Measuring trade in value added: how valid is the proportionality assumption?

For countries that have only aggregate (‘competitive type’) input–output (IO) tables, value added in exports is commonly estimated using the ‘proportionality assumption’ to separate imported-inputs from domestically procured inputs. We test the validity of this assumption using non-competitive type IO tables, which contain separately compiled domestic- and imported-input matrices, for Indonesia, Thailand, Malaysia, Taiwan, and Australia. The results show that the proportionality assumption leads to an overestimation of domestic value-added in exports, and that the magnitude of the bias becomes amplified when the export composition of a country shifts from primary products to manufactured goods through integration into global production networks.