Hello Statalist,

I am running the gravity model of trade at the product level.

So first I generated log of exports like this:

gen lnX = ln (X+1)

After that run the ppml with:

xi: ppml lnX ....

Is this approach wrong? Can I use log of dependent variable in PPML and add 1 before generating lnX. Because without 1, stata is giving "dependent variable lnX has negative values".

Or it`s better to run the model without log of dependent variable?