[Richa Khurana]

Dear Sir,

Regarding zero mean. There are many other variables which have been dropped but their mean is not equal to zero. Why would that be ?

Best regards
Richa

[Joao Santos Silva]

Richa,

That may be due to collinearity or for the estimator to exist. Please familiarize yourself with the papers in the list of references, OK?

Best wishes,

Joao

[Richa Khurana]

Dear Sir,

Okay.. Thank you so much for your guidance!

Best Regards
Richa

[Richa Khurana]

Hello Sir
I am facing some problems with the estimation again. Something went wrong with assigning pair_ids earlier and therefore, wrong estimates were generated. But now trying the same estimation with ppml, is not reporting the standard errors (when I am estimating with importer time, exporter time and country pair fixed effects along with my variables of interest which are also dummies). There are no singletons in my data. Trying with options like strict, noconstant and strict and noconstant together is also not working for me. It gives the warning "variance matrix is nonsymmetric and highly singular". Can you please suggest some other way to go about it.
Many thanks

Best regards
Richa

[Richa Khurana]

Also my panel is unbalanced. Is balanced panel required for this to work ?

[Joao Santos Silva]

Richa,

Unbalanced panels are not a problem. With the information you provide, I suppose that the problem is related to high collinearity in your data. Do you really need the country-pair fixed effects? Anyway, there are other things that you can do, for example run the regression with sub-sets of the regressors to see if you can identify the variable causing the problem. Rescaling you data may also help.

Best wishes,

Joao

[Richa Khurana]

Dear Sir,
Yes Sir the collinearity is very high and I was trying to follow the literature which says the MRTs are controlled for properly with country time fixed effects and country- pair fixed effects. Sir please could you elaborate on "running the regression with sub-sets of regressors" . Can I email you my dataset Sir? Will it be possible for you to have a look at it and identify the problem?

Thank you
Best Regards
Richa