SUR or Fixed Effects Reg. unbalanced panel

[Drobi]

Hi everyone,

I would like to run a SUR regression model or Fixed effects regression model on my unbalanced panel data.
SUR:
log(gini_net)=c(1)+c(2)*log(gdp_p_c_usd_ppp)^2+c(3)*log(gdp_p_c_usd_ppp)+c(4)*trade_openness+c(6)*inflation+c(7)*unemployment+c(8)*fc_total
log(gini_market)=c(9)+c(10)*log(gdp_p_c_usd_ppp)^2+c(11)*log(gdp_p_c_usd_ppp)+c(12)*trade_openness+c(14)*inflation+c(15)*unemployment
FE:
LS(?,CX=F,PER=F) LOG(GINI_NET) C LOG(GDP_P_C_USD_PPP) (LOG(GDP_P_C_USD_PPP))^2 TRADE_OPENNESS UNEMPLOYMENT INFLATION FC_TOTAL
For the fixed effects regression I employed time fixed and country fixed effects. I would like to use Driscoll-Kraay standard errors to account for cross sectional and temporal dependence. How do I best account for heteroskedastic and autocorrelation (poss. two lags) in the FE model?

Second, the SUR and FE model should give (ideally) similar results however I suspect it in its current version, does not account for time or country fixed effects nor forms of heteroskedastic and autocorrelation.

Thank you for any tips.

[EViews Gareth]

I'm not sure what your question is.

[Drobi]

Sorry that it was a little unclear.
The results of both regressions are somewhat disappointing. I wonder, how to best control for heteroskedastic and autocorrelation in the FE model. Using White correlation coefficient matrix?

Second, the SUR results in completely different results, which I suspect might be due to the missing control for country and time fixed effects. How can I include that in the SUR model is there an option I can tick within Eviews to include these dummies?

Thank you

[EViews Gareth]

EViews does not offer built in fixed effects for system estimation, including SUR, no.

[Drobi]

Could I include @expand(country ,@dropfirst) in my SUR system to create cross section dummies and something similar for period effect? If yes, would it look like in the attachment? How could I do similar dummies for period effects.

thanks for the fast replies :D

[EViews Gareth]

You cannot. You'll have to add them manually.

[Drobi]

Hi,

I now created a group of dummy variables for each country. However, I have difficulties including the group into the SUR model above. Or is it necessary to split up the group into separate series and include those manually into the SUR model. If so, how can I efficiently split up a group of dummies variables into series where each column represents one series?

Thank you for your help