[Joao Santos Silva]

That is correct :-)

Joao

[Ainhoa Oses]

Perfect, Joao. Thanks a million for everything :-)

Ainhoa

[Ainhoa Oses]

Dear Joao,

Now that I have my PPML gravity model and the corresponding predictions ready (through -xtpoisson- and with country-pair fixed effects), I wanted to ask how it would be possible to get confidence intervals for my predictions. I saw in this thread https://www.statalist.org/forums/forum/general-stata-discussion/general/1307684-ppml-prediction-gravity that it might be difficult for some Stata versions, but my version is Stata 15.1 and so I wanted to know if there's a relatively simple way of getting those.

Many thanks,

Ainhoa

[Joao Santos Silva]

Dear Ainhoa,

Do you want to compute a CI for the prediction or a prediction interval. I do not think it is possible to compute reasonable prediction intervals and I do not see how CI for the prediction can be computed in a simple way.

Best wishes,

Joao

[John Alice]

Dear Mr Silva
I am sorry to bother you again!I have seen the form of ppml_panel_sg. on user-written before.But i do not know what the _panel_sg means in my model?In this paper, I use the ppml method to estimate the bilateral trade flows between countries based on the gravity model. Finally, based on the ppml estimation results, we measure the openness of China's service industry from the perspective of tariff equivalents and make international comparisons. Now I want to show you the questions asked are summarized as follows:
(1) I want to measure and compare the tariff equivalents of the service industries of the seven major service sectors of finance, telecommunications, tourism, and transportation services and so on in China and other countries. The choice of model can only be based on the same gravitational model?or Can i analysis based on multiple gravitational models?
(2) "Based on the results of the ppml estimation, the openness of China's service industry is measured from the perspective of tariff equivalents and international comparisons" Do you think this topic has research value?
(3) Is it necessary to do OLS regression before doing the empirical analysis of ppml? If the OLS regression parameter coefficient result is not significant, can I directly perform an empirical analysis of ppml?

[Joao Santos Silva]

Dear John Alice

I believe I answered your question here:

I am afraid I cannot comment on the particular research questions you are considering. I suggest you install pplm_panel_sg and check the help file. Finally, you do not need to use OLS; PPML is enough.

Best wishes,

Joao

[John Alice]

Dear Mr Silva
Thanks again for your reply!I may not be clear about the question i asked. My second question is actually asking, I may need to do seven estimates and regressions of the ppml model in the paper, the model in the empirical test must be the same model each time? For example, if i change the explanatory variable form of the base model or the explanatory variable to perform different ppml regression or estimation?

[Joao Santos Silva]

It depends on what you are doing; it may be OK to use different models but I see no reason for that.

Best wishes,

Joao

[Alexandre Asselin]

Dear Joao,

I am trying to measure the impact of diversity (measured by diversity in age and compensation among US firms) on innovation (spending in RD).

My hypothesis is that greater pay diversity and greater age diversity will increase the spending in RD.

I want to take into account the size of the firms (total assets, employees, EBIT) in order to cancel the effects of bigger firms having more spending in RD. Would using a command such as xtpqml seem appropriate to you? (I am not very familiar with Stata, so I hope to have provided you with enough information)

Best,

Alex

[Joao Santos Silva]

Dear Alexandre Asselin,

If you have panel data, then xtpqml (or xtpoisson) would be appropriate.

Best wishes,

Joao

[Alexandre Asselin]

Dear Joao,
Thanks a lot !

[Ainhoa Oses]

Dear Ainhoa,

Do you want to compute a CI for the prediction or a prediction interval. I do not think it is possible to compute reasonable prediction intervals and I do not see how CI for the prediction can be computed in a simple way.

Best wishes,

Joao

Dear Joao,

I'd like to get the error bands for my predictions. I just want to show a measure of uncertainty on my predictions and was wondering whether there was a relatively easy way to do so.

Thanks a million,

Ainhoa

[Joao Santos Silva]

Dear Ainhoa,

You should be able to get a confidence interval for the prediction, but not a prediction interval. That is, you can construct an interval the will contain the prediction with a certain frequency, but not an interval that will contain the observation you want to predict with that frequency.

Best wishes,

Joao

[John Alice]

Dear Mr Silva:
Sorry to bother you! Regarding the PPML model regression, I would like to ask you a few questions:
(1) Is there a requirement for sample observations using PPML regression? Is it generally necessary to have more than 3,000 sample observations?
(2) Add the time and country-pair fixed effects to the gravity model, and use the PPML regression command: ppml Y X1 X2 X3 .... i.countrypair i.year, cluster(countrypair) or PPML/Xtpossion Y X1 X2 X3, fe i(countrypair)？
Many thanks for your valuable help.
Alice

[Joao Santos Silva]

Dear John Alice

1) Sample requirements are as for any other estimator: you need large samples especially if you have many parameters, but there is no magic minimum number.

2) The second approach does not include the year dummies, but apart from that they are equivalent.

Best wishes,

Joao