ppml with a large number of fixed effects

[David Larson]

I am trying to estimate a gravity equation (cross section) with a large number of fixed effects. For OLS I use
xi: areg depvar indepvar, absorb(fixed_effect)
Is there an equivalent to "absorb" for ppml? The number of fixed effects is very large it will not work with i.fixed_effect.

Best,
D

[Joao Santos Silva]

Try xtpoisson with the fe option.

Best wishes,

Joao

[David Larson]

Dear Joao,
Thank you for the reply.
I just wanted to clarify one additional detail. I am an applied person and I am not familiar with the technical details of your RESTAT paper. If I use xtpoisson command, should I write that I have used PPML or Poisson?

[Joao Santos Silva]

Hi David,

If you use robust (or clustered) standard errors as you should, you can say that you used PPML (it does not hurt to add a footnote saying that you implemented it by using the xtpoisson command with the FE option).

Best wishes,

Joao

[David Larson]

Thank you very much!

[Bashir Muhammad]

Dear Professor,
If lambda_hat is significant in gravity model such as regression and fixed effect so what is the solution for this problem. Is there any test or any other solution for this problem.

Thanks so much

[Joao Santos Silva]

Sorry, what is lambda_hat?

Joao

[Bashir Muhammad]

First i obtain 0,...,0,lambda_hat_it,0,...0: from the following commands.

forvalues i = 2002 2003 to 2012 {
quietly probit dummy z1 if year==`i'
quietly predict xb_hat_`i', xb
quietly gen lambda_hat_`i' = normalden(xb_hat_`i')/normal(xb_hat_`i') if dummy==1
quietly gen lambda_hat_d_`i' = lambda_hat_`i'
quietly replace lambda_hat_d_`i' = 0 if year!=`i'
quietly drop xb_hat_`i' lambda_hat_`i'
}

After this all i run the gravity model such as regression and fixed effect. for Wooldridge, J. M. (1995) sample selection but the probit lambda_hat is negative significant.

so what is the solution for this problem. Is there any test or any other solution for this problem.

Thanks so much

[Joao Santos Silva]

Dear Bashir Muhammad

My advice is that you should not estimate gravity equations using a sample selection model, just use ppml.

Best wishes,

Joao

[Bashir Muhammad]

Dear Joao Santos Silva,
I know but the problem is that my observation are more and cant use ppml because not work

[Joao Santos Silva]

Sample size should not be a problem for ppml, it the problem is that you have too many fixed effects, try xtpoisson or ppml_panel_sg.

Best wishes,

Joao

[Bashir Muhammad]

Ok thanks so much