Unveiling the hidden costs of critical dependencies
1 Recent trends in global trade
Trade integration, a key driver of economic growth in recent decades, is at a crossroads amid growing geopolitical rivalries and protectionist policies. By liberalising cross-border flows of goods, services, investment and technology, the multilateral rules-based trading system boosted growth and reduced global inequalities (Frankel and Romer, 1999; Feyrer, 2019). A corollary of this deeper globalisation has been the increasing sensitivity of domestic output to external shocks (Chart 1), particularly in the 1990s and the 2000s – a period marked by the proliferation of multilateral trade agreements (e.g. the North American Free Trade Agreement in 1992) and the establishment of the World Trade Organization in 1995, which China joined in 2001. While greater trade openness has enabled countries to cushion domestic shocks through greater reliance on foreign demand (International Relations Committee, 2023), it has also heightened their exposure to global economic disturbances. Recent global shocks – such as the COVID-19 pandemic and Russia’s unjustified war against Ukraine – have brought this duality into sharp focus. As a result, trade integration is viewed increasingly through the lens of vulnerability and strategic risk, prompting a shift towards more inward-looking economic policies by many governments. This shift, exemplified by events such as Brexit (2016), the US-China trade tensions (2018-19) and the recent US tariff hikes, represents a paradigm change in global trade relations, characterised by growing scepticism towards trade liberalisation and increasing strain on the open global trading system.
Chart 1
Impact of a 1% shock to foreign total factor productivity on output growth in the first year
(percentage points)

Sources: Boeckelmann et al. (2025), World Input-Output Database, Asian Development Bank and ECB staff calculations.
Notes: Impulse response functions (IRFs) are computed using a dynamic multi-country, multi-sector model featuring sectoral linkages through production networks for intermediate and capital goods. IRFs for the different time periods are derived by calibrating the model with the corresponding vintages of global input-output matrices. Panel b): only shocks outside the euro area are considered.
Governments are reorienting their trade strategies, shifting the focus from liberalisation and efficiency towards security and resilience. There is a growing focus on reducing reliance on foreign suppliers, especially for critical raw materials, strategic technologies and other critical goods – areas where countries with a dominant position in terms of their supply may deliberately restrict access to harm dependent countries. Recent events have underscored this shift. The Organisation for Economic Co-operation and Development (OECD, 2025) highlights a fivefold rise in the number of vital inputs (e.g. lithium, cobalt) under export restrictions since 2009, a trend which reflects the growing demand for these inputs due to the digital and green transitions (e.g. an electric vehicle requires six times the amount of mineral inputs needed for a conventional car). Examples of such restrictions include the controls on US exports of semiconductors to China, which were launched in 2022, and the restrictions on exports of critical metals introduced by China in response.[1] In 2025 China implemented export restrictions on seven rare earth minerals, while the US Administration started to investigate risks to national security posed by reliance on imported critical minerals. In addition, the EU has introduced measures to address critical dependencies (see Box 1).
Box 1
Addressing critical dependencies: the EU approach
Since 2020 the EU has elevated strategic autonomy into a central policy objective. An array of policy initiatives has been introduced to identify, monitor and reduce critical dependencies, including rare earth minerals, platinum group metals and materials needed for green and digital infrastructures (International Relations Committee, 2023). The EU’s framework has evolved to balance greater internal capacity with trade openness. The initiatives can be categorised into two interlinked strands: (1) strengthening domestic production capacity; and (2) diversifying and enhancing the resilience of existing global supply chains.
In the first strand, the EU has sought to reduce critical dependencies by boosting internal extraction, processing and manufacturing capabilities. Initiatives include:
- fiscal incentives, such as tax credits and subsidies for producers of clean technologies and semiconductors, including under the Green Deal Industrial Plan and the Chips Act[2] (both adopted in 2023);
- public and private investment schemes to support industrial innovation and the scaling-up of production, such as the Important Projects of Common European Interest (IPCEIs) on microelectronics, hydrogen, batteries, cloud infrastructure and digital communication;
- benchmarks and targets to diversify and enhance EU supplies, including under the Critical Raw Materials Act[3] (2024);
- regulations to make state aid rules more flexible and to streamline permitting procedures for mining and processing operations for critical raw materials, including under the Net-Zero Industry Act[4] (2024) and the Critical Raw Materials Act (2024);
- protective trade instruments and revised public procurement policies to shield domestic producers from unfair competition and to target foreign subsidies distorting competition in the EU internal market, including under the Foreign Subsidies Regulation[5] (2023) and the International Procurement Instrument[6] (2022).
In the second strand, the EU has implemented policies to stabilise and diversify supply sources, notably:
- strategic partnerships and bilateral agreements with resource-rich and like-minded partners (e.g. United States, Canada, Australia, Chile, Namibia), incorporating environmental, social and governance standards;
- regulations on the strategic stockpiling and joint procurement of critical goods, including via pooled purchasing mechanisms modelled on the COVID-19 pandemic vaccine strategy, aimed at ensuring continuity in supply.
Policies have been designed to foster collaboration between EU institutions, EU Member States and private stakeholders. For example, both the Critical Raw Materials Act and the Chips Act foresee the creation of boards that include representatives of Member States and the European Commission to advise on and coordinate the implementation of measures and oversee the designation and support of strategic projects within the EU. Under the Green Deal Industrial Plan and the Chips Act, public-private partnerships have been established to encourage research and development in semiconductors and industrial decarbonisation. In general, most policies have been funded through a combination of funds from the EU, Member States and the private sector, except for IPCEIs, which are financed primarily by national budgets.
These policies have yielded tangible progress in reshoring production, stimulating investment and launching cross-border industrial projects. Arjona et al. (2025) find that EU imports are shifting away from countries without trade agreements with the EU and gravitating both inwards and towards regional neighbours and partners engaged in active trade initiatives. At the same time, in some cases, long approval processes – particularly for mining, renewable infrastructure and manufacturing facilities – have delayed project deployment. In addition, while some Member States have mobilised significant public support, fiscal disparities across the EU have led to uneven industrial scaling. Lastly, the implementation of policies can be somewhat fragmented across Member States, owing to varying levels of commitment or administrative capacity.
The European Commission has proposed several new policies for 2025 and beyond. These include the creation of an EU Critical Raw Material (CRM) platform to serve as a coordination hub with a view to enhancing CRM supply chain monitoring, and to further support joint procurement and stockpiling. In addition, the Commission has proposed the creation of a European Sovereignty Fund, aimed at addressing investment asymmetries between Member States and at bolstering EU-level support for strategic sectors. Lastly, new strategic partnerships in CRM and clean technologies are under negotiation with countries in Latin America, Africa and South-East Asia.
Recent data point to a reconfiguration of trade along geopolitical fault lines for some products of strategic relevance, though the situation remains fluid. Aggregate indicators of trade integration, such as participation in global value chains, suggest that integration has plateaued rather than declined (Chart 2, panel a). At the same time, as shown in Attinasi et al. (2024), aggregate trends mask some reconfiguration of trade along geopolitical fault lines, particularly for goods of strategic importance from a national security standpoint, such as advanced technology products (e.g. integrated circuits, biotechnology devices). This reconfiguration, as evidenced by a sharp drop in US imports from China and a steep decline in EU imports from Russia, has accelerated since 2021, notably as a result of western sanctions on Russia (Conteduca et al., 2024; Airaudo et al., 2025). Moreover, the recent imposition by the United States of higher tariffs on its main trading partners might reshape global trade flows more significantly.
Chart 2
Trade integration and firms’ de-risking strategies
a) Trade along global value chains
(percentage of total trade)

b) Actions to reduce exposure to China
(percentage of firms surveyed relying on critical Chinese inputs)

Sources: Attinasi et al. (2024), Balteanu et al. (2024) and ECB staff calculations.
Notes: Panel a): trade along global value chains refers to merchandise trade crossing more than one border. Panel b): the surveys were conducted in 2023 and covered only manufacturing firms.
Surveys of manufacturing firms and supply chain analysis reveal a trend towards de-risking and supplier diversification strategies, notably among top US technological firms. Multinational companies report plans to relocate production owing to geopolitical tensions (HSBC, 2024), including in the euro area, where firms express a desire to implement more de-risking strategies in the future (Attinasi et al., 2023). This is particularly the case for firms relying on geopolitical rivals for key inputs. According to surveys conducted in 2023, in selected euro area countries, around half of the manufacturing companies sourcing inputs deemed critical from China had already implemented strategies to reduce supply chain risks, or were planning to do so by the end of 2024 (Chart 2, panel b). Beyond surveys, an analysis of supply chain data points to key US technological firms adjusting their supplier networks to reduce reliance on China. Both Apple and Tesla have significantly decreased their numbers of China-based suppliers since 2022 (Chart 3). While Chinese suppliers were replaced to some extent by firms in South-East Asia, the decline in the total number of suppliers would suggest that both firms consolidated their input providers into fewer but more reliable suppliers.
Chart 3
Evolution of suppliers by origin
(number of direct suppliers)

Sources: Bloomberg and ECB staff calculations.
This article sheds light on the evolution and economic relevance of critical dependencies across the United States, the euro area and China. Critical dependencies are defined as goods with limited import diversification, meaning that any disruption to their supply could have a severe impact on strategic sectors. We complement this importer perspective with a network analysis focused on exports in order to assess supply concentration and potential disruption risks if a few dominant players restrict access to critical dependencies. Lastly, we conduct a model-based analysis to examine the extent to which sudden supply disruptions to critical dependencies could entail significant economic costs despite their small share in total trade.
2 Evolution of critical dependencies
Critical dependencies are identified as strategically important inputs for which there is heavy reliance on a small number of foreign suppliers. This article identifies critical dependencies following the framework of Arjona et al. (2023) for the EU, which we adapt to cover the United States, China and the euro area (see Box 2).[7] The methodology relies on product-level trade data to single out critical dependencies from 5,000 commodities by identifying products with a low import diversification, a scarcity of global supply and a limited presence of domestic capacity. Products that have those three characteristics and feature in a pre-defined list of strategic sectors (European Commission, 2021), such as health products, batteries, hydrogen and electronic chips, are pinpointed as “critical dependencies”.
Box 2
A data-driven methodology for identifying critical dependencies
Identifying critical dependencies is essential because disruptions to key inputs can pose significant risks to strategic sectors. However, the breadth of global trade – spanning many goods and partners – makes their identification challenging. We leverage the BACI dataset (Gaulier and Zignago, 2010) covering 5,000 goods and 238 regions, and follow the methodology of Arjona et al. (2023), originally developed for the EU, which we adapt to a global perspective. The methodology consists of three steps: (1) computing core dependency indicators (CDIs); (2) filtering for products with high levels of dependency; and (3) selecting products belonging to strategic sectors. The result is a set of goods with high import concentration, elevated external reliance and limited domestic substitutability.
Step 1: Computing CDIs
The first indicator measures import diversification. For country and product , this is the sum of squared import shares as follows:
where are country ’s imports of product from country , and are country ’s total imports of product . The threshold for selecting products with low import diversification is 0.4, as in Arjona et al. (2023), which means filtering for products where imports originate from fewer than three countries.
The second indicator accounts for scarcity of global supply. For each product , we calculate the ratio of a country’s imports () over global imports ():
and apply country-specific thresholds: a product is filtered out if a country’s share in global imports for that product is higher than the country’s share in global imports across all imported products.[8]
The third indicator is a proxy for domestic capacity. It assesses the degree to which imports can be substituted with domestic production, using the ratio of the country’s imports of product () to exports of that same product ():
As in Arjona et al. (2023), we select products for which is above 1 – meaning if a country imports more than it exports.
Step 2: Filtering for highly dependent products
We identify dependencies by applying the thresholds defined above. The main benefit of this method is simplicity, though the threshold choices might seem arbitrary. Hence, we use a complementary approach in line with Arjona et al. (2023) by filtering for products in the top 10% of the aggregate core dependency, computed as a simple average of ranks across the three CDIs. The final selection meets both conditions – CDIs exceeding thresholds and ranking in the top 10% of the aggregate core dependency.
Step 3: Identifying products in strategic sectors
The final step cross-checks the list of highly dependent products against the list of products of strategic importance. The classification of strategically important products is based on the list of sensitive ecosystems established by European Commission (2021) and used in Arjona et al. (2023). It includes products such as batteries, electronic chips and critical raw materials.[9] Although strategic sectors may differ across economies, using a common definition facilitates comparability.
Over the past 30 years, the number of critical dependencies has declined slightly for the euro area and China, and remained comparatively high for the United States. In the 2020s the number of critical dependencies was around 100 for the euro area and China, and 120 for the United States (Chart 4, panel a), representing, on average, 7% of total imports.[10] Since the 1990s the composition of critical dependencies has varied across economies, reflecting differences in industrial policies and the positioning along global value chains. For the United States, the overall number of critical dependencies has remained largely unchanged, as reduced dependencies on intermediate inputs were offset by a steep rise in dependencies on final goods, especially consumer electronics (e.g. radios, televisions). The euro area has followed a similar pattern, albeit to a lesser extent, with fewer critical dependencies on intermediate goods and increasing dependencies on final products. In 2023 critical dependencies for the euro area included critical raw materials (e.g. uranium, manganese), pharmaceuticals (e.g. hormones, antibiotics) and household appliances (e.g. toasters, vacuum cleaners). By contrast, China saw a steady decline in critical dependencies between the 1990s and the 2010s, mainly on account of final goods, as the country expanded its manufacturing base and made efforts to reduce reliance on foreign partners (e.g. “Made in China 2025” plan in 2015, Dual Circulation Strategy in 2020). In the 2020s there was a slight rebound, driven mainly by rising demand for minerals (e.g. copper, nickel, beryllium) used in the rapidly developing technology sector.
Chart 4
Evolution of critical dependencies
a) By end use
(number of products)

b) By geographical source
(number of products)

Sources: BACI (HS 92, 6-digit level), OECD and ECB staff calculations.
Notes: Averages over decades (“1990s” refers to the period 1995-99 and “2020s” refers to the period 2020-23 owing to data availability). The euro area is treated as a single entity, abstracting from intra-euro area trade. Panel a): the classification is based on the OECD Bilateral Trade in Goods by Industry and End-use database. Panel b): geographical source is defined as the main source of imports.
The limited variation in total dependencies masks a reshuffling along the geographical dimension. Breaking down dependencies by origin shows that China has reduced reliance on advanced economies, while increasing reliance on emerging economies, notably Indonesia, Thailand and Russia (Chart 4, panel b). This reflects the expansion of the Chinese manufacturing sector, which has lowered its dependency on industrial products from advanced economies (e.g. electric motors, machinery) but increased reliance on raw minerals (e.g. nickel, zinc) from emerging economies. At the same time, both the United States and the euro area saw a marked increase in critical dependencies from China in the 2000s and 2010s and a move away from other advanced economies. In the 2020s signs of a trend reversal emerged in the United States, while in the euro area there was a plateauing of critical dependencies from China. To some extent, these developments reflect a strategic reconfiguration of supply chains as the United States tried to curb its exposure to China (e.g. 2018-19 trade war, the Inflation Reduction Act of 2022 targeting inputs from non-allied countries). Given the importance of geopolitical considerations, the remainder of this article focuses on reciprocal dependencies between western economies and China.
The dependence of the United States and the euro area on China increased the most in the electronics and chemicals sectors, reflecting China’s growing role as a global manufacturing hub (Chart 5, panel a). For most of the products in these sectors, the United States was not critically dependent in the 1990s, but progressively became so as China gained a central position as an exporter of items such as graphite, plastics and consumer electronics. For the euro area, critical dependencies from China followed a broadly similar pattern, although it was more pronounced for chemicals than electronics. At the product level, the United States and the euro area share some similar critical dependencies from China, in particular consumer electronics and health products (e.g. vitamins and hormones, cooking appliances, data storage units). For China, critical dependencies from the euro area and the United States are lower, decreasing and less concentrated. This partly reflects China’s position as a manufacturing superpower, as well as the Chinese government’s efforts to gain strongholds in strategic sectors (Baldwin, 2024). For instance, China significantly reduced its dependencies on vehicles, machinery and aircraft.
Chart 5
Critical dependencies between China and western economies
a) By sector
(number of products)

b) By supply chain position
(number of products)

Sources: BACI (HS 92, 6-digit level) and ECB staff calculations.
Notes: For the euro area and the United States, the chart shows dependencies from China, while for China it shows dependencies from both the euro area and the United States. Averages over decades (“1990s” refers to the period 1995-99 and “2020s” refers to the period 2020-23 owing to data availability). The euro area is treated as a single entity, abstracting from intra-euro area trade. Geographical source of critical dependencies is defined as the main source of imports. Panel a): the sector classification is based on the United Nations International Standard Industrial Classification of All Economic Activities (ISIC) Rev. 4. “Chemicals” refers to divisions 19-22, “Electronics” to 26-27, “Metals” to 24-25 and “Mining” to 3-5. Panel b): the supply chain position is based on upstreamness indices from Antràs et al. (2012) with “Downstream” and “Upstream” corresponding to the lower 33 and upper 67 percentiles of the distribution of indices respectively.
From a supply chain perspective, the United States and the euro area have become increasingly reliant on Chinese downstream products (Chart 5, panel b). This is most noticeable in the United States, where dependence on downstream goods has grown fivefold since the 1990s, while reliance on upstream products has remained relatively constant, meaning that the critical exposure of the United States to China is mainly in consumption goods, in particular consumer electronics. By contrast, for the euro area, dependencies on upstream and downstream products have increased in a broadly similar way. In the meantime, China has lowered its dependency on downstream products from the euro area and the United States. This signals a shift in supply chains and shows how China has become a key hub for the assembly of consumer goods shipped to advanced economies.
3 Risk of supply disruptions: a network analysis
A network analysis complements the assessment of critical dependencies by identifying exporters with the capacity to influence access to critical goods. While the analysis of critical dependencies focused on the importer perspective, the network analysis looks at the exporter perspective, capturing whether some countries hold dominant positions in the supply of critical products. In line with Arjona et al. (2023), we rely on two metrics: export concentration and network centrality (see Box 3). The two complement each other, as looking at export concentration alone does not necessarily fully capture a country’s influence within the global production network. In fact, while a country with a high share of global exports would be in an influential position, countries that act as distribution intermediaries can play an equally important role. For this reason, considering both metrics facilitates a more accurate identification of what Arjona et al. (2023) refer to as “single point of failures” (SPOFs), which are the nodes in trade networks where supply disruptions can be the most harmful. We first document risks for critical dependencies in general and then zoom in on critical raw materials.
Chart 6
Single point of failures among critical dependencies between China and western economies
(concentration and centrality indices for each product deemed a critical dependency)

Sources: BACI (HS 92, 6-digit level) and ECB staff calculations.
Notes: SPOFs stands for single point of failures. Each dot represents a critical dependency. For the euro area and the United States, the chart shows only dependencies from China, while for China it shows dependencies from both the euro area and United States. The euro area is treated as a single entity, abstracting from intra-euro area trade.
For the euro area and the United States, around 30% and 40% respectively of critical dependencies from China were at risk of being SPOFs in 2023. Critical dependencies at risk of being SPOFs are those products that have both high export concentration and high network centrality (top right-hand quadrants of panels a) and b) in Chart 6). Focusing on critical dependencies from China in 2023 (panel b), SPOFs are identified in 30% and 40% of cases for the euro area and the United States (blue and yellow dots) respectively. Notably, the high-risk products include health products (e.g. antibiotics, vitamins) and consumer electronics (e.g. radios). The number of SPOFs for critical dependencies from China has grown substantially since 1995 (panel a), reflecting both the increase in critical dependencies from China and the more central role played by China in global trade. By contrast, China managed to eliminate almost all of its high-risk critical dependencies from the euro area and the United States – with the exception of some key industrial chemicals (e.g. cellulose, hexamethylenediamine).
Box 3
A complementary network analysis
The identification of critical dependencies is complemented by a network analysis that pinpoints key bottlenecks in the global supply of strategic goods. The evaluation follows Arjona et al. (2023) and relies on two metrics: (1) concentration of global exports and (2) network centrality.
Export concentration is measured using a Herfindahl-Hirschman Index (HHI). It is formally measured as follows:
where are exports of product from country , and are global exports of product . The higher the HHI, the more concentrated is the global market, suggesting the market power is split among only a few countries.
We complement the HHI with a measure of network centrality that assesses whether a country acts as a key hub. As in Barrat et al. (2004), the centrality of country in the global trade network for product relies on the market share of country in imports of other countries. The formula is:
where are exports of product from country to country , and are total imports of product by country . Since the number of importing countries () can differ across products and time, for comparability we normalise by the maximum centrality (), which is an extreme situation where one country serves all the others.[11] The index is bounded between 0 and 100, indicating low and high influence respectively. Changes in centrality are driven by either the extensive margin (if a country trades with more partners) or the intensive margin (if it deepens its market share in existing partners).[12]
Lastly, we compute the aggregate network centrality of a product as the standard deviation of network centrality indices across all countries, as in Korniyenko et al. (2017). A higher standard deviation indicates a more centralised network where a few countries dominate.
Since the 2000s China has expanded its role as a key hub for critical raw materials. China dominates the supply chain of minerals that are essential for modern technologies. It refines around 73% of the cobalt and 40% of the lithium in the world (Vivoda, 2023) and accounts for over 95% of global rare earth production. As a result, China’s central role in the supply chain of critical minerals makes the country a key player in the security of supply.
Regarding critical raw materials, a key concern has been the supply security of dual-use minerals, such as cobalt, magnesium and lithium. Dual-use minerals are those that have both military and civilian applications. For instance, the uses of cobalt include battery technology for electric vehicles and in the defence sector – and the euro area relies on Asia for 75% of its cobalt imports. Magnesium, a critical component in the defence, aerospace and automotive sectors, is sourced mainly from China for the euro area (85% of imports) and from Israel for the United States (58% of imports). In the case of lithium, which is vital for clean energy and defence, the United States is dependent mainly on Chile (94% of imports), while the euro area is dependent on Chile and China (39% and 18% of imports respectively).[13] The evolution of trade in these dual-use minerals provides an illustration of the relevance of network centrality, with China having secured a dominant position in each case.
- China’s network centrality for cobalt does not stem from high export concentration. The cobalt supply is highly concentrated in the Democratic Republic of the Congo (DRC), which dominates global exports (Chart 7, panel a). However, China has emerged as a key distribution intermediary despite its limited domestic extraction capacities (Chart 7, panel b), as the DRC exports mostly to China – its share in the DRC’s cobalt exports rose from below 10% in the 1990s to 75% in the 2020s.[14] By re-exporting to a growing number of partners, China has become as central a player as the DRC and now rivals the United States, despite the latter’s dominance in the 1990s (Chart 7, panel c).[15]
- Over time China has expanded its central position in exports of magnesium. In the 1990s China was already a key exporter of unwrought magnesium, with a 30% global share. Since then it has significantly increased its export share, to 75% in the 2020s, leading to a sharp rise in export concentration (Chart 7, panel a). China has also consolidated its central position in the network to much higher levels than those of the United States and the euro area (Chart 7, panel b). This increase in centrality is explained by both penetration of new markets and the strengthening of existing trade links (Chart 7, panel c).
- China has consolidated its dominant position in exports of lithium in more recent years. During the 1990s China was a marginal exporter of lithium, whereas in the 2020s its share in global lithium exports rose to 75%, overtaking that of western suppliers (Chart 7, panel a). Until the 2010s this diversification meant that export concentration was limited, but China’s growing dominance pushed concentration to record levels in the 2020s owing to massive production and refining expansion which, to some extent, has crowded out other producers. Starting from a low network centrality (Chart 7, panel b), China expanded its trade network and deepened ties with existing partners (Chart 7, panel c) to become the most central global player by the 2020s, displacing the United States.
Chart 7
Network analysis of selected critical raw minerals
a) Export concentration
(index)

b) Network centrality
(index)

c) Changes in China’s network centrality
(index)

Sources: BACI and ECB staff calculations.
Notes: DRC stands for Democratic Republic of the Congo. The euro area is treated as a single entity, abstracting from intra-euro area trade. Panels a) and b): averages over decades (“1990s” refers to the period 1995-99 and “2020s” refers to the period 2020-23 owing to data availability). Panel b): the DRC is singled out only for cobalt because of the large share of cobalt in the country’s exports.
4 Costs of critical dependencies: a model-based assessment
We model the economic costs of potential disruptions to the supply of critical dependencies using the multi-country, multi-sector model of Baqaee and Farhi (2024). The model simulates the effects of supply shocks and their propagation through global production networks, including to downstream consumers and suppliers, also taking into account non-linear effects of shocks across countries and sectors. To allow for the low degree of substitutability of critical inputs, the Baqaee-Farhi model is calibrated as in Attinasi et al. (2024), which notably embeds low elasticities of substitution from Boehm et al. (2023).
To model the impact of shocks related to shortages of critical dependencies, we rely on a new methodology for building granular input-output (IO) tables. Models such as Baqaee-Farhi are usually calibrated with standard IO tables, which have a high level of sectoral aggregation, making it challenging to simulate shocks to specific products such as those identified in the analysis of critical dependencies. For example, a standard IO table bundles cobalt with many non-critical commodities (e.g. marble, sandstone) in a “mining and quarrying” sector, hence making it impossible to simulate targeted shocks and subsequent propagation across sectors and countries. To overcome these limitations, Conteduca et al. (2025) propose a data-driven methodology to disaggregate IO tables and isolate relevant niche products. We apply this methodology to build an IO table tailored to the dependencies identified in Box 2.
For each country, we study a sudden halt to the supply of products for which that country is critically dependent on foreign suppliers. The simulations assume a large increase in the trade costs of imports from the China-led eastern bloc for the United States or the euro area, and vice versa.[16] To gauge how the economic significance of these critical dependencies has evolved over time, we run this scenario twice for each country, using the dependencies and IO tables for 1995 and 2023.
Over time the costs of a sudden halt to the supply of critical dependencies has risen for the euro area and the United States but declined for China, although they remain much higher for China (Chart 8, panel a). While China remains more vulnerable to western supply disruptions to critical dependencies, its push for self-reliance (see Section 2) has reduced its vulnerability over the past 30 years. Losses in Chinese final demand from a sudden halt to the supply of western-produced critical dependencies is estimated to have decreased from 2.1% in 1995 to 1.4% in 2023. However, the opposite holds for the euro area and the United States. Since 1995 losses in final demand owing to a sudden halt to the supply of critical dependencies from the eastern bloc have risen tenfold in the euro area (from 0.04% to 0.41%), driven by a higher dependency on Chinese inputs and a more widespread use of these inputs in production. Such losses also increased significantly for the United States (from 0.08% to 0.32%), although to a lesser extent than in the euro area given the higher starting point and efforts by US Administrations to curb dependency on China. At the same time, losses in Chinese final demand from a sudden halt to the supply of critical dependencies remain well above final demand losses in the euro area and the United States.
Chart 8
Impact of a supply shock to critical dependencies
a) Final demand
(percentage deviations from steady state)

b) Shares of supply disrupted and final demand affected
(percentages)

Sources: Baqaee and Farhi (2024), Conteduca et al. (2025), BACI, OECD and ECB staff calculations.
Note: Panel b): values refer to 2023.
The key role of critical dependencies as production inputs, as well as their low degree of substitutability, amplifies the effects of sudden shortages. While critical dependencies represent only a tiny fraction of the total intermediate inputs used for production (0.01% in the United States and 0.07% in China), the impact of a sudden halt to their supply on final demand is disproportionate, being around 20 times larger than the proportion of inputs that is disrupted (Chart 8, panel b).
The model-based results are likely lower-bound estimates. Results from the Baqaee-Farhi model abstract from short-term amplification mechanisms, especially if disruptions to the supply of critical inputs give rise to uncertainty or to episodes of financial turmoil. In addition, the loss of critical inputs can cause temporary production stoppages, which are not taken into account in the model. Lastly, elasticities of substitution for highly specialised inputs (e.g. rare earth minerals) might be smaller than assumed in our calibration. Irrespective of the list of critical dependencies, all these effects would exacerbate losses.
5 Conclusions
This article sheds light on the economic risks associated with critical dependencies. Since the 1990s trade liberalisation and specialisation of production across geographically dispersed networks have enabled substantial efficiency gains. However, this has resulted in asymmetric dependencies because some countries have secured dominant positions in global supply chains (e.g. China in strategic raw minerals such as lithium, magnesium and cobalt), while others have become acutely reliant on foreign inputs. These dependencies create strategic vulnerabilities, as their disruption by geopolitical rivals can entail significant economic costs.
As advanced economies have deepened their exposure to foreign inputs, the impact of trade disruptions has intensified considerably. Trade conflicts now have far greater economic repercussions than in earlier decades, as the euro area and the United States have seen rising costs associated with their dependencies on Chinese inputs.[17] Conversely, China has reduced its dependence on foreign inputs.
In the wake of recent global shocks, governments are reassessing their approach to trade, which is increasingly subject to geopolitical influence. Focus is placed on critical inputs which, though small in value, are difficult to replace and to which supply disruptions can severely amplify inflation and dampen demand.
Despite these risks, the macroeconomic costs of a full decoupling of geopolitical blocs would likely exceed those associated with critical dependencies – while failing to eliminate them. Model simulations suggest that a comprehensive decoupling of geopolitical blocs could reduce global GDP by up to 12% in the long run and temporarily push up inflation by as much as 4 percentage points in the first year (Goes and Bekkers, 2022; Attinasi et al., 2025a; 2025b). These costs stem from output losses, rising input prices and inefficiencies due to fragmented trade flows. Moreover, protectionist measures may prove ineffective at eliminating dependencies, as trade often reroutes through neutral third countries (Attinasi et al., 2024).
Policymakers therefore face a trade-off between strengthening supply chain resilience and preserving the benefits of openness. To navigate this dilemma, rather than resorting to blanket protectionism, governments should adopt targeted and coordinated de-risking strategies. Such strategies should aim to address specific vulnerabilities while preserving the economic gains of global integration. This balanced approach is key to ensuring both resilience and long-term prosperity. To help in the tailoring of policies to relevant dependencies, the type of data-driven analysis carried out in this article provides some insight into how to unveil critical dependencies that are not visible in aggregate data.
References
Airaudo, F., de Soyres, F., Gaillard, A. and Santacreu, A.M. (2025), “Recent evolutions in the global trade system: from integration to strategic realignment”, ECB Forum on Central Banking 2025 – Adapting to change: macroeconomic shifts and policy responses, ECB.
Antràs, P., Chor, D., Fally, T. and Hillberry, R. (2012), “Measuring the Upstreamness of Production and Trade Flows”, American Economic Review, Vol. 102, No 3, May, pp. 412-416.
Arjona, R., Connell, W. and Herghelegiu, C. (2023), “An enhanced methodology to monitor the EU’s strategic dependencies and vulnerabilities”, Single Market Economics Papers, No 14, European Commission, March.
Arjona, R., Connell, W. and Herghelegiu, C. (2025), “EU supply chain tectonics”, VoxEU Columns, Centre for Economic Policy Research, 3 April.
Attinasi, M.G., Ioannou, D., Lebastard, L. and Morris, R. (2023), “Global production and supply chain risks: insights from a survey of leading companies”, Economic Bulletin, Issue 7, ECB.
Attinasi, M.G. et al. (2024), “Navigating a fragmenting global trading system: insights for central banks”, Occasional Paper Series, No 365, ECB.
Attinasi, M.G., Boeckelmann, L. and Meunier, B. (2025a), “The economic costs of supply chain decoupling”, The World Economy, Vol. 48, No 3, March, pp. 598-627.
Attinasi, M.G., Boeckelmann, L., Gerinovics, R. and Meunier, B. (2025b), “Steeling losses: sectoral strains from the return of tariffs on steel and aluminium”, Econbrowser, 13 March.
Baldwin, R. (2024), “China is the world’s sole manufacturing superpower: A line sketch of the rise”, VoxEU Columns, Centre for Economic Policy Research, 17 January.
Balteanu, I., Bottone, M., Fernández-Cerezo, A., Ioannou, D., Kutten, A., Mancini, M. and Morris, R. (2024), “European firms facing geopolitical risk: Evidence from recent Eurosystem surveys”, VoxEU Columns, Centre for Economic Policy Research, 18 May.
Baqaee, D. and Farhi, E. (2024), “Networks, Barriers, and Trade”, Econometrica, Vol. 92, No 2, March, pp. 505-541.
Barrat, A., Barthelemy, M., Pastor-Satorras, R. and Vespignani, A. (2004), “The architecture of complex weighted networks”, Proceedings of the National Academy of Sciences, Vol. 101, No 11, March, pp. 3747-3752.
Boeckelmann, L., Imbs, J. and Pauwels, L. (2025), “Accounting for propagation”, Working Paper Series, forthcoming, ECB.
Boehm, C., Levchenko, A. and Pandalai-Nayar, N. (2023), “The Long and Short (Run) of Trade Elasticities”, American Economic Review, Vol. 113, No 4, April, pp. 861-905.
Bonneau, C. and Nakaa, M. (2020), “Vulnerability of French and European imports”, Tresor-Economics, No 274, Ministère de l’Économie et des Finances, December.
Conteduca, F.P., Giglioli, S., Giordano, C., Mancini, M. and Panon, L. (2024), “Trade fragmentation unveiled: five facts on the reconfiguration of global, US and EU trade”, Occasional Papers, No 881, Banca d’Italia, October.
Conteduca, F.P., Mancini, M., Romanini, G., Giglioli, S., Borin, A., Attinasi, M.G., Boeckelmann, L. and Meunier, B. (2025), “Fragmentation and the future of GVCs”, Occasional Papers, No 932, Banca d’Italia, April.
de Benedictis, L., Nenci, S., Santoni, G., Tajoli, L. and Vicarelli, C. (2014), “Network Analysis of World Trade using the BACI-CEPII dataset”, Global Economy Journal, Vol. 14, No 03n04, October, pp. 287-343.
European Commission (2021), “Strategic dependencies and capacities”, Staff Working Document, SWD(2021) 352 final, Brussels, May.
Feyrer, J. (2019), “Trade and Income – Exploiting Time Series in Geography”, American Economic Journal: Applied Economics, Vol. 11, No 4, October, pp. 1-35.
Frankel, J. and Romer, D. (1999), “Does Trade Cause Growth?”, American Economic Review, Vol. 89, No 3, June, pp. 379-399.
Gaulier, G. and Zignago, S. (2010), “BACI: International Trade Database at the Product-Level. The 1994-2007 Version”, CEPII Working Paper, No 2010-23, October.
Goes, C. and Bekkers, E. (2022), “The impact of geopolitical conflicts on trade, growth, and innovation”, Staff Working Papers, No 9, World Trade Organization, July.
HSBC (2024), “The great relocation – how supply chains are shifting”, General Research Insights, Global Research, March.
International Relations Committee (2023), “The EU’s Open Strategic Autonomy from a central banking perspective – challenges to the monetary policy landscape from a changing geopolitical environment”, Occasional Paper Series, No 311, ECB, March.
Korniyenko, Y., Pinat, M. and Dew, B. (2017), “Assessing the Fragility of Global Trade: The Impact of Localized Supply Shocks Using Network Analysis”, Working Papers, No 2017/030, International Monetary Fund, February.
López, L., Odendahl, F., Párraga, S. and Silgado-Gómez, E. (2024), “The pass-through to inflation of gas price shocks”, Working Paper Series, No 2968, August, ECB.
Organisation for Economic Co-operation and Development (OECD, 2025), OECD Inventory of Export Restrictions on Industrial Raw Materials 2025: Monitoring the Use of Export Restrictions Amid Growing Market and Policy Tensions, OECD Publishing, Paris, May.
Vivoda, V. (2023), “Friend-shoring and critical minerals: Exploring the role of the Minerals Security Partnership”, Energy Research & Social Science, Vol. 100, No 103085, June.
US export controls on semiconductors were tightened further in 2023 and 2024. China’s restrictions on exports included gallium and germanium in 2023 and were expanded to antimony, graphite and superhard materials in 2024.
Regulation (EU) 2023/1781 of the European Parliament and of the Council of 13 September 2023 establishing a framework of measures for strengthening Europe’s semiconductor ecosystem and amending Regulation (EU) 2021/694 (Chips Act) (OJ L 229, 18.9.2023, p. 1).
Regulation (EU) 2024/1252 of the European Parliament and of the Council of 11 April 2024 establishing a framework for ensuring a secure and sustainable supply of critical raw materials and amending Regulations (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1724 and (EU) 2019/1020 (OJ L 2024/1252, 3.5.2024).
Regulation (EU) 2024/1735 of the European Parliament and of the Council of 13 June 2024 on establishing a framework of measures for strengthening Europe’s net-zero technology manufacturing ecosystem and amending Regulation (EU) 2018/1724 (OJ L 2024/1735, 28.6.2024).
Regulation (EU) 2022/2560 of the European Parliament and of the Council of 14 December 2022 on foreign subsidies distorting the internal market (OJ L 330, 23.12.2022, p. 1).
Regulation (EU) 2022/1031 of the European Parliament and of the Council of 23 June 2022 on the access of third-country economic operators, goods and services to the Union’s public procurement and concession markets and procedures supporting negotiations on access of Union economic operators, goods and services to the public procurement and concession markets of third countries (International Procurement Instrument – IPI) (OJ L 173, 30.6.2022, p. 1).
While there are other methodologies (Korniyenko et al., 2017; Bonneau and Nakaa, 2020), we use Arjona et al. (2023) because of its: (1) comprehensiveness, as it combines several methods used separately in the literature, notably import concentration and export network analyses; (2) robustness, as it leverages multiple indicators; and (3) high granularity.
For the EU, Arjona et al. (2023) use the ratio of extra-EU imports over total EU imports. Given our global perspective, we modify this indicator slightly.
The fourth list of critical raw materials includes rare earth minerals and 28 other materials (e.g. bauxite, tungsten).
The number of critical dependencies for the euro area (around 100) is lower than the 200 reported in Arjona et al. (2023). This is due to: (1) differences in geographical scope, i.e. EU in Arjona et al. (2023) and euro area in this article; (2) adjustments to the methodology; (3) the fact that, in order to facilitate comparability over the period 1995-2023, we use the 1992 version of Harmonized System (HS) codes, covering around 5,000 goods, while Arjona et al. (2023) uses the 2017 version, covering 10% more goods; and (4) the use of TRADE-FIGARO-EUROSTAT data in Arjona et al. (2023), which corrects for re-exports extensively for the EU but to a lesser extent for third countries, while this article uses BACI data, which are uncorrected for re-exports but ensure a symmetric treatment of EU countries and third countries such as the United States and China. For Arjona et al. (2023), using BACI data results in around 30% fewer critical dependencies than when using TRADE-FIGARO-EUROSTAT data.
If a country does not import product , we normalise by since it would be the maximum number of foreign destinations. While conceptually similar, our index differs from Arjona et al. (2023) because (1) we use market shares instead of export values divided by average value of imports, and (2) we normalise the index as in de Benedictis et al. (2014).
There is a third effect from changes in the number of importing countries (), but it is generally small and therefore allocated proportionally to other channels in our analysis.
Its high share in US and euro area lithium imports notwithstanding, Chile’s centrality is low as it exports to only a few countries.
China’s high share in the DRC’s cobalt exports is due to Chinese ownership of major cobalt mines in the DRC and China’s dominant position in refining.
The euro area’s role in the global cobalt network is mostly due to Finland’s cobalt refining activities.
For completeness, it should be noted that for this analysis, it is assumed that the world is divided into the three geopolitical blocs (western, eastern and neutral) along the lines of Attinasi et al. (2024). The scenario simulations are based on the list of critical dependencies without filtering for strategic sectors, as in step 3 in Box 2.
This is also the case for prices, which are not covered here. They are, however, shown in López et al. (2024) for the Russian gas shock.