Can I run a Pseudo-Poisson Maximum Likelihood in panel data when not using a Gravity Model? What is the STATA command?

[Mulenga Chonzi]

I want to run the Pseudo-Poisson Maximum Likelihood (PPML) in a panel data framework as my dependent variable has many zeroes. However, my challenge is that from all the literature I have read on the PPML, it seems to only work in gravity model type of estimation.
1. Is it possible to run a PPML using panel data for a non-gravity type of model?
2. If it is possible, what is the STATA command to use?

[Joao Santos Silva]

Dear Mulenga Chonzi ,

Indeed you can use PPML other kinds of data and there are many examples of that. With panel data, consider using xtpoisson and the user-written ppmlhdfe.

Best wishes,

Joao

[Mulenga Chonzi]

Thank you very much for your response. It was very helpful

[Mulenga Chonzi]

I would like to do a post estimation check for the goodness of fit after running my PPML estimation. Iam using STATA 13. However when i type the command estat gof, Iam getting an error of "invalid subcommand gof'. Does anyone know another command I can use or if there is something Iam not doing right?

[Jeff Wooldridge]

The PPML estimator was being using for nonnegative responses long before it was being used for gravity models. The xtpoisson command with the vce(robust) option works fine. Be sure to include time fixed effects.

It's not easy to get meaningful goodness of fit measures for nonlinear panel data models. And there's not much need. You don't have much choice when it comes to models and estimation, anyway. At least none that are as robust and simple.

JW

[Qiyangfan Feng]

Dear all,

I am new for stata....It's my first time to contact ppml.and i got some questions recently.

Is ppml applicable to the other aspects rather than gravity model? Can it be used if i want to measrue the effect of FDI flows to economic growth even if there are no zero? Should i log the dependent var(economic growth) or keep them original?