In early May, I discovered an error in the export data set used in Tables 5-8; for some observations (I think it’s values of cty2<cty1) gdp1, gdp2, gdppc1 and gdppc2 are incorrect (gdp1/gdppc1 should be gdp2/gdppc2 and vice versa). This turns out to be of no consequence for literally any result in the paper (I’ve re-run everything … if you’re interested, it’s because of the symmetric constraints in Tables 7&8 combined with exporter/importer-time FE which obviate use of output [except to delineate samples]). However, it may be consequential if one used this data set for a reason other than replication. Since I provide my data sets for the purposes of replication, I’m not that fussed. But … be warned!