hetroscedasticity in panel data

[kazmi89]

dear members can you please tell me how i can test hetroscedasticity in panel data fixed effect model? i am using eviews 8 demo version

[EViews Gareth]

EViews does not have built in heteroskedasticity tests for panel data.

[kazmi89]

EViews does not have built in heteroskedasticity tests for panel data.

then sir how i can test it manually in eviews?

[EViews Gareth]

You'll have to calculate whatever test you want to perform by hand.

[kazmi89]

You'll have to calculate whatever test you want to perform by hand.

thank you sir

[adc]

Dear,

In order to test for heteroscedasticity, you can change the structure of your workfile from panel data to unstructured and perform the White heteroscedasticity test. However, if I want to correct for heteroscedasticity when using panel data, I should know whether I have cross-section / period / contemporaneous heteroscedasticit. How can I know/which test can I run on my unstructured data in order to know this?

Many thanks in advance.

[kazmi89]

Dear,

In order to test for heteroscedasticity, you can change the structure of your workfile from panel data to unstructured and perform the White heteroscedasticity test. However, if I want to correct for heteroscedasticity when using panel data, I should know whether I have cross-section / period / contemporaneous heteroscedasticit. How can I know/which test can I run on my unstructured data in order to know this?

Many thanks in advance.

dear can you please comment on the following procedure which i have done for white test.
first of all i got the residuals from fixed effect estimation after that i converted my file into unstructured and then i regressed square of those error term(which i obtained from fixed effect estimation) on my independent variables, their squares and their cross products. my procedure is ok or not?

[joshchipunza]

i would like to ask how i can use Eviews 8 to correct for heteroscedasticity and clustered errors in a fixed panel model.

[joshchipunza]

I meant to ask how Eviews 8 can be used to correct for heteroscedasticity and correct for clustered standard errors in a fixed panel model.
Could anyone please assist me in this regard because i am stuck with my methodology

[EViews Glenn]

http://forums.eviews.com/viewtopic.php? ... ster#p1191

Also see the discussion in the manual of weighted least squares in panel estimation and related topics.

[sihas]

Dear Sir,

I have followed previous posts and kindly ask you to reconfirm if the White test for heteroskedasticity using unstructured dataset (by converting panel struture to unstructured) is valid for the original dated panel. Should it be the case, how can I detect the type of heteroskedasticity in Eviews 8 and also test for correlation?

I have a panel with cross section fixed effects, where T<N. Can I use cross section weights and White diagonal (instead of white period) ?
How can I find the optimum choice?

Many thanks for your time and kind reply.

Kind regards,
Vivi

[sihas]

Dear Members,

In a panel with cross section fixed effects, where T=14 nad N=27, can I use cross section weights and White diagonal (instead of white period because period weights don't work with fixed effects and white period yields a warning message concernig the rank of the covariance matrix) ? Can I simply disregard the warning? (by the way, I followed your advice concerning the commands wgt=persur @expand(c,@crossid) and near singular matrix is displayed.

I really hope you can help!

Thanks a lot!
V.