Seemingly unrelated pseudo poisson regression

[Akira Soto Nishimura]

Hello Statalist,

I am trying to do a seemngly unrelated pseudo poisson regression with fixed effects. As far as I can tell there seems to be no commands to do this. I read this paper https://repec.its.waikato.ac.nz/wai/econwp/1902.pdf where the authors did what I want to do. To quote the article,

"Finally, since we still need to consider the interrelationships between the two equations, we combine SUR and PPML together. To achieve this, we adopt the original two-stage SUR procedure. That is, we obtain the residuals from first-stage PPML models, and add each residual as an additional variable in the opposite PPML model in the second stage."

I know the authors did this in Stata but I do not know how to do it. Any advice on how would be much appreciated. I have been using the ppmlhdfe command and after the regression using the predict command but I get the error option residuals not allowed. I have an unbalanced panel dataset.

This is basically what I have tried

ppmlhdfe DV1 V1 V2 V3, absorb(year dyad)

predict newvar, residuals

[Akira Soto Nishimura]

Sergio Correia Any idea on how to get residuals after ppmlhdfe?

[Jia Gao]

Have you tried this package? https://sites.google.com/view/abbiezhang/stata-package. I have not tried it yet, so I do not know whether it works. Just scratching my head around this.