Working Paper: NBER ID: w30890
Authors: Emmanuel Dhyne; Ayumu Ken Kikkawa; Magne Mogstad; Felix Tintelnot
Abstract: We use Belgian data on domestic firm-to-firm transactions and ask how the measurement of the share of imports in final consumption is affected when one uses data recorded at higher levels of aggregation. We find that aggregating detailed firm-to-firm transaction data to the firm level and imposing homogeneity assumptions in the composition of firms’ input and output do not substantially affect the measurement of the share of imports in final consumption. However, using the national IO tables alone may understate the share of imports in final consumption and, thereby, the gains from trade.
Keywords: No keywords provided
JEL Codes: F1
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| data aggregation (C43) | measurement of the share of imports in final consumption (E20) |
| national IO tables (D57) | understatement of the share of imports in final consumption (F62) |
| detailed firm-level data (L20) | higher median firm's total import share (F14) |
| national IO tables (D57) | lower share of imports in final consumption compared to baseline measurements (F62) |
| treatment of wholesale and retail firms in national IO tables (L81) | significant underestimation of total import shares (F14) |