[Amlan Ray]

Hi Joao, in case of 3-way gravity with pair, importer-year and exporter-year fixed effects, will I have to remove GDP and Population from the model to avoid multi collinearity?

[Joao Santos Silva]

Yes, they drop automatically if you use ppmlhdfe.

[Amlan Ray]

Thank you Joao

[Emir Yilmaz]

Dear Joao,

I am working on a gravity model with just pairwise FE estimating the effect of economic support and containment stringency on intra eu trade flows during Covid-19. When I include time FE, the significance of standard gravity components vanishes. Would omitting time FE be a sound approach given the timeframe is just 2019-2021 (Monthly).

All the best

[Joao Santos Silva]

I understand form our email exchange that the best solution is to use exporter-quarter, importer-quarter and pairwise FE.