Declaring panel on bilateral trade data

[Maria Petersen]

Hi, I am relatively new to STATA and I have tried several hours to find this out on my own, unfortunately without success. I am working with a dataset on bilateral trade between countries in three different product sectors over the period 2016-2020. I have a heteroscedasticity problem after trying the BP-test, and I read that one way to solve this is to use fixed effects by declaring panel data. However, as I have several observations matching each other for example:

Exporter	Importer	Product	Year	Value	Weight
Denmark	Norway	Aluminium	2016	xxx eur	xxx kg
Denmark	Norway	Iron and Steel	2016	xxx eur	xxx kg
Denmark	Finland	Iron and steel	2016	xxx eur	xxx kg

except that it covers all of the EU countries in imports and exports from each other. However, there is only one observation for each combination of Exporter, Importer, Product and Year.

I get the error code

"repeated time values within panel"

when I try to use xtset. The only advice I've seen is to delete data, which would be counterproductive as it is data that I need..

Someone who knows how I can solve this? I want to do some form of regression analysis determining the effects of carbon pricing on the trade weights between the countries, so I will gladly accept all the help I can get.

[Maria Petersen]

I have also tried to include all these in the panel by using

xtset ExporterID ImporterID Year Resourcegroup

but I only get the error message : too many variables specified

[Maxence Morlet]

Use the egen group() function. You can only declare two variables in xtset, so in your case Maria Petersen I would do this:

egen unit_level=group(exporterid importerid resourcegroup)

xtset unit_level Year, yearly