Good afternoon,

I am studying the determinants of asset trade flows across 40ish countries, from 1997 to 2015, using the gravity equation. Following Piermartini and Yotov's Estimating Trade Policy Effects with Structural Gravity (2016) paper, I wish to use the PPML estimator instead of OLS to deal with the various problems they highlight.


In Stata 14 I have tried the following code:

Code:
 ppml asset_flows ln_distw contig comlang_off common_leg eu i.o_country#i.year i.d_country#i.year
where ln_distw is the log of distance, comlang_off is a dummy for whether the two countries share the same official language, common_leg is a dummy for whether they share the same legal system, eu is a dummy equal to 1 if both countries are in the European Union, o_country is the exporter country and d_country is the importer country.

Stata gives back the error: factor variables and time-series operators not allowed
r(101);

If instead I try:
Code:
 ppmlhdfe ln_asset_flows ln_distw comlang_off eu common_leg contig, absorb(i.o_country#i.year i.d_country#i.year)
I get no error. I have three questions:
1) What is the error I get with ppml? I looked at other discussions but I can't seem to find a way to solve that applies to my case.
2) What is the difference between ppml and ppmlhdfe? It is my understanding there isn't ay from an econometric point of view.
3) If there isn't any difference between the two, how can I explain the fact that ppmlhdfe does not give me an error message?

Thank you for your help,

Matteo