[Joao Santos Silva]

Dear Kumuthini,

I am not aware of any better method, so I would stick to ppml. Maybe you want to try different sets of regressors?

Best wishes,

Joao

[Kumuthini Sivathas]

Yes. I understand Prof. Joao, Thanks so much.


To show the robustness of PPML, is that fine to include time in-varying variables namely ln_Distance, ln_CommonLanguage, ln_ColonialConnections etc to the model?

Kindly advise me.

Thanks & regards
Kumuthini

[Kumuthini Sivathas]

Dear Prof. Joao,


In addition to the above question, i have the following as well.


1. Can i rely on R^2 or adjusted R^2 test to discuss the explanatory power of the PPML (Its very high in my PPML results) compared to OLS?
2. Having included destination fixed effects in to the model, Would it make sense to use RESET test ?

Kindly advise me.


Thanks you

Kind regards
Kumuthini

[Joao Santos Silva]

Dear Kumuthini,

The R2 is rarely interesting, so I would not put much weight on it. The RESET is always potentially interesting, but the more variables you have in the model the more demanding the test becomes.

Best wishes,

Joao

[Kumuthini Sivathas]

Noted with many thanks Prof. Joao.

Yes, I found the RESET test to fail. But I will continue with it.

Thanks and Regards
Kumuthini

[Kumuthini Sivathas]

Dear Prof. Joao,


Greetings to you!!


If you could please see this output. Usually the fixed effects used to be eliminated from the regression. But in this case (where I gave a condition to choose some set of countries), ln_GDP and TFA_EX(a dummy variable) also had been excluded.

This is not due to collinearity. What’s the implication of this?

Could you please help me with this.



. xi: ppml EXPORTS ln_GDP_SL ln_GDP ln_PC_GDP_SL ln_PC_GDP ln_POP_SL ln_POP ln_TARIFF TFA_EX i.year i.countr
> y_id if country_group_id == 777, cluster (country_id)


i.year _Iyear_1995-2016 (naturally coded; _Iyear_1995 omitted)
i.country_id _Icountry_i_2-77 (naturally coded; _Icountry_i_2 omitted)

note: checking the existence of the estimates

Number of regressors excluded to ensure that the estimates exist: 73
Excluded regressors: ln_GDP TFA_EX _Iyear_1996 _Iyear_1997 _Iyear_2016 _Icountry_i_3 _Icountry_i_4 _Icountr
> y_i_6 _Icountry_i_7 _Icountry_i_9 _Icountry_i_10 _Icountry_i_11 _Icountry_i_12 _Icountry_i_13 _Icountry_i_
> 14 _Icountry_i_15 _Icountry_i_16 _Icountry_i_17 _Icountry_i_18 _Icountry_i_19 _Icountry_i_20 _Icountry_i_2
> 1 _Icountry_i_22 _Icountry_i_23 _Icountry_i_24 _Icountry_i_25 _Icountry_i_26 _Icountry_i_27 _Icountry_i_28
> _Icountry_i_29 _Icountry_i_31 _Icountry_i_32 _Icountry_i_33 _Icountry_i_34 _Icountry_i_35 _Icountry_i_36
> _Icountry_i_37 _Icountry_i_38 _Icountry_i_39 _Icountry_i_40 _Icountry_i_41 _Icountry_i_42 _Icountry_i_43 _
> Icountry_i_44 _Icountry_i_45 _Icountry_i_46 _Icountry_i_47 _Icountry_i_48 _Icountry_i_50 _Icountry_i_51 _I
> country_i_52 _Icountry_i_53 _Icountry_i_55 _Icountry_i_56 _Icountry_i_57 _Icountry_i_58 _Icountry_i_59 _Ic
> ountry_i_60 _Icountry_i_61 _Icountry_i_62 _Icountry_i_63 _Icountry_i_64 _Icountry_i_65 _Icountry_i_66 _Ico
> untry_i_68 _Icountry_i_69 _Icountry_i_71 _Icountry_i_72 _Icountry_i_73 _Icountry_i_74 _Icountry_i_75 _Icou
> ntry_i_76 _Icountry_i_77
Number of observations excluded: 1

note: ln_PC_GDP_SL omitted because of collinearity
note: ln_POP_SL omitted because of collinearity

note: starting ppml estimation
note: EXPORTS has noninteger values

Iteration 1: deviance = 303.7585
Iteration 2: deviance = 120.8608
Iteration 3: deviance = 78.63341
Iteration 4: deviance = 70.3778
Iteration 5: deviance = 68.32475
Iteration 6: deviance = 67.65021
Iteration 7: deviance = 67.45487
Iteration 8: deviance = 67.41638
Iteration 9: deviance = 67.41336
Iteration 10: deviance = 67.41332
Iteration 11: deviance = 67.41332

Number of parameters: 27
Number of observations: 77
Pseudo log-likelihood: -202.09714
R-squared: .96627972
Option strict is: off
(Std. Err. adjusted for 5 clusters in country_id)
--------------------------------------------------------------------------------
| Robust
EXPORTS | Coef. Std. Err. z P>|z| [95% Conf. Interval]
---------------+----------------------------------------------------------------
ln_GDP_SL | .0694801 .7926377 0.09 0.930 -1.484061 1.623021
ln_PC_GDP | 1.04572 .6330145 1.65 0.099 -.194966 2.286405
ln_POP | -4.016381 1.262029 -3.18 0.001 -6.489913 -1.54285
ln_TARIFF | -3.707038 1.076158 -3.44 0.001 -5.816269 -1.597806
_Iyear_1998 | -.3853567 .3100148 -1.24 0.214 -.9929745 .2222611
_Iyear_1999 | -.7047279 .1627425 -4.33 0.000 -1.023697 -.3857585
_Iyear_2000 | -.7003651 .1289156 -5.43 0.000 -.9530351 -.4476952
_Iyear_2001 | -.9255743 .2604073 -3.55 0.000 -1.435963 -.4151854
_Iyear_2002 | -.7101523 .2278252 -3.12 0.002 -1.156682 -.2636231
_Iyear_2003 | -.5953068 .2443264 -2.44 0.015 -1.074178 -.1164358
_Iyear_2004 | -.5887271 .2540867 -2.32 0.021 -1.086728 -.0907262
_Iyear_2005 | -.6612038 .3013202 -2.19 0.028 -1.251781 -.070627
_Iyear_2006 | -.557996 .3110963 -1.79 0.073 -1.167734 .0517416
_Iyear_2007 | -.5958093 .219775 -2.71 0.007 -1.02656 -.1650582
_Iyear_2008 | -.5362667 .12927 -4.15 0.000 -.7896312 -.2829022
_Iyear_2009 | -.6122251 .1878175 -3.26 0.001 -.9803407 -.2441095
_Iyear_2010 | -.5844723 .1267693 -4.61 0.000 -.8329355 -.3360091
_Iyear_2011 | -.4372145 .0750644 -5.82 0.000 -.5843379 -.2900911
_Iyear_2012 | -.2709049 .1979359 -1.37 0.171 -.6588522 .1170424
_Iyear_2013 | -.3033954 .1022903 -2.97 0.003 -.5038808 -.1029101
_Iyear_2014 | -.1874146 .0247396 -7.58 0.000 -.2359034 -.1389258
_Iyear_2015 | -.0521911 .0132413 -3.94 0.000 -.0781436 -.0262386
_Icountry_i_5 | 25.08654 7.002227 3.58 0.000 11.36243 38.81065
_Icountry_i_8 | -4.736253 .795196 -5.96 0.000 -6.294808 -3.177697
_Icountry_i_49 | 15.76431 4.789388 3.29 0.001 6.377281 25.15134
_Icountry_i_54 | 26.22793 7.212084 3.64 0.000 12.09251 40.36336
_cons | -1.341223 8.82448 -0.15 0.879 -18.63689 15.95444
--------------------------------------------------------------------------------


Thank you very much.

Kind regards
Kumuthini

[Joao Santos Silva]

Dear Kumuthini Sivathas,

Themain problem here is that your sample is not big enough to estimate this model.

Best wishes,

Joao

[Kumuthini Sivathas]

Yes, Prof. Joao. I understand that.

Thanks so much.


Kind regards,
Kumuthini

[Bashir Muhammad]

Dear Professor, when we take log of Bilateral FDI then log drop the 0 values from Bilateral FDI.
Through which problems log drop the 0 values and what is the good solution for this.
Thanks!
Bashir Muhammad

[Bashir Muhammad]

Dear @Joao Santos Silva,
when we take log of Bilateral FDI then log drop the 0 values from Bilateral FDI.
Through which problems log drop the 0 values and what is the good solution for this.
Thanks!

Bashir Muhammad

[Joao Santos Silva]

Do not take logs and use PPML instead.

Best wishes,

Joao

[Bashir Muhammad]

but if we want to take log so it will create some problem in results so which test can we use on this which remove this problem from results.
Thanks

[Bashir Muhammad]

Dear @Joao Santos Silva, I understand.

Thanks so much

Bashir

[Joao Santos Silva]

Excellent; thanks for letting me know.

Best wishes,

Joao