[John Alice]

Dear Joao
Thank you very much for your answer. The missing values of the bilateral service sector between the two countries I encountered during the research process refers to the unreported trade values in the OECD Tisp. The unreported trade values can be understood as the trade volume between country A and country B is relatively too small so it is not reported? Can a zero value be used even if there is a large range of unreported trade values?
Thanks in advance,
Alice

[Joao Santos Silva]

Dear John Alice,

As I said before, I do not know the data well enough to comment, but if the missing values are small it should be reasonable to replace them with zeros.

Best wishes,

Joao

[Manmeet Ajmani]

Dear Prof. @Joao Santos Silva,

As per suggestion, I used the following commands for RESET test:

qui ppml DefTradeMill `cties' contig comlang_off comcol col45 ln_distw _fe*, cluster(en_ctypair)
 
where DefTradeMill: Deflated trade in Million USD, `cties’ = country-pair dummy and _fe* = exporter and importer fixed effects
 
*RESET test
predict XB,xb
qui su XB
gen XB2 = (XB-r(mean))^2
ppml DefTradeMill `cties' contig comlang_off comcol col45 ln_distw XB2 _fe*, cluster(en_ctypair)
test XB2 = 0

The results are as follows:

Number of parameters: 4289
Number of observations: 186378
Pseudo log-likelihood: -3432757.5
R-squared: .86834016
Option strict is: off
                               (Std. Err. adjusted for 10,776 clusters in en_ctypair)
-------------------------------------------------------------------------------------
                    |               Robust
       DefTradeMill |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
            BRN_KHM |  -3.796096   .8692616    -4.37   0.000    -5.499818   -2.092375
            BRN_IDN |   -2.15949   .5372618    -4.02   0.000    -3.212504   -1.106477
            BRN_LAO |  -1.550168    .904569    -1.71   0.087    -3.323091    .2227544
            BRN_MYS |    1.38816   .5760659     2.41   0.016     .2590914    2.517228
            BRN_MMR |  -3.604317   .8420464    -4.28   0.000    -5.254697   -1.953936
            BRN_PHL |  -2.131688   .5197988    -4.10   0.000    -3.150475   -1.112901
            BRN_SGP |   .2513248   .5761639     0.44   0.663    -.8779357    1.380585
            BRN_THA |  -.1077163   .5159846    -0.21   0.835    -1.119028    .9035949
….*dummies removed for brevity
 
test XB2 = 0
 
 ( 1)  XB2 = 0
 
           chi2(  1) =    5.76
         Prob > chi2 =    0.0164

Can I say that my model specification is reliable at 1 percent?

If yes, then can you please guide me how to interpret my result, which is as follows:
If BRN_PHL = -2.13, then Brunei (BRN) is under-exporting to Philippines (PHL) by (e(-2.13) – 1) = -0.88 million USD. Is my interpretation correct? This is in absolute terms and I will consider the country pair trade in order to have results in relative terms.

If my interpretation is not correct, then how will you interpret it?

Looking forward to your reply.

Regards,
Manmeet Ajmani

[CAO DUC SON]

thank you, Professor Joao Santos Silva.
May I ask you a favor? ( my data abut 25 exporters, 197 importer, year:2001-2019)
-Can I use the Ramsey test to select the appreciated model? (3 fixed effects Time-Exporter; Time-Importer and Time-Exporter-Importer estimated by PPML)
-Can I use the coefficient to estimate the potential trade for an individual country to its importer?

The topic: https://www.statalist.org/forums/for...tobit-and-pplm

[Joao Santos Silva]

Dear CAO DUC SON

I suggest you choose the model based on economic theory rather than the RESET test. Yes, people use these models to estimate trade potentials.

Best wishes,

Joao

[Sofiene Mahjoubi]

Dear Isabel,

The warning is displayed when regressors have values larger than log of one million (in absolute value) so you can get summary statistics of the GDP variables and decide on the appropriate scale. Anyway, if you get convergence (which you do) you can ignore the warning.

Best wishes,

Joao

Sir, how to create time-invariant origin fixed effects and time-varying destination fixed effects. My data is 24 neighbors and 3 years (1994, 2004 and 2014

[Joao Santos Silva]

Dear Sofiene Mahjoubi,

Apologies for the late reply. Please use the user-contributed ppmlhdfe command and check the help file to see how this can be done.

Best wishes,

Joao

[Sofiene Mahjoubi]

Dear Joao Santos Silva, thanks for your answer, but i need to estimate spatial panel data with PPML method of gravity model but in stata now command is available. Just ppmlhdfe, ppml and ppml_panel_sg. my question, how to estimate spatial panel data in stata of gravity model with ppml method??

[Joao Santos Silva]

Dear Sofiene Mahjoubi,

I am not sure if I understand your question. All gravity equations are "spatial" in the sense that they are models for flows of something (goods, services, people, etc.) between regions of space. If you mean "spatial" in the sense of spatial econometrics where you account for spatial auto-correlation, I am not even aware of any theoretical results on that, let alone a Stata command.

Best wishes,

Joao

[Sofiene Mahjoubi]

Dear Joao Santos Silva, my data is spatial of 24 x 23 country (migration of country i to country j) but i don't konw what's the command to estimate this data?? ppml, ppmlhdfe, ppml_panel_sg or ppml_fe_bias??

[Joao Santos Silva]

I would start with ppml or ppmlhdfe.

[Sofiene Mahjoubi]

Dear Joao Santos Silva, what's the command use in this paper? https://sci-hub.st/10.1016/j.gloenvcha.2016.04.014

[Jared Greathouse]

Sofiene Mahjoubi You likely seek the ppml command.

[Sofiene Mahjoubi]

Dear Jared Greathouse, ok, with ssc install ppml in stata?

[Sofiene Mahjoubi]

Joao Santos Silva, the result of R² = 0.428 (reghdfe) but Pseudo R² is 0.021 (ppmlhdfe), this result is correctly or not?