Effect of common-sourcing completion on comparative performance

Determine the direction and magnitude of the effect that detailed common-sourcing assumptions have on the comparative performance of completion methods for partially observed Japanese interregional input-output networks.

Background

The Japanese interregional input-output benchmark is constructed using common sourcing coefficients across purchasing industries, whereas the paper’s aggregated buyer groups do not necessarily satisfy that same restriction. The authors note that imposing detailed common sourcing may favor completion methods that embody related assumptions, but they do not establish whether this improves or worsens the reported comparisons, or by how much.

Resolving this issue would clarify whether the relative performance of proportional, gravity-based, collapsed-bilateral, and buyer-group completion methods reflects their informational content or instead partly reflects compatibility with the benchmark table’s construction procedure.