Comparing coefficients across models using Newey-West?

[Karun Advani]

Hi all,

I currently have multiple time-series regressions with Newey-West standard errors (as heteroscedasticity and autocorrelation are present). I need to conduct a pairwise Wald test across the different models to test for the equality for a specific independent variable. The command "suest" does not seem to work with Newey-west standard errors. Also, when trying to use cluster errors around the model ID, all standard errors disappear. Please advise as I need to conduct the test but in a way that would use standard errors that take into account heteroscedasticity and serially correlated errors.

Thanks

[Attaullah Shah]

See this post, and the z-test in #4 if that can be of any help http://www.statalist.org/forums/foru...oss-the-models

[Karun Advani]

Hi Attaullah,

Thank you for your reply. As you described in the referred post, it seems to me a manual calculation. Would you be able to recommend how we could carry this out if a Wald-test for the joint-significance (equality) of multiple variables needs to be carried out (chi2 distribution)?