Examining the Trade Potential of the UAE using a Gravity Model and a PPML Estimator

[Salim Ghazvini Kor]

We are estimating a PPML model for the UAE using a total of 32 exporting countries and about 200 importers. However, from the results that we obtain we are only interested to discuss the UAE's predicted values of exports (to estimate the Trade Potential using the conventional method) and determinants. If I understand this correctly: Despite the fact that the scope of the research is centered on the UAE, we must still include many exporters in our econometric estimations, because if we do not, by restricting the sample to the bilateral exports and imports of the UAE with the rest of the world, then we would miss out on trade diversion effects and we would not be able to take into account the multilateral resistances in the model. Is this correct? If so, is this adequate justification in order to discuss results for the UAE obtained with a sample from 32 exporting countries?

[Tom Zylkin]

Hi Salim:
If I understand your concerns correctly (you want to account for multilateral resistances), then yes you want to have more than one exporter and more than one importer.
Regards,
Tom

[Salim Ghazvini Kor]

Thank you for your reply professor Zylkin. So my question was that :

1. basically any conclusions we draw about the UAE which is one of the exporting countries in our data set, we can also draw for the remaining exporting countries in our data as well? So that we could concentrate in a second paper, with the same results, on another one of the countries?

And if I may ask a few more short questions:
2. Does the ppml_panel_sg command allow us to retain the residuals so that we can conduct the Reset test?
3. PPML and ppml_panel_sg can handle high level data. We have exporters, importers and aggregated (1 digit) products and we employ exporter-time FE, importer-time FE as well as country pair FE. The proper way to do this for more dissagregated data (let's say 6-digit data), we would first need to define the panel (i) for each category of exporter-importer-sector (i.e. by creating a variable through the command cap egen categi = group(exporter importer category) and then include both country-pair Fixed Effects and exporter-sector-time and importer-sector-time Fixed Effects? Is that correct?
4. We are using 2-3 and 4 year data intervals for our econometric estimations. Is this practice necessary only to account for trade policy changes or is this necessary in every estimation of PPML gravity models?


Thank you again and sorry for not finding clear answers to those questions in the literature.