Adding robust option to fixed effects regression? Good or bad for statistical outcome ?

[Nico Peters]

Dear econometricians,
I just have a small question as i'm conducting at the moment a data analysis which tries to test effects of immigrants on GDP per capita.
my first regression is ln(gdppercapitagrowth)=ß0+ß1ln(populationgrowth)+ß 2ln(laborforcegrowth)+ß3ln(investmentgrowth)+ß4ln( inflowgrowth)
in which the inflow growth variable represents the growth rate of immigrant inflows.For the first regression i want to test that immigration inflows lead to lower capital per worker and hence decrease GDP per capita.

In the second regression i split up the inflow of immigrants by their level of educational attainment lowskilledgrowth & highskillegrowth and exclude the inflowgrowth variable (the rest stay's the same). With the second regression i want to test that highskilled immigrants have a larger postive impact on Gdp per capita then low skilled immigrants.

I conducted already the Hausmann test and came to the result to use fixed effects. But now i saw that it is even possible to add the robust option to an fixed effects model. It gaves me the following outputs :

So my question is if it is a good choice to add the robust option as for the first to outputs it leads to a even better outcome in terms of significance. But for the other 2 i.e. the ones with low/high skilled growth it make the outcome worse. Would highly appreciate it if somebody could quickly approve my methodology or give me some comments or suggestions.

Thanks for your time

Nico

[Carlo Lazzaro]

Nico:
statistical significance is not the gold standard here.
Under -xtreg- robust or clustered SEs (they do the same yob, unlike with -regress-) take heteroskedasticity and/or autocorrelation into account: that's all.
That said, it does not seem to me that in your case a wide difference exists with default or robust SEs.

[Nico Peters]

Thanks Carlo for the fast response,
just for clarification it is already sufficient to just use the xtreg var ..., fe option without adding further robustness option right ?

[Carlo Lazzaro]

Nico:
if heteroskedasticity and/or autocorrelation are not issues with your dataset, go default SE.