[Karen Jyo]

Dear Maks,

Here are some answers:

1 - I assume your N is much larger than T, so do not worry about non-stationarity.

2 - Your -xtpoisson- regression is RE; you need to include the FE option and then the results should be the same.

3 - Just ignore the estimates based on the linear models as they are not reliable (and be very skeptical of RE estimators anyway).

4 - If you mean testing the significance of the difference, I think that there is a command that allows you to do it, but I cannot recall which one it is.

Best wishes,

Joao

Dear Joao,

Generally, it's not a matter when N is much larger than T.
But if the short panel only covers limited years and provinces, all the time-variant variables are stationary at first differences and cointegrated,
Can I perform PPML?

Best regards,

Karen

[Joao Santos Silva]

Dear Karen Jyo,

Can you please say more about your dataset?

Best wishes,

Joao

[Karen Jyo]

Dear @Joao Santos Silva

Thanks for your response.
Sorry, I gave the wrong message (countries rather provinces).
I am preparing a study about China's agricultural export (exp), which includes China's agricultural export value as the independent variable, and importer's GDP (lngdp) and population (lnpop) as time-variant variables.
distance, common border, common language, RTA and APEC are dummy variables.
My dataset includes 147 importers and 24 years (3528 observations).
The unit root test and cointegration test showed evidence of I(1) and cointegration relationship.
Hence, I want to know whether the dataset can be estimated by PPML.

Best regards,

Karen

[Joao Santos Silva]

Dear Karen Jyo,

Your dataset is rather small, but I would still use PPML.

Best wishes,

Joao

[Karen Jyo]

Dear @Joao Santos Silva,

Thanks very much for the response!

Same as you said, I can estimate the dataset by PPML.
Actually, I read some papers but no one clarifies the relationship between the cointegration test and the PPLM logically.
I hope to know the proof of their relationship because it seems that the cointegration test only shows the validity of OLS.
Hence, how can I explain the application of PPML on cointegrated panel data?

Best regards,

Karen

[Joao Santos Silva]

Dear Karen Jyo,

PPML is generally used in situations where the number of time periods is small relatively to the other dimensions of the data (e.g., number of countries) and therefore stationarity and co-integration are not relevant. However, in cases where T is the dominant dimension, you would have to worry about the time-series properties of the data, but I am not surprised if no one looked into that.

Best wishes,

Joao