Benefit-principle CO2 accounting for the global textile and clothing value chain

Benefit-principle CO₂ accounting for the global textile and clothing value chain This repository provides the R code used to reproduce the benefit-principle CO₂ accounting analysis for the global textile and clothing value chain. The analysis uses input-output data from: Bhusal, L. B. (2025). Harmonized ADB Multi-Regional Input-Output (MRIO) Panel: Current Prices (2000-2024) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.17694852 The underlying input-output source is the Asian Development Bank Multi-Regional Input-Output database. CO₂ emissions are taken from the Asian Development Bank Environmentally Extended Multiregional Input-Output Tables for Air Emissions (https://kidb.adb.org/globalization/adb_environmentally_extended_multiregional_inputoutput_tables). The accounting framework The code implements a benefit-principle emissions framework for CO₂. The main accounting object is E_be = C_hat G^T F_hat L Y where Y is the bilateral final-demand matrix, L is the Leontief inverse, F_hat is the diagonal value-added coefficient matrix, G^T is the transposed Ghosh inverse, and C_hat is the diagonal CO₂ intensity matrix. Read from right to left, the equation traces final demand backward through supply chains, assigns value added to factor providers, carries that value added forward to emission sites, and converts sustained output into CO₂. The repository keeps four country positions visible in the same accounting path: E_ber,l(s,k) = C_hat^r (Gˢʳ)^T F_hat^s Lˢᵏ Yᵏˡ The indices are: Index Meaning r producer, where CO₂ is physically released l consumer, whose final demand starts the chain s factor provider, whose labor, capital, and other value-added components receive income k final producer, where the final good is completed before reaching the consumer This four-index structure separates the physical producer from the factor provider and the final producer. It is suited to textile and clothing supply chains where upstream inputs, assembly, final demand, and value capture can occur in different economies. Route decomposition The route decomposition partitions the final-producer index k: Route Condition Interpretation FIN k = r final-goods trade from the emission producer INT k = l intermediate inputs finalized in the consumer economy GVC k != r and k != l final production through a third economy The factor-provider index s remains unrestricted inside each route. Route labels describe how the final product reaches the consumer, while benefit-principle attribution assigns CO₂ to factor providers. Main metrics Metric Meaning phi(s,l) factor-provider share of consumer l's benefit-principle emissions attributed to provider s lambda(s,m) CO₂ per unit of factor-provider value added under route m rho(s,l,m) benefit-principle attribution to provider s divided by CBE physically occurring in s for consumer l and route m The repository also computes sectoral emitted-BPE and consumed-BPE summaries. Emitted BPE identifies sectors where CO₂ physically occurs. Consumed BPE identifies sectors whose final products generate demand. This distinction separates, for example, electricity-sector emissions from textile and clothing final demand. The selected-year check applies the same method to 2017, 2019, 2021, and 2023 from the same ADB MRIO panel. It is not a time-series analysis. It checks whether the core structural indicators persist under the same method and data family. What each script does Script Role Main outputs 00_core_functions.R Shared configuration, loaders, matrix builders, BPE functions, FIN-INT-GVC route functions, metric functions, and export helpers none by itself 01_prepare_data.R Loads the MRIO panel, checks dimensions, builds bilateral final demand, computes or reads input coefficients and inverse matrices, and saves year-specific caches outputs/cache/prepared_YYYY.rds, outputs/preparation_checks.csv 02_run_analysis.R Computes BPE, CBE, route components, phi, lambda, rho, sector summaries, economy-wide BPE/PBE, and selected-year checks outputs/BenefitPrinciple_TextileGVC_2023.xlsx, outputs/selected_year_check.csv, result RDS files Run order Run from the repository root: source("code/01_prepare_data.R") source("code/02_run_analysis.R") 00_core_functions.R is sourced by the two runner scripts and is not normally run alone. Data note The data are not bundled in this repository. Users must place the prepared MRIO panel RDS in the data/ folder or edit the input path in the configuration block. If CO₂ emissions are not already stored in the RDS, the code can read a separate emissions CSV with year, country, sector, and CO₂ quantity columns. Method sources The benefit-principle allocation follows the BPE logic of Chen et al. (2025), in which the Leontief and Ghosh inverses are used together in a static accounting framework. The FIN, INT, and GVC route classification follows the production-route logic of Meng et al. (2018), adapted here to the final-producer index inside the BPE equation. Reproducibility note The pipeline uses no random numbers. Results depend on the input MRIO panel, the CO₂ emissions vector, country codes, sector ordering, final-demand category ordering, and the selected years listed in the configuration.