Dear Joao Santos Silva

I have a panel data set with a count dependent variable and several count regressors, all which have a large number of 0 values. I ran the PPML command to find 3 warnings for 3 different regressors about the size of the values, but none were excluded from the estimation that followed.

In your note, http://personal.lse.ac.uk/tenreyro/poisson.pdf, you consider situations where y has too large values and give some suggestions, but what if it is the independent variables that have too large values, what is the solution then?

I have looked on the ppml help and it says:
" ppml also issues a warning if the dependent variable or any of the regressors has large values that are likely to cause numerical problems and lead to convergence difficulties. Once the subset of regressors and observations to be used is defined, ppml uses the [R] glm command to perform the estimation."

Does this mean it solves the issue by implementing glm itself?

Regards,
Anthony