What can you find on this page?
This page is mainly based on Mejean and Rousseaux (2024), the third chapter of Paris Report 2: Europe’s Economic Security, published by CEPR and Bruegel in May 2024 and edited by Jean Pisani-Ferry, Beatrice Weder di Mauro and Jeromin Zettelmeyer. It also draws on the extension of this work published in the European Investment Bank’s Investment Report 2024/2025 (Chapter 5, Box C).
The purpose is to explain how imported trade vulnerabilities can be identified statistically and to document what the resulting diagnostic says about the European Union: which products are vulnerable, how vulnerabilities evolve over time, where they come from, which sectors they belong to, and why some matter much more than others.
Once vulnerabilities have been identified, the statistical list is crossed with four non-exhaustive risk dimensions developed in Mejean and Rousseaux (2024): geopolitical exposure; position in global value chains and the risk of downstream propagation; shortages of critical goods, including pharmaceutical inputs; and exposure in technologies that matter for future European competitiveness, especially the green transition. The final section discusses the policy instruments available to strengthen resilience.
Why should we pay attention to EU imported trade dependencies?
Global value chains coordinate production across firms, sectors and countries, from upstream inputs to final consumption. Their international fragmentation has generated large gains from specialisation and can provide risk-sharing against country-specific shocks. At the same time, production can become concentrated at particular nodes of the network. When those nodes are disrupted, shocks may propagate through downstream firms and industries (Backus et al., 1992; Antràs and Chor, 2021; di Giovanni et al., 2020; Bonadio et al., 2021; Boehm et al., 2019).
The Covid-19 pandemic and Russia’s invasion of Ukraine made these exposures particularly visible, including for pharmaceuticals, energy and other critical inputs. The policy problem is therefore not to eliminate international specialisation, but to identify the relatively small set of exposures for which concentration is combined with limited domestic capacity and limited possibilities for substitution. This is the efficiency–resilience trade-off at the heart of Mejean and Rousseaux (2024).
The same measurement framework is relevant in periods of trade tension. It can be used to identify EU reliance on particular foreign suppliers and, in reverse, foreign reliance on European supply. These measures should nevertheless be interpreted as exposure indicators rather than direct measures of geopolitical leverage: inventories, alternative technologies, supplier switching and the position of a product in the value chain all affect the consequences of a disruption.
How many trade vulnerabilities does the EU import?
Starting from the roughly 5,381 HS6 products imported by the consolidated EU27 over 2015–2019, the European Commission's three criteria identify 378 vulnerable products (its "bottom-up" list). Adding the domestic-absorption criterion — restricting to products for which extra-EU imports cover more than half of EU absorption — narrows this to 228 products. Finally, keeping only those with very low ex-post substitutability between suppliers leaves 49 strategic dependencies: about 0.5% of total EU27 import value, concentrated in energy, mining, basic metals and chemicals, and mostly sourced from China.
Live replication: the funnel, from the underlying data
How has this evolved since 2002?
Reproducing the diagnostic year by year from 2002 to 2024 shows a clear before/after break around the 2008 financial crisis, followed by a durable plateau — regardless of which methodology is used. Production data (needed for the absorption and full five-criteria lists) lags the trade data by a year or two, so the most recent year(s) on the chart can show the EC list only until Prodcom catches up.
Number of vulnerable products, by year and methodology
Pooled by period
Vulnerabilities are persistent. Using the full five-criteria definition, 41% of the dependencies identified before the global financial crisis remain in the immediately following period, and 35% remain in the subsequent recovery period. The Covid period is associated with a further increase in the number of products identified by the European Commission criteria.
Sector, geography and concentration
Everything below — the sector chart, the map of top suppliers, the upstreamness × HHI scatter, the distribution plots, and the product explorer at the bottom of the page — reacts to the same period toggle. Pick a period first, the charts redraw immediately below it.
Sectoral distribution
Top countries of origin
Position in the value chain × import concentration
Ex-post substitutability ("stickiness")
Upstreamness
Domestic absorption
Explore the underlying product list
Search, sort and filter the website replication directly — still split by tier and, now, by period. Showing the top 20 matches only; for the full list, email pierre.rousseaux@ensae.fr.
| HS6 | Product description | Tier | HHI ▾ | Upstreamness | Top supplier |
|---|
Four normative risk dimensions
Not every statistically vulnerable product carries the same consequence for the European economy. Mejean and Rousseaux (2024) cross the vulnerability list with four non-exhaustive normative arguments.
Between 2015 and 2019, 70–80% of identified trade vulnerabilities were primarily sourced from non-NATO countries. Linking these dependencies to the Geopolitical Risk Index (Caldara and Iacoviello, 2022) — built from systematic tracking of geopolitical events (the Gulf War, 9/11, Crimea 2014) in major US newspapers, across 23 non-NATO countries — shows that between 4 and 18 strategic dependencies sourced from non-NATO countries exceed the median geopolitical-risk level.
Examples: quebracho extract from Argentina (96% share, HHI 0.92), lithium carbonate from Chile (85%, HHI 0.73), iodine from Chile (73%, HHI 0.56), vulcanised-rubber gloves from Malaysia (63%, HHI 0.42), and refractory clay from Ukraine (55%). China's share of EU import vulnerabilities is rising, while the shares of the United States and the rest of the world have fallen by 3 to 10 percentage points — and China's share is especially high among products the EU cannot easily produce itself.
Shocks can propagate downstream through supply chains (di Giovanni et al., 2020; Bonadio et al., 2021; Boehm et al., 2019), which matters most when they threaten the competitiveness of domestic firms further down the chain (Baur and Flach, 2022). Over 50% of the EU's trade vulnerabilities vis-à-vis China involve final consumption goods — but the more strategically relevant question is how far upstream a vulnerable product sits.
Linking dependencies to the upstreamness metric of Antràs et al. (2012) identifies 22 products more than three production stages away from final consumers — concentrated in energy, mining, basic metals and chemicals, and predominantly imported from China. Examples: acyclic hydrocarbons from Russia (70% share, upstreamness 4.20), trichloroethylene from the US (90%, upstreamness 3.85), quebracho extract from Argentina (96%, upstreamness 3.85), raw beryllium from the US (62%, upstreamness 3.42), and magnesium products from China (91%, upstreamness 3.40). On average across periods, highly upstream products make up around 49% of full-criteria vulnerabilities.
Shortages of critical goods carry consequences well beyond economic cost, including in the worst cases human costs — and the private-versus-societal tradeoff between low-probability risk and productive efficiency justifies public intervention (Baldwin and Freeman, 2021). The pharmaceutical sector is the clearest illustration: drug shortages in the US have increased tenfold since the 2000s, closely linked to the offshoring of active-ingredient manufacturing to emerging economies.
Active pharmaceutical ingredients (APIs): 169 product categories, only 2 under the full five-criteria filter
Two APIs survive the full five-criteria framework: daprodustat (chronic kidney disease anaemia) and metharbital (epilepsy), both sourced >97% from India with an HHI of 0.94. The EC criteria identify 159 of these API product categories; 136 remain after adding the absorption criterion — including pregabalin (epilepsy, anxiety), atorvastatin-type cholesterol treatments, and bimatoprost (glaucoma) — mostly sourced from Singapore and the United States at market shares often above 70%.
The baseline exercise is backward-looking because it identifies vulnerabilities from observed trade and production patterns. Mejean and Rousseaux therefore complement it with a forward-looking discussion of technologies that matter for the green transition. None of the products in the 2015–2019 five-criteria list is directly identified as an input into the selected green technologies, but several of their upstream materials are traded in highly concentrated world markets. This exercise is best read as a prospective risk screen rather than as an extension of the five-criteria vulnerability list.
- Lithium-ion batteries: lithium carbonate exports are concentrated in Chile (57%) and Argentina (16%), HHI 0.367; cobalt ore and concentrate exports are 88% from the Democratic Republic of Congo, HHI 0.778. Batteries themselves are less concentrated (HHI 0.209; China 41%, Korea 15%), with diversification opening up via Canada and technologies such as Northvolt's NMC 811.
- Solar panels: moderate overall concentration (HHI 0.216; China 43%, Malaysia 10%) masks heavier Chinese dominance in specific components such as mounted piezo-electric crystals, diodes and transistors.
- Hydrogen and rare earths: the hydrogen market is comparatively competitive (Netherlands 38%, Canada 33%, HHI 0.264); rare earths are more concentrated (Vietnam 33%, China 29%, HHI 0.234), with China's dominance of global mining reserves compounding the risk — nearly half of natural graphite powder exports, for instance.