In our fourth model, we
simultaneously introduce country-and-time fixed effects (by generating a full set of exporter-and-time and importer-and-time dummy
variables) and country-pair fixed effects to correct the bias induced by unobserved time-varying multilateral resistance terms.