China-HQ green FDI readiness / Method appendix
Back to the story

Competition in third markets

Purpose. Methodology note for the two third-markets competition figures on the station page [1]. The figures adapt Figure 1 of the Kiel Institute report on China's manufacturing surge [2] to clean-tech scope: per Solar and Battery manufacturing product, they plot the change in China's share of third-market imports against the change in the tech-specific incumbent bloc's share, between a 2012-2014 pre-window and a 2022-2024 post-window. The upper-left quadrant, where China gained share while the incumbents lost it, is the declining-incumbent story zone.

Spine framing

The station argues that China-HQ green FDI is a revealed-capability signal. This layer supplies the competitive backdrop that makes the signal consequential. The 2010 clean-tech leaders were Germany, Japan, and the United States in solar and Japan, Korea, and the United States in batteries. The figures show those incumbents losing the export competition for neutral markets across most of the value chain, and China taking the share they shed. The outward FDI documented elsewhere in the station is the channel through which that lead now propagates into new manufacturing entrants.

Data sources

Bilateral HS6 trade is read from two BACI 202601 releases (CEPII): the post-window 2022-2024 from BACI_HS17_V202601.zip (HS2017 nomenclature, which already backs the station's trade-outcome layer) and the pre-window 2012-2014 from BACI_HS12_V202601.zip (HS2012 nomenclature, added for the 2012 baseline). The Solar and Battery manufacturing baskets are the same HS6 sets used throughout the station (data/processed/hs6_baskets.json, manufacturing cut, which excludes raw materials). Supply-chain stage labels come from the green dictionary [3]. The build is scripts/build_third_market_shares.py; the figure generator is scripts/make_third_market_competition.py. Both raw BACI zips are kept out of version control.

Third-market definition

For each technology, third markets exclude, as destinations, China plus that technology's incumbent bloc:

Excluding these destinations isolates the contested neutral markets (emerging economies and the rest of the world) and removes two confounds: home-market sales and trade internal to the incumbent bloc. The incumbent bloc is excluded only as a buyer. It is retained as a seller, so the figure measures the bloc's combined supplier share among third-market imports. This mirrors the Kiel construction, which excludes the EU+7 bloc as a destination but plots its aggregate exporter share.

Share computation

For each product and year, an exporter's third-market share is its value of exports to third-country destinations divided by all exporters' third-market imports of that product. Shares are averaged over the years in each window that carry positive trade, matching the three-year-mean convention used elsewhere in the station. The reported change is the post-window mean share minus the pre-window mean share, in percentage points, computed for China and for the incumbent bloc as an aggregate, with each bloc member also reported individually for the tooltip. Dots are colored by supply-chain stage and sized by post-window third-market trade value.

Concordance

Scatter points are keyed on the live HS2017 code present in the post-window. The pre-window value is read under that code's HS2012 antecedent. For stable six-digit lines, which are the large majority, the antecedent is the identical code, confirmed present in the HS2012 data. Three codes were split in the HS2017 amendment and require a curated antecedent: HS 382499 reads under HS 382490 (electrolyte and other chemical products), HS 390761 reads under HS 390760 (PET solar film), and HS 285390 reads under HS 285300 (NCM precursor). These three antecedents are broader HS2012 parents than their HS2017 child, so the pre-window value mixes in some trade that later moved to sibling codes; those points carry a broad-parent flag in the figure tooltip and in the data. The resolved mapping is written to sources/hs17_hs12_concordance.csv.

Basket codes that exist only in the HS2022 nomenclature have no HS2017 post-window data and drop out: the photovoltaic-cell split HS 854142, 854143, and 854149, the photovoltaic-generator codes HS 850171, 850172, and 850180, and the furnace code HS 851432 for solar; and three machine-tool and intermediate codes for battery. They are reported in data/processed/_third_market_qa.md, not plotted. A publish gate fails the build if any high-value live code resolves to no pre-window trade, which is the signature of a missing concordance entry; the gate passes for the current basket.

Findings

Most clean-tech manufacturing products sit in the China-gains, incumbents-lose quadrant: 24 of 34 plotted solar products and 37 of 54 battery products. The median solar product saw China gain 6.3 points of third-market share and the incumbent bloc lose 6.5 points; the median battery product saw China gain 9.6 points and the bloc lose 4.0 points.

The two finished-product anchors carry the clearest signal. In solar cells and modules (HS 854140), China's third-market share rose from 39.1 percent to 64.0 percent while the Germany-Japan-US bloc fell from 26.9 percent to 11.9 percent, with Germany and Japan each shedding about 6.5 points. In lithium-ion cells (HS 850760), China rose from 43.2 percent to 54.1 percent while the Japan-Korea-US bloc collapsed from 32.3 percent to 9.7 percent, with Japan, the original commercial lithium-ion producer, losing 11.7 points and Korea 8.8 points. The aggregate value-weighted picture is consistent: China's third-market value over the post-window equals 88 percent of its all-destination solar manufacturing exports and 71 percent of its battery manufacturing exports, the gap reflecting the excluded bloc destinations.

Caveats

These are descriptive share shifts, not causal attribution. The figures show where China gained and incumbents lost, not why, and the FDI layer is not invoked as the cause of the trade shift.

The solar finished-product proxy HS 854140 bundles photovoltaic cells and modules with light-emitting diodes in HS2017 nomenclature. The bundling is stable across HS2012 and HS2017, so the pre-to-post comparison is internally consistent, but the level should be read as a module proxy rather than a clean module count. The HS2022 split into HS 854142 and HS 854143 would separate the two; it is unavailable for the post-window without abandoning the consistent nomenclature.

Several large dots are multi-use commodities rather than tech-specific products: unwrought aluminium (HS 760120) in the solar basket, alumina (HS 281820) in both baskets, and ammonia (HS 281410) in the solar basket. Their share shifts reflect general industrial and petrochemical trade, not solar-specific or battery-specific competition. They are colored as processed materials and cluster near the origin, where neither China nor the bloc moved much. This is the same multi-use-commodity limit documented for the supply-chain spectrum layer [4].

The three broad-parent concordance points carry an inflated and noisier pre-window baseline. The clearest case is the NCM precursor (HS 285390), whose HS2012 parent HS 285300 is a catch-all inorganic-chemicals heading; its pre-window shares mix in unrelated trade and its position should be read with the broad-parent flag in mind. Products trading under 50 million dollars in the post-window are drawn faint and dashed because a small denominator makes their share volatile.

The window is deliberately long, 2012-2014 to 2022-2024, to capture the full arc of incumbent decline rather than a recent slice. This is a different design choice from the station's two-year FDI-outcome windows and is the right one for a structural-competition question.

References

Markdown source on GitHub