Autocorrelation and heteroskedasticity in Panel Data

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

NimUnited
Posts: 4
Joined: Sun Mar 18, 2018 11:26 am

Autocorrelation and heteroskedasticity in Panel Data

Postby NimUnited » Sun Mar 18, 2018 11:22 pm

I am using Eviews 9, and I am having some troubles with panel data. I have a panel data set of 400 Companies with observations from 2007-2016. It is almost balanced, but not totally, i.e. I have some missing observations now and then.

I have 2 questions:
1. How do I test for autocorrelation and heteroskedasticity in my dataset? I can't seem to find the same tests as with cross-sectional.
2. If autocorrelation and heteroskedasticity exists, how do make my standarderrors robust to it?

I have been told, that Eviews can't handle the above questions, and that I should use Stata, but I think that sounds a bit weird? However I do not seem to find any guides on how to deal with it on Eviews online-forum.

When I do my panel regressions I can choose White Period standarderrors, but I am not sure that they do the job I want them to do in this particular example? (question 2), and it does not say anything about whether autocorrelation and heteroskedasticity actually exists (question 1).

I hope someone can help me.

Best regards.

NimUnited
Posts: 4
Joined: Sun Mar 18, 2018 11:26 am

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby NimUnited » Tue Mar 20, 2018 11:12 am

Do anyone has an answer for this? I truly hope to get some help.

Best regards

sarchi
Posts: 19
Joined: Wed Jan 24, 2018 9:10 am

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby sarchi » Thu Apr 19, 2018 9:45 am

Hi,

That is my question too. Can anybody help us?
I use EViews 10, but in this version ,is there any possibility for white test. I want to solve Hetroscedastisty Problem with white, but Cent find it.
Is it enough , if we choose white cross-section In Equastion Estimation, in Panel Option ? or should I choose fix Effect in Effect spesification too ?

Im looking for any answer please!!!

Thank you

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11725
Joined: Tue Sep 16, 2008 5:38 pm

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby EViews Gareth » Thu Apr 19, 2018 9:54 am

The standard introductory approaches to autocorrelation and heteroskedasticity don't apply in panel data. Thus there is no "White test" or simple "White standard errors". The idea of heteroskedaticity is more complicated in a panel setting - is the covariance between errors within cross-sections only, within period only, or a mixture?

Thus the options for testing and adjusting covariances are different. You have to figure out what you want to correct for, then choose the appropriate covariance option.
Follow us on Twitter @IHSEViews

sarchi
Posts: 19
Joined: Wed Jan 24, 2018 9:10 am

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby sarchi » Thu Apr 19, 2018 10:25 am

Thanks for your answer,

I have workfile mentioned below. Durbin-watson stat is near zero. That means there is Heteroscedastisity here. What can I do to handle Heteroscedastisity? is it enough to Choos "white cross- section" in "Coef covariance method " ?

Thank you so much
Attachments
Hetros.tiff
Hetros.tiff (33.54 KiB) Viewed 163 times

sarchi
Posts: 19
Joined: Wed Jan 24, 2018 9:10 am

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby sarchi » Thu Apr 19, 2018 11:06 am

white diagonal means correct both cross-sections and within period ?

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11725
Joined: Tue Sep 16, 2008 5:38 pm

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby EViews Gareth » Thu Apr 19, 2018 11:53 am

Follow us on Twitter @IHSEViews

sarchi
Posts: 19
Joined: Wed Jan 24, 2018 9:10 am

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby sarchi » Thu Apr 19, 2018 12:46 pm

thank you very much. that was very helpful.
but now i have another question.
"Ignoring cross-sectional dependence in estimation can have serious consequences, with unaccounted for residual dependence resulting in estimator efficiency loss and invalid test statistics."
what should I do, wenn I have cross-section Dependance in my Panel data ?
in EViews Help just says how can we test it. But did not say how can we solve it.

Thanks a lot

NimUnited
Posts: 4
Joined: Sun Mar 18, 2018 11:26 am

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby NimUnited » Fri Apr 20, 2018 12:39 pm

Hi Gareth,
We have clustering in both cross-section and in time. We use yearly time-dummies to fix the time-clustering. Would you recommend White Period to fix cross-section clustering?
Thanks in advance.

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11725
Joined: Tue Sep 16, 2008 5:38 pm

Re: Autocorrelation and heteroskedasticity in Panel Data

Postby EViews Gareth » Fri Apr 20, 2018 12:45 pm

I do not give econometric advice
Follow us on Twitter @IHSEViews


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 3 guests