An attempt to create a consistent international trade dataset for the course STA2101.
I implemented a gravity-type model to remove transportation costs and mismatches by imputing around 10% of the exports data for the period 2002-2012, and also removed flows with no attributed origin/destination.
The idea of imputing some rows is because I applied a criteria of replacing the observed values with data from the model if and only if that reduced the mismatching between trading partners.
These tables discount re-imports and re-exports besides harmonizing mismatching flows.
This is similar to BACI: International Trade Database at the Product-Level. The 1994-2007 Version.