How to Correct for Heteroscedasticity and Autocorrelation IN THE SAME REGRESSION COMMAND in a Fixed Effects Panel Data Model?

[Anou shka]

Hello
I'm running a panel data regression, with 5 independent variables and 28 firms over 5 years. I have 140 observations.

After running a Hausman test, i found that a FE reg is to be used. Next I tested for heteroscedasticity - using the Cook-Weisberg httest for residuals - and autocorrelation - using the xtserial command for panel data. Both turned positive. My data is characterized by both heteroscedasticity and autocorrelation.

I then looked for ways to correct for them. I learned the following:
>> heteroscedasticity - use robust (eg. xtreg dep, var1, var2....., fe vce(robust))
>> autocorrelation - use Cochranne Orcutt method (prais dep, var1, var2...., corc)

But I need to correct them simultaneously in a single regression. I did found something using google :the Newey-west method...
I ran newey dep, var1, var2....., lag (1) force

My questions and problems are as follows:
> How to obtain r-squared when running newey?
> is there another more efficient way of correcting for both autocorrelation and heteroscedasticity?
> should i run dfuller to ensure stationarity? if yes, how to interpret dfuller please?

Any help would be much appreciated....

[Carlo Lazzaro]

Anou: Example 3 under -xtreg- entry in Stata 13.1 .pdf manual, recommends -vce(robust) for dealing with suspected heteroskedasticity and within panel autocorrelation in the idiosyncratic error term.

[Carlo Lazzaro]

Anou: Example 3 under -xtreg- entry in Stata 13.1 .pdf manual, recommends -vce(robust) for dealing with suspected heteroskedasticity or within panel autocorrelation in the idiosyncratic error term. However, the subsequent Technical note states: "Clustering on the panel variable produces an estimator of the VCE that is robust to cross-sectional heteroskedasticity and within-panel (serial) correlation..."

[Nick Cox]

http://email.about.com/od/netiquette...e-Shouting.htm applies.

[Abhishek Shinde]

Check if this is useful:

http://fmwww.bc.edu/repec/bocode/x/xtscc_paper.pdf

xtreg cluster

[Anou shka]

So you mean to say vce(robust) controls for both heteroscedasticity and autocorrelation? that's enough?

[Anou shka]

Check if this is useful:

http://fmwww.bc.edu/repec/bocode/x/xtscc_paper.pdf

xtreg cluster

The table in that pdf is really useful!
Gonna follow it to derive my model

Thank you

[Jeroen vanDam]

Dear all,

I have more or less the same question. I read the article suggested in this post but I'm a bit confused which analysis to use in STATA to generate the right results. The analysis of my unbalanced panel dataset implies that the FE model has to be used, next to this both heteroskedasticity and autocorrelation are present. I did a lot of research on the internet and articles and different options show up on how to deal with this, I'm not sure which model is the most valid for this particular case. The options that I found and are also present in the suggested article are:

xtreg, fe robust - however, my results turn up to be non-significant when using this analysis
xtreg, fe vce(robust) - however, this option does not control for autocorrelation according to the article of Hoechle.
xtscc, fe

As far as I understand the xtscc, fe option turns out to be the best option. However, I'm not sure if the sample is cross-sectionally dependent. Can someone please explain me the differences between those options and which one turns out to be the best option in my case? Thank you in advance!


Kind regards,

Jeroen

[Carlo Lazzaro]

#8 is a duplicate post. Please, see: http://www.statalist.org/forums/foru...utocorrelation

[Syeda Afsheen]

Hey carlo
I have 234 companies with 11 years data (unbalanced) my problem is that I get fixed model via hausman, but my results are insignificant, given the fact that there is heteroscedasticity problem problem present as well. Could you please guide me

[Carlo Lazzaro]

Syeda:
welcome to the list.
What if you -cluster()- your standard errors?

[Syeda Afsheen]

Hello
I'm running a panel data regression, with 5 independent variables and 28 firms over 5 years. I have 140 observations.

After running a Hausman test, i found that a FE reg is to be used. Next I tested for heteroscedasticity - using the Cook-Weisberg httest for residuals - and autocorrelation - using the xtserial command for panel data. Both turned positive. My data is characterized by both heteroscedasticity and autocorrelation.

I then looked for ways to correct for them. I learned the following:
>> heteroscedasticity - use robust (eg. xtreg dep, var1, var2....., fe vce(robust))
>> autocorrelation - use Cochranne Orcutt method (prais dep, var1, var2...., corc)

But I need to correct them simultaneously in a single regression. I did found something using google :the Newey-west method...
I ran newey dep, var1, var2....., lag (1) force

My questions and problems are as follows:
> How to obtain r-squared when running newey?
> is there another more efficient way of correcting for both autocorrelation and heteroscedasticity?
> should i run dfuller to ensure stationarity? if yes, how to interpret dfuller please?

Any help would be much appreciated....

Anusha the command that you are looking for is

xtreg dep independ, fe vce(cluster id)

[Syeda Afsheen]

Thanks Carlo, I just saw your post. I did se both cluster and vce command to make my results more robust

[Carlo Lazzaro]

Syeda:
please note that -vce- (variance-covariance estimate) adds nothng to -cluster()-, as it is simply the way Stata uses to introduce different flavours of standard errors.