Coded dataset and prototype disaggregation weights for "Inherited, not chosen: agricultural sector resolution in environmental input–output analysis"

Supporting data, analysis code, and a prototype weight set for the manuscript "Inherited, not chosen: agricultural sector resolution in environmental input–output analysis – a systematic review and a prototype set of disaggregation weights" (submitted to Economic Systems Research). The deposit contains three components. This is version v1 of the deposit. It adds flat CSV exports of the coded database; the weight set itself is unchanged and remains v0, as described in the manuscript. Version v0 of the deposit is archived at https://doi.org/10.5281/zenodo.21835671 1. Coded dataset. A relational database (papers.db) of the 155 studies coded for the review, of which 124 form the PRISMA sample. Each record carries bibliographic metadata and the coded facets, including the review's primary axis (agri_representation), the input–output database used, the environmental indicators reported, and the reporting-transparency items. 2. Analysis code. Scripts that rebuild the database from the coded pages and regenerate every table, figure and reported statistic in the manuscript, with pinned dependencies (requirements.txt, Python 3.10.12) and the order in which to run them. A frozen record of the inter-coder reliability run (kappa_run_output.md) reports Cohen's kappa for each coded axis together with the full disagreement table. 3. Prototype agricultural disaggregation weights (v0). A versioned concordance mapping 184 FAOSTAT commodities to fourteen EXIOBASE-aligned agricultural sub-sectors, country-level split weights for 131 countries derived from FAOSTAT gross production value (2022), and the pipeline that builds them and applies them to an aggregate agricultural sector. Coverage is 100% of reported production value, every weight vector sums to one, and the shipped file has been independently re-derived from the raw extract by a separate code path. Not included: the reviewed publications themselves, which are third-party copyright and are identified by DOI in the database; and the reliability answer key together with the second coder's raw sheet, withheld so that the coding exercise can be repeated blind by an independent coder. Intended use and limitations (v0 prototype) This deposit accompanies a systematic review. The disaggregation weights it releases are a v0 prototype – a first step toward a shared resource, not a finished one. Please read the following before applying them. Economic baseline, not emission allocation. The weights are derived from economic scale (FAOSTAT gross production value) and record the economic size of each sub-sector rather than its emission intensity. They must not be read as an emission allocation on their own. Proportional splitting assumes each sub-sector shares the parent sector's cost structure, which is the transparent baseline rather than the frontier. Temporal scope. Weights are static: a 2022 single-year build ships as the default, with a pooled 2020–2022 alternative supplied alongside and a per-country sensitivity analysis of the base-year choice. The deposit does not provide year-by-year weights and cannot represent structural change over time in longitudinal panel work. Inter-coder reliability. Cohen's kappa for each coded axis, the complete disagreement tables, and the spelling-normalisation log are in analysis/reliability/kappa_run_output.md. The answer key and the second coder's raw sheet are withheld so that an independent coder can repeat the exercise blind. Data formats. The review database ships as SQLite (papers.db) with flat CSV exports alongside it, including long-format tables for the multi-valued facets. The prototype weights ship in both wide and long CSV.