Geoeconomic dependence and supplier market power

Method

GeoDep stores one importer-HS6-year row with import dependence, concentration fields, a dependent flag, the dominant supplier in first_odpt and that supplier's share in share_odpt.

The country surface selects rows with dependent = 1, orders them by import_dpt and reports up to 30 critical products. It also reports pre-aggregated totals for critical products, high concentration, single sourcing, average import dependence and strategic-sector counts.

Supplier market power is represented by precomputed dominant, high-share and monopoly counts and average supplier share. The endpoint also lists the top 20 importer-product rows on which the requested ISO3 is first_odpt and the 15 importing economies with the most such rows.

ISO3 input is uppercased. CEPII GeoDep represents the EU27 as EUN, so requests for member-state ISO3 codes transparently read EUN and return a note that the response is an EU27 aggregate.

The default year is the literal 2022.

Equations

Validation

• Direct parquet inspection on 2026-08-27 found 2,945,881 importer-product-year rows in geodep, 800 importer-year summary rows and 743 supplier-year market-power rows. All three cover 2019-2022, so the route default of 2022 is within shipped coverage.
• BGD and USA each have a 2022 country-summary row, and CHN has a 2022 market-power row, validating the examples.
• The named sources document no external benchmark or reproduction exercise. Validation is limited to shipped-table coverage and agreement between route fields and parquet schemas.

Limitations

• The named sources do not document the formula or thresholds used to construct import_dpt, dependent, high_concentration, single_source or monopoly_count. None is inferred here.
• EU member-state requests return EUN aggregate evidence, not member-state-specific dependence. The response exposes source_iso3 and a note so this is visible.
• A product can carry several sector flags, but the country endpoint assigns the first matching flag as its primary display sector.
• The compare endpoint accepts at most eight comma-separated countries and silently truncates additional codes.
• The country endpoint joins products_all for product names, while the market-power endpoint joins the narrower products table. Recent-revision HS6 codes can therefore receive a fallback name on the market-power surface.
• The endpoints return HTTP 404 if their required GeoDep tables are unavailable. They also use a hardcoded default year of 2022.
• No causal interpretation is documented. The tables describe dependence and dominant-supplier exposure.
• GeoDep covers 2019-2022 and has no dedicated export in src/lib/vintage.ts. The registry field therefore carries the linked supply-chain surface's platform BACI label, not a GeoDep release date.

References

• CEPII GeoDep geoeconomic dependence database.

Endpoints