Replication Package for "Resource Productivity and Material Import Dependency: A Joint Decomposition across Organisation for Economic Co-operation and Development Economies"

This repository contains the replication package for the manuscript “Resource Productivity and Material Import Dependency: A Joint Decomposition across Organisation for Economic Co-operation and Development Economies.” The study examines the empirical accounting relationship between resource productivity (RP) and material import dependency (MID) across 38 OECD economies. The primary analysis uses a balanced country-year panel covering 2000–2023, with an additional available-through-2024 sensitivity analysis. The study decomposes changes in RP and MID to distinguish same-flow accounting covariance, cross-flow covariance, non-shared covariance, and material within/composition components. The replication package contains the archived harmonised analytical extract used in the study, Python analysis and figure-generation scripts, machine-readable table and figure source data, validation outputs, metadata, and documentation required to reproduce the reported analyses. The workflow reproduces the main and supplementary analyses, including: one-, two-, three-, and five-year RP–MID co-movement estimates; the annual 4 × 2 RP-flow × MID-flow covariance decomposition; same-flow, cross-flow, and non-shared covariance contributions; the material within/composition decomposition; country-cluster bootstrap inference; alternative exact allocation conventions and Shapley feasibility checks; population-, GDP-, DMI-, and DMC-weighted sensitivity analyses; leave-one-material-group-out sensitivity analyses; trade-throughput and cross-country heterogeneity diagnostics; multi-year and alternative inference checks; supplementary LMDI context; rolling-origin forecasting diagnostics; and the available-through-2024 sensitivity analysis. The package reproduces the reported analyses from the archived harmonised OECD/World Bank analytical extract. It is not intended to reproduce data acquisition from live OECD or World Bank APIs. Third-party source data remain subject to the applicable terms and conditions of the originating data providers. The primary balanced panel contains 38 economies and 912 country-year observations for 2000–2023. The annual primary regression uses 874 annual changes, while the category-complete sample used for the four-material-group composition decomposition contains 870 annual changes. These sample definitions correspond to those reported in the manuscript. To reproduce the analysis, download and extract the package and run: python run_all.py from the package root directory. Full instructions, software requirements, output mappings, data provenance, and validation information are provided in the accompanying README.md and validation files.