Summing observations with bilateral trade data

[Leo Picard]

Dear Statalist community,
I'm working on bilateral trade openness between countries, and I am searching for a way to sum observations from country a to b with b to a. Here is an example of the dataset I am dealing with:

iso3_o    iso3_d    year    import_openness
DEU    FRA    1978    .0306583
DEU    FRA    1979    .0312403
FRA    DEU    1978    .0185003
FRA    DEU    1979    .020072

Here,I would like to sum the import openness from Germany to France and from France to Germany for each year. I was wondering if there was a way to compute this for a large dataset (400k+ observations). I've succeeded using the command -reshape- accordingly, but it does not work for the complete dataset as there is too many values.
The goal is to generate a new variable that looks like this:

With each element being the import openness between country i and j at time t (import flow divided by the importer GDP).

Thank you in advance !

[Nick Cox]

I don't that reshape is either needed or even helpful here. The sum of whatever over e.g. exporters is

egen double total_i = total(whatever), by(iso3_o year) 

and that over importers is similar. Those appear to be your denominator terms.

For handling dyads some tricks are explained within https://www.stata-journal.com/articl...article=dm0043

[Leo Picard]

Dear Nick,
Thank you for your swift answer. I found what I needed using your link, and it works like a charm.

generate first = cond(iso3_o < iso3_d, iso3_o, iso3_d)
generate second = cond(iso3_o > iso3_d, iso3_o, iso3_d)

bysort first second year: egen bil_open = sum(import_openness)
drop import_openness first second

Thanks again for your support !

Cheers
LP

[Ahmed Adib]

Dear Nick,
Thank you for your swift answer. I found what I needed using your link, and it works like a charm.

generate first = cond(iso3_o < iso3_d, iso3_o, iso3_d)
generate second = cond(iso3_o > iso3_d, iso3_o, iso3_d)

bysort first second year: egen bil_open = sum(import_openness)
drop import_openness first second

Thanks again for your support !

Cheers
LP

This has been immensely helpful! Thank you so much!