.

Also how come it test for a martingale(unbiases Random Walk) and in this case if Homoscedasticity, a Random walk?
Do I need to be concerned or overall there are both really testing for random walk, but one of them assumes it is biased?

[trubador]

The rejection of hypothesis in the first attempt is very much likely due to not accounting for the heteroscedasticity. The results of heteroscedasticity test also imply that. Although they test the same thing, the results do not have to agree as they have different assumptions regarding the data. If your data in fact not have a constant variance, then you should go with the results of second test.

Scedasticity and random walk are related, but two different concepts. And there is really (much) more to random walk than meets the eye.

[richa61416]

Ok. Just to be clear. By go with the second test you mean the one testing for Heteroscedasticity.

[trubador]

Yes, that is correct.

[richa61416]

Thank you so very much.