Hi Forum,

I am using a PPML model for my thesis which looks at export diversification (count dependent variable). I have one importer and 133 exporters. I am using exporter dummies and time dummies. I notice that the use of time dummies causes the GDPj and POPj variables to drop out (j is importer). Initially, only GDP was dropping and as this was my only time-variant control for the importer I also added Population to see how the model behave. It also drops the POPj!

I can't seem to figure out why this could be. Since, I am not interacting importer*year fixed effects here. I understand that this means these variables are perfectly collinear with the time dummies (i addition to the time-demeaning due to the fixed-effects).

What would you suggest to overcome this problem?

(Attachment)

Thank you in advance

Ollie Price