PSCE model with FIXED effects

[Der Lancken]

Hello everyone,
I'm trying to run a PSCE model with fixed effects as Im trying to recreate a regression from this article, https://www.jstor.org/stable/24695992, where they use this model. I've tried running this code,

> xtreg DC l.DC REALI M2 GFCF GDP_GR ln_GDP_PC INEQ DEREG PORTF i.Years, fe vce(cluster countryID) which aims to do the same thing but obviously this isnt a PSCE model. So I have been told to use xtpcse but this command on this own doesnt work with a standard, fe. So I think i need to manually need to add dummies and now I don't know how to go forward.
And when i run my xtpcse it just gives me this error code, even before i add my dummies to account for fixed effects: no time periods are common to all panels, cannot estimate disturbance
covariance matrix using casewise inclusion

This is my xtreg code that "works"

eststo: xtreg DC l.DC REALI M2 GFCF GDP_GR ln_GDP_PC INEQ DEREG PORTF i.Years, fe vce(cluster countryID)

this is my xtpcse that doesnt work right now

eststo: xtpcse DC l.DC i.Years REALI M2 GFCF GDP_GR ln_GDP_PC INEQ DEREG PORTF


So im looking for your input, help would be greatly appreciated since this is apart of my bachelors thesis ha ha ha

[Maxence Morlet]

If my memory serves, Beck and Katz (1995) pcse standard errors are appropriate if T>N, which happens seldomly. Is this your case?

[Der Lancken]

If my memory serves, Beck and Katz (1995) pcse standard errors are appropriate if T>N, which happens seldomly. Is this your case?

No you're right its not. I will try to estimate a GMM sys modell to deal with the problems that the PCSE will have, but I'd first like to see what the PCSE gives me. But maybe that is a waste of time...

[Der Lancken]

If my memory serves, Beck and Katz (1995) pcse standard errors are appropriate if T>N, which happens seldomly. Is this your case?

My last post is incorrect, T is actually > N, i have 20 countries and 37 years.

[Maxence Morlet]

I do not know pcse that well, but again, if my memory serves, in terms of type 1 errors, in this case with T>N, Driscoll & Kraay (1998) standard errors may outperform Beck and Katz (1995) standard errors. They can be implemented using the community contributed command xtscc.