Replication package: Trade shocks redraw the global carbon map but do not shrink it

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:Provides a rigorous empirical framework and replication data for studying trade-carbon linkages.

🏢実務担当者:Highlights how trade policy changes can affect supply-chain carbon footprints, relevant for Scope 3 reporting.

🏛政策担当者:Informs CBAM and trade policy design by showing potential carbon leakage effects.

Replication package for Trade shocks redraw the global carbon map but do not shrink it: product-level evidence from integration and fragmentation. The package rebuilds every number, table and figure in the paper from the bilateral embodied-carbon panel it is built on: 22.4 million flows at HS2 × country-pair × year, 231 exporters, 1995–2023, constructed by mapping cradle-to-gate life-cycle intensities onto BACI trade and audited in a companion measurement paper. Contents: data_r2d/ — the bilateral embodied-carbon panel ( edges_hs2_year.parquet , 22.4 M rows) and the gravity and capacity layers the counterfactuals need. This is the audited R2D basis; basis_a_superseded/ holds the earlier basis and is included only so the correction is auditable. code/ — the analysis scripts behind the published results: the within-exporter event studies, the Callaway–Sant'Anna designs, the gravity and CBAM counterfactuals, the capacitated transport re-solve, and the figure system. code/fonts/ vendors the Latin Modern faces (GUST Font License) so the figures render identically without a TeX installation. code/review_2026_08/ — the analyses added over three pre-submission review rounds, with their own README: randomization inference where a single treated unit invalidates clustering, relative-magnitudes sensitivity to pre-treatment trends, retention of displaced carbon against a counterfactual growth path, the elasticity re-estimated on the products the counterfactual prices, propagation of the life-cycle factors' uncertainty through 200 complete redraws, and the entrepôt and absorber-control tests. Two scripts are kept as negative records, marked as such, because they produce results the paper does not use and the reason matters: a raw retention ratio that credits trend growth as reappearance, and an HS6 retention regression whose slope is set by a handful of extreme cells. results_r2d/ — the published figures. provenance/ — the audit trail: which script produces which number, what was corrected and when, the figure system's design decisions, and the weakness audit that drove the last round of work. Every table in the manuscript is emitted by a script in this package from a logged run, so no reported number is transcribed by hand. Trade data are BACI (CEPII), which requires free registration and is therefore not redistributed here; life-cycle factors come from ecoinvent and Agribalyse through the ADEME Base Empreinte and are likewise not redistributed. The Global Sanctions Data Base requires a non-commercial request to its maintainers. Everything derived from those sources that the paper reports is included.

• openalex https://doi.org/10.5281/zenodo.21889662first seen 2026-08-13 05:03:44