Technological Change and Labor Income Share in China: Evidence from Skill Complementarity and Industry-Level Transmission
Description
This repository contains the data and code required to reproduce the empirical results in "Technological Change and Labor Income Share in China: Evidence from Skill Complementarity and Industry-Level Transmission". Study overview. The paper examines how industry-level total factor productivity (TFP) affects China’s aggregate and industry-level labor income share over 2000–2014, distinguishing direct effects, supplier and customer linkage effects, final-demand effects, and between-industry reallocation. The analysis uses WIOD-based Chinese industry panel data (42 industries, 15 years), distributed-lag weighted least squares regressions, IV robustness checks, and heterogeneity analysis by period and sector. Contents of the replication package: - 1.code.ipynb: data cleaning, price-index rebasing, construction of solow_tfp; Figures 1–2; Tables 1–3. - 2.code.ipynb: industrial linkage weights; construction of supplier/customer TFP and U.S. TFP; baseline and IV regressions; robustness and heterogeneity analysis; Figures 3–4; Tables 4–10. - dataset_01/: raw input data (WIOD IO tables, China industry accounts, capital price indices, U.S. industry panel data) and analysis/regression output files. - README.txt: step-by-step replication instructions mapping each table and figure to notebook cells. - tables in the paper.xlsx and figures in the paper/: final tables and figures as reported in the manuscript. Data sources. Primary data are from the World Input–Output Database (WIOD), 2016 release (2000–2014). Supplementary materials include fixed-asset investment price indices and U.S. industry panel data for robustness. All data are publicly available; sources are documented in the paper. How to reproduce. Run 1.code.ipynb and then 2.code.ipynb in order, following the cell titles listed in README.txt. Main analysis file: chn_data_panelv2.xlsx. Software: Jupyter Notebook(Python) Interpretation. Industry TFP is measured as a Solow residual. Supplier and customer TFP are constructed from normalized input–output linkage weights (own-industry flows excluded and renormalized). U.S. TFP is computed using the growth-accounting approach. Regression coefficients report the cumulative effect of a one-standard-deviation TFP shock over lags 0–4. See the paper for variable definitions and table–figure correspondence.