FedProcCarbon v1.0：米国連邦政府物品調達の内包GHG排出量、FY2015-FY2024、NAICS-6月次分解

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

🔬研究者:Provides a high-resolution dataset for analyzing government procurement emissions and testing Scope 3 methodologies.

🏢実務担当者:Offers a template for building procurement-based emission inventories, useful for companies with government contracts.

🏛政策担当者:Highlights the feasibility of tracking embodied emissions in public spending, informing sustainable procurement policies.

Monthly obligated contract spend and embodied greenhouse-gas emissions for 616 goods-producing NAICS-6 commodities over 120 United States federal fiscal months (FY2015-FY2024), covering USD 2,766.8 billion of obligations and 486.9 MtCO2e of embodied emissions, with an agency-resolved annual companion table spanning 23 awarding agencies. The dataset joins two public sources that have not previously been released together as a commodity-resolved time series: contract obligations from the Federal Procurement Data System via the USAspending.gov API, and the U.S. EPA Supply Chain Greenhouse Gas Emission Factors v1.4.0. The join is exact rather than probabilistic, because every federal contract action carries a NAICS-6 code assigned by the contracting officer and NAICS-6 is the emission factor table's own key: 616 of 616 commodities (100%) match a factor. TECHNICAL VALIDATION. Annual totals reconstructed from an independent USAspending endpoint and aggregation path agree with the panel to a maximum relative discrepancy of 0.0000% in all ten fiscal years. A random sample of 25 commodities (3,000 monthly values) reproduced byte-identically on re-query. The panel is a complete 616 x 120 matrix with zero null and zero non-finite entries; monthly cells aggregate to annual cells with a difference of USD 0.000000. First-digit frequencies of the 57,893 positive obligation cells are consistent with Benford's law (chi-square = 10.1 on 8 degrees of freedom). The agency table captures 99.8-99.9% of panel obligations in every year. FILES. fedproccarbon_monthly_naics6.csv (73,920 rows, primary panel); fedproccarbon_agency_naics_fy.csv (46,984 rows); commodity_dimension.csv (616 commodities, v1.4.0 and superseded v1.2 factors, USEEIO reference sector); annual_summary.csv; naics_2022_to_2017_crosswalk.csv (1,150 pairs with ambiguity flags). KNOWN LIMITATIONS, stated in the README. Obligations are nominal while factors are per 2024 USD, so cross-year real-activity claims require deflation first. Obligations are financial commitments rather than deliveries. Emission factors are sector averages and cannot resolve individual suppliers, so the dataset must not be used for supplier benchmarking. EPA's 1,016 published NAICS-6 factors contain only 260 distinct values because they derive from a 392-sector USEEIO model, so cross-commodity statistical inference should cluster on the supplied useeio_reference_code column.

• Zenodo https://zenodo.org/records/22037222first seen 2026-08-21 04:11:58 · last seen 2026-09-04 04:36:44