Industry Chain Map

Problem

IO-table work often stalls on sector granularity, matrix algebra, graph path search, and plotting. @user_053d27d0/industry-chain-map targets a 15-sector structure aggregated from the Chinese NBS 42-sector table for 2020, and exposes upstream/downstream dependency, key sector detection, supply-chain path tracing, multipliers, import reliance, clustering, and shock simulation as a Python API.

How It Works

• IOTable wraps the intermediate flow matrix Z, direct consumption matrix A, Leontief inverse, and Ghosh inverse, with total output, value added rates, and final demand.
• upstream_dependency() and downstream_influence() compute backward/forward linkage to identify sectors that pull upstream suppliers or support downstream demand.
• key_sectors() ranks sectors by linkage strength; supply_chain_path() uses BFS to trace the shortest transmission path between two sectors with intermediate flows and coefficients.
• value_multiplier() measures the output pull of a one-unit final demand increase; impact_analysis() simulates a sector shock using ΔX = (I-A)^(-1) * ΔF.
• import_dependency() estimates import reliance from intermediate input share and HHI concentration; structural_holes() uses betweenness centrality to find bridge sectors; cluster_sectors() groups sectors by IO similarity.
• Results are returned as pandas.DataFrame, with color mappings in config/sectors.yaml for external plotting in matplotlib, plotly, or ECharts.

Boundaries

Best for static, script-based analysis of China's industrial structure and supply-chain risk. It is not a replacement for import-level detail, employment estimates, time-series analysis, or a GUI app. The data are embedded 2020 approximations at a fixed 15-sector level; load_io_table(year=...) is only a label, and changing years requires replacing matrices in io_table.py. Domestic and imported intermediate inputs are not separated, so import reliance is a proxy estimate.