Gravity model on trade war

[Siphelele Ndzoyiya]

Hi everyone, I am measuring the effects of the trade war between the US and China and their impact on Saudi Arabia's trading volume with its trading partners. I have collected all the required data, and here are my variables:

ln(Tij)=α+β1ln(GDPi)+β2ln(GDPj)+β3ln(Distanceij)+β4ExchangeRatej+β5TradeWarDummy+β6(TradeWarDummy×ln(GDPi))+β7(TradeWarDummy×ln(GDPj))+β8ln(Popi)+β9ln(Popj)+β10RegionalTradeAgreementij+β11Borderij+β12PandemicDummy+β13(PandemicDummy×ln(GDPi))+β1 4(PandemicDummy×ln(GDPj))+ϵij

So I am new to Stata and using the Gravity model. Many studies suggested I should use ppml. So, I do not know where to begin or how I should start to get better results. I want to measure before the trade war, during the trade war(2018-2023), and during the trade war in the presence of COVID-19 (2020-2022). And after Covid-19.

You can suggest anything, which stata commands and variables that could help me.

[Joao Santos Silva]

Dear Siphelele Ndzoyiya,

I suggest you use the user-contributed command ppmlhdfe to estimate via PPML. Check the help file for examples.

Best wishes,

Joao

[Siphelele Ndzoyiya]

Okay Thank you, Mr Silva. where can I get the help file?

[Joao Santos Silva]

Type

ssc install ppmlhdfe

and then

help ppmlhdfe

[Siphelele Ndzoyiya]

Okay here are my results, what can you say about these results. Some variables are significant and others are insignificant but some are not. What can you suggest?

[Joao Santos Silva]

Dear Siphelele Ndzoyiya,

My suggestion is that you start by reading about the method you are using and about gravity equations. For example, you will find that when you estimate by PPML, the dependent variable is not in logs. Also, your sample is very small.

Best wishes,

Joao

[Siphelele Ndzoyiya]

Okay, thank you very much, sir. Can you please tell me which are the key things I should know about ppml and gravity model, how do I do a diagnosis test using ppml, and how do I know if my model is fit. I do have other variables such as population, which other variables would you recommend so I can increase my sample size?

[Siphelele Ndzoyiya]

@Joao Santos Silva

Please look at my results below. What can you say or suggest?

[Joao Santos Silva]

Dear Siphelele Ndzoyiya,

It is difficult to comment without knowing much more about what you are doing, but I note that you are not using any fixed effects, which is unusual. Also, when you estimate using sub-samples, you really do not have enough observations.

Best wishes,

Joao