Question on Cook's Distance with xtpcse regression

[Abe Percival]

I have been trying to calculate Cook's Distance for an xtpcse regression. The form of my regression is:

xtpcse depvar indvars, corr(ar1)

When I use "predict D, cooksd" I get a message that says that "option cooksd not allowed".

My question is twofold. Are there any ways for calculating Cook's Distance in Stata for xtpcse regressions (I get the impression that xtpcse has limited postestimation options for some reason)? If not, does anyone have a .do-file or a list of commands that I could use to manually compute Cook's Distance for an xtpcse regression?

Thanks.

[Rich Ganske]

I'd like to bump this question. I too have found that I need to use PCSEs because I have contemporaneously cross-sectional correlations in my panels (I'm looking monadically at economic recovery from wars). I have used fixed effects in the following code format:

xi: xtpcse dv iv ... i.ccode, corr(psar1) p

What would really make my life easy is if there were a -lvr2plot- available so I could examine my outlier cases, but also my high leverage cases as well. Alas, the postestimation command for xtpcse being what they are, I can get at the outliers by using the predict xb and stdp options.

But I'm all out of ideas of how to work around towards calculating leverage. Your suggestions are greatly appreciated.

Thanks, Rich