Economic Analysis of Supply Chain Vulnerabilities (Summary)

Opening Remarks

FUKAO Kyoji (Chairman, RIETI / University Professor, IER, Hitotsubashi University)

This symposium, Economic Analysis of Supply Chain Vulnerabilities, is held as part of METI’s Economic Security Global Forum Weeks. The Japanese government is considering establishing a comprehensive think tank on economic security, with RIETI as a potential base. Economic security analysis is built on two pillars: crisis response and crisis preparation. Response focuses on rapid action, mainly by governments, to minimize damage from shocks like supply disruptions or cyberattacks. Preparation includes supply chain monitoring, information sharing, and simulations, where economic research institutes like RIETI can make major contributions. The symposium examines supply chain vulnerabilities, global value chains, simulation analysis, and the Chinese economy, presenting research from RIETI, JOGMEC, IDE-JETRO, and international experts to advance policy, research, and cross-sector collaboration.

TAKAHARA Ichiro (Chairman& CEO, JOGMEC)

This symposium focuses on supply chain vulnerabilities, a pressing economic security challenge highlighted by geopolitical tensions, pandemics, and natural disasters. These recurring global risks have revealed how fragile current supply chains can be, making international cooperation increasingly important. JOGMEC works to secure stable supplies of energy and mineral resources through procurement, stockpiling, and support for technological innovation, with the goal of building sustainable supply chains. I hope this symposium will provide a venue for sharing expertise on various topics and for generating new ideas through which we can find future solutions.

IMAIZUMI Shinya (Executive Vice President, IDE-JETRO)

This symposium examines economic security from a data driven perspective. Earlier discussions highlighted how growing international instability, economic interdependence, and digitalization have transformed goods, money, and information into tools of economic coercion. Building on this context, the symposium applies quantitative analysis and simulations to supply chain vulnerabilities to clarify current risks and test countermeasures. IDE-JETRO brings expertise in value chain analysis and simulation modeling, working with RIETI to generate knowledge and strengthen cooperation in security research.

Session 1: How can we mitigate supply chain risk?

Session Chair:
Satoshi HASHIMOTO (Deputy Director General, JOGMEC)

Presentation 1

Stéphane BOURG (Director of OFREMi, BRGM)

The French Observatory of Mineral Resources for Industrial Sectors (OFREMi) operates within a national interministerial framework coordinating critical raw materials policy. OFREMi serves as the technical arm, providing both broad and specific criticality assessments, vulnerability analysis, and decision-making support, alongside national and European investment mechanisms linked to the Critical Raw Materials Act and the EU resources action plan.

Our work prioritizes the specific needs of our economy. Objectives such as energy transition and geopolitical sovereignty translate into requirements in mobility, energy, and digital technologies, while specific technologies also require specific metals, including base and specialty materials. To understand risks, we conduct criticality assessments and enterprise-specific analyses. To understand vulnerabilities, we carry out stress tests based on value chain mapping, impact analysis, and mitigation planning, supported by alert systems and interactive dashboards.

Stress testing requires deep engagement with industry. Each value chain is structured differently, so we work through confidential industry working groups to build iterative value chain maps, assess impacts, and evaluate response capacity. A key concept is the cost of inaction. Modest additional costs to secure supply chains can function as an insurance premium when compared with the potential loss of entire industrial sectors.

We also use tabletop exercises to raise awareness and test responses across government and industry. The main lessons are clear: involve industry early, strengthen public private collaboration, understand geopolitical drivers, and build partnerships to reduce strategic vulnerabilities.

Presentation 2

Siyamend AL BARAZI (Head of Unit – Mineral Economics – German Mineral Resources Agency, DERA)

Germany faces challenges similar to Japan as a highly import-dependent economy. While Germany does not have an institution equivalent to JOGMEC, a Raw Materials Fund was established last year, with one billion euros available to invest in mining, refining, and recycling projects along the value chain.

I work at the German Mineral Resources Agency, part of the Federal Institute for Geoscience and Natural Resources under the Federal Ministry of Economic Affairs and Energy. Our role is to support industry and government through monitoring, data, and risk assessment. In recent years, repeated disruptions and export restrictions made it clear that traditional tools were insufficient.

We launched a pilot to track incidents affecting mining, refining, and trading worldwide. As this could not be developed in house, we partnered with external specialists and now work with Everstream Analytics. The system monitors supplier, climate and environmental risks, geopolitical and socio-political risks, as well as other risks with particular focus on upstream production disruptions. It currently covers 908 mining and refining sites across ten commodities and is being expanded. Relevant incidents are reviewed and assessed before alerts are shared with industry and government. A test phase is currently underway with German industry.

This work demonstrates that high quality, regularly updated data is essential. Precise location data, production volumes, and knowledge of key market operators are critical to reducing price and supply risks. Monitoring alone is not sufficient, but it provides a necessary foundation for effective risk mitigation.

Presentation 3

Simon WEIMER (Senior Manager, Raw Material Strategy and Risk Management, BMW Group)

The BMW Group has been working on responsible raw materials management and resilient procurement for many years, well before recent European Union initiatives such as the Critical Raw Materials Act or the Resource EU Action Plan announced last week. Drawing on long experience in international raw material markets, we have optimized our approach to current challenges and, two years ago, established a dedicated department for Raw Materials Strategy and Risk Management.

Our goal is to ensure responsible mining and processing while building resilient raw material supply chains. We start with a multi-stage raw material process and regularly analyze our raw materials portfolio to create transparency related to risks. This holistic assessment covers availability, ESG related risks, and price risks, and allows us to derive targeted mitigation measures.

Automotive supply chains are highly complex, particularly in Germany, so we apply a simple but effective approach to risk mitigation. To secure availability, especially for battery raw materials, we directly purchase materials such as cobalt and lithium and supply them to our value chain. To manage price risks, we hedge selected raw materials on financial markets. This proactive approach enables rapid responses to market changes.

All actions must balance resilience, competitiveness, and sustainability. Resilience means ensuring supply security, competitiveness focuses on cost and quality, and sustainability includes environmental responsibility and the use of secondary raw materials. Addressing these conflicting priorities requires diversification, a strong presence in global markets, and a European raw materials policy embedded in broader industrial policy. Promoting circular-economy solutions and recycling can reduce dependencies, but this demands consistent policy support. Japan’s close cooperation between government and industry provides a valuable model for Europe.

Discussion

HASHIMOTO:
JOGMEC is under strong policy pressure to reduce risks related to critical minerals and provides unique support such as financing and stockpiling. Demand for cooperation with private companies is rapidly increasing, but limited internal capacity is a major challenge.
First, what is the biggest current challenge facing your organizations?

BOURG:
The primary challenge is convincing industry that supply chains will never return to a stable, predictable state. Volatility, disruptions, and price swings are now permanent features. Recent events help to illustrate this reality. A second challenge is encouraging downstream industry to invest upstream. Minerals are extracted, refined, and processed not so that products can be sold, but so that cars can be built and planes can fly. Companies need secure access to these inputs and must participate directly, including through investment. This ultimately requires a public private system, since supply chains serve society as a whole. To support this shift, we must raise awareness and deploy tools such as real-time value chain monitoring to demonstrate to companies how quickly disruptions propagate so that they might be more prone to invest appropriately. Diversification, technological alternatives, and de-risking strategies are essential.

AL BARAZI:
Germany faces very similar challenges. Export controls on materials such as gallium, germanium, antimony, and tungsten have significantly affected industry, especially permanent magnets. Germany is the world’s largest importer of magnets and therefore highly import dependent. We are encouraging companies to invest more proactively across the supply chain, but this is difficult. Large-scale mining expansion in Europe is unlikely, and high energy costs limit certain activities in Germany. Stronger cooperation with global partners is therefore essential. These challenges will persist, but industry engagement is slowly increasing. Collaboration, including between Europe and Japan, is important.

WEIMER:
From our perspective, companies are reaching their limits. Supply chains are dominated by SMEs, which struggle to manage sustainability, diversification, and competitiveness simultaneously. Support mechanisms are needed, including cross-industry cooperation and public analysis tools. Focusing only on mining or OEM actions is insufficient.

HASHIMOTO:
What are your hopes for collaboration with Japan and JOGMEC?

WEIMER:
Cooperation already exists, but we need action, not more paperwork or memoranda. Germany and Europe are too slow to move from analysis to implementation. Japan offers a strong example, particularly through trading houses and investment mechanisms. Germany lacks equivalent structures, and partnerships with Japan or a German version of JOGMEC should be considered.

AL BARAZI:
There is clear scope for deeper collaboration. With a government fund now in place, joint investment is possible. The key is matching expertise in Germany and Japan to specific critical supply chains.

BOURG:
Economic security depends on public-private partnerships and cooperation among like-minded countries. JOGMEC is a key partner. Joint investments, such as rare earth projects in France, show that shared efforts enable outcomes neither side could achieve alone. We can also share methodologies and tools for risk and value chain analysis, while keeping sensitive data confidential.

HASHIMOTO:
What about academia?

BOURG:
We rely on research institutions for methodologies, including traceability, life cycle analysis, and monitoring tools. Key needs include new raw materials traceability technologies and domestically controlled IT capabilities, developed through close university industry collaboration.

AL BARAZI:
Germany has lost processing know-how for selected materials and faces high re-entry costs. There needs to be a new understanding of the importance of safeguarding industrial know-how, as once it is lost, is really has the potential to disappear for the long term. Academia is essential, but investment and action must accelerate.

WEIMER:
There is close cooperation with universities, particularly through initiatives such as the European mining master’s program, which trains the next generation of engineers needed to address future supply chain challenges.

Session 2: Firm-level data analysis on supply chain vulnerabilities

Session Chair:
Keiko ITO (Professor, Graduate School of Social Sciences, Chiba University)

Presentation 1

KAWAKUBO Takafumi (Fellow (Specially Appointed), RIETI / Assistant Professor, The University of Osaka)

Over the past decade, supply chains have faced repeated disruptions from Brexit, the COVID-19 pandemic, natural disasters, trade wars, and the war in Ukraine. These shocks are likely to become more frequent due to climate change and geopolitical tensions. At the same time, economic security has become a central issue for policymakers, reflected in measures such as revisions to Japan’s Economic Security Promotion Act.

Concern about supply chains has risen sharply alongside geopolitical fragmentation. To understand vulnerabilities, their impacts, and possible countermeasures, granular micro-datasets are essential. These include administrative data such as customs and Value Added Tax (VAT) data, which reveal international and domestic firm-to-firm linkages, and proprietary datasets such as those from Tokyo Shoko Research, as well as logistics data such as automatic identification system (AIS) ship tracking and bill-of-lading records. Together, these datasets enable analysis at the firm and transaction levels and allow researchers to map supply chain networks.

Recent research examines three major sources of disruption: natural disasters, trade wars, and conflicts. Studies of earthquakes, floods, and pandemic lockdowns show that average effects often mask substantial heterogeneity across firms. Firms with long-term supplier relationships suffer significant losses and struggle to switch suppliers, while firms that can substitute quickly are far more resilient. This highlights the importance of firms’ ability to reorganize supply chains. A key policy implication is that governments should improve firms’ access to information on alternative suppliers.

Research on trade wars shows clear trade diversion effects, such as Vietnam’s increased exports to the United States and imports from China following U.S.-China decoupling. Studies of geopolitical conflicts show large impacts, but also demonstrate that substitution can sharply reduce damage. When inputs can be replaced, even imperfectly, economic losses fall dramatically.

Ongoing work using AIS data shows that geopolitical shocks to maritime routes, such as Houthi attacks in the Red Sea, have forced rerouting around the Cape of Good Hope, substantially increasing fuel costs and CO2 emissions.

These findings underline the need for a comprehensive economic security think tank supported by continuous access to granular data. Combining customs, VAT, proprietary, and logistics datasets allows precise identification of vulnerabilities and supports timely and evidence-based policy responses. Academic research capacity, coordinated through institutions such as RIETI, is essential for strengthening economic security in an increasingly unstable and complex global environment.

Presentation 2

Steffen MUELLER (Professor of Economics and Head of Department for Structural Change and Productivity, Halle Institute for Economic Research (IWH), Germany)

Natural gas is a critical input for electricity generation, household heating, and many manufacturing processes. Before 2022, around 50% of German gas imports came from Russia, creating serious economic vulnerability and political exposure. Industry groups warned that a gas cutoff would cause immediate collapse of chemicals and widespread downstream failures, but the government lacked the data needed to verify these claims.

The economic consequences were unclear for two reasons. First, existing macro models were too aggregated and relied on strong assumptions about substitution, especially in the short run, and struggled to capture cascade effects along value chains. Second, Germany lacked sufficiently granular data. Input-output tables were too coarse, and there was no ready access to micro data on gas usage.

In response, we merged two product-level datasets from the German statistical office covering all manufacturing establishments with at least 20 employees. This produced detailed data on gas consumption and sales by product. It allowed us to identify gas-heavy products and gas-intensive products, defined as gas input per euro of sales. Gas use was extremely concentrated. Out of around 1,600 products, 300 accounted for 89% of gas consumption, and just 16 products accounted for 25%,　meaning that analysis of those products would have a significant impact.

To address downstream risks, we focused not on energy substitution, but on import substitution. By combining the product-level data with U.N. Comtrade data, we assessed which gas-intensive products could plausibly be imported, excluding risky suppliers. This showed that many highly gas-intensive products are also highly substitutable through imports. Eliminating production of the most gas-intensive and importable products could save large amounts of gas at relatively low economic cost, with no downstream cascade effects.

The policy message is to use granular data to identify a small set of critical products, assess value added per unit of gas, examine substitution possibilities at each stage of the value chain, and design targeted emergency plans. This approach is far more effective than uniform rationing and allows governments to reduce vulnerability while minimizing economic damage.

Comment

Chad P. BOWN (Reginald Jones Senior Fellow, Peterson Institute for International Economics (PIIE))

The work presented by Dr. Mueller and Dr. Kawakubo represents the best use of highly granular microeconomic data in identifying supply chain vulnerabilities and assessing how economies can respond to shocks. This kind of research is exactly what policymakers need in order to understand risks and build preparedness.

The key additional contribution is methodological. Ideally, governments would have had this type of analysis in real time during the Great East Japan Earthquake or immediately after the Russian invasion of Ukraine. These papers show what data, tools, and institutional capacity are required to analyze vulnerabilities as crises unfold, not years later. Policymakers should take seriously the implicit call to invest in data access, analytical infrastructure, and research capability to support both crisis response and crisis preparation.

From a policy perspective, however, the analysis must go further. Concepts such as short-run and long-run elasticities are not sufficient. Policymakers need specific timelines. Is substitution possible in three months, or in six months? Will industry resolve the problem before public support can be deployed? These details matter for prioritization and budgeting.

Similarly, recommendations to help firms find alternative suppliers must confront real-world constraints. Firms are often unwilling to share supplier information due to competitive concerns. Research must account for these institutional and strategic barriers. The same applies to proposals such as import substitution or subsidies for gas intensive sectors. Policymakers need concrete options, including whether measures are temporary or structural, and how they should be implemented.

This research is extraordinarily valuable, but its impact depends on translating findings into specific, operational policy choices. That translation is now the critical next step.

KAWAKUBO:
There is often a gap between the pace of academic research and the needs of real-time policy decision making. Rigorous empirical research relies on detailed administrative and transaction-level data, which typically become available for analysis only after careful processing and verification. As a result, even when researchers work with high-quality datasets such as customs records, the most recent data available for research may still lag behind current developments. Improving timely access to granular data would significantly enhance the ability of researchers to support policy discussions on issues such as tariffs, export controls, and other policy shocks.

It is also important to build substantial research capacity to translate these data into policy-relevant analysis. A larger group of researchers should support policy analysis while continuing to produce academic research. To attract leading researchers, it is important that the government allow academic publications based on work conducted in such a research environment. Strengthening this connection between data access, research capacity, and policy analysis highlights the value of establishing a comprehensive economic security think tank. With a strong research base at its core, such a think tank could also bring together academic expertise, industry knowledge, and ministerial policy experience to support rigorous and evidence-based policy analysis.

MUELLER:
I want to caution against moving too far toward government intervention. Proposals involving subsidies, equity stakes, or direct intervention risk resembling central planning. I grew up in East Germany and saw how decades of central planning damaged economic performance. Firms, especially private firms, should ultimately be responsible for securing their own value chains. If firms face risks from small or fragile suppliers, they can internalize those activities rather than relying on government support.

Micro level evidence shows that firms are far more flexible than aggregated data suggest. There are many substitution possibilities that governments do not understand. If governments begin organizing critical inputs, this creates distorted incentives, reduces innovation, and concentrates power, raising risks such as corruption.

The appropriate role for government is monitoring, risk assessment, and understanding production processes, not organizing private sector supply chains in concrete terms. Institutions such as RIETI can support policymaking by analyzing vulnerabilities and informing infrastructure and security decisions without directing firm behavior.

ITO:
Could imported products themselves be produced using Russian gas, circumventing sanctions?

MUELLER:
In this case, the answer is clear. Products such as ammonia imported from the United States are not produced using Russian gas. For rare materials, this issue can be more complex, but supplier exclusion and careful sourcing make it manageable.

Session 3: Analysis of supply chain vulnerabilities using world input-output tables and trade data

Session Chair:
TAMURA Akihiko (Senior Advisor, RIETI / Director General JETRO Paris)

Presentation 1

Richard BALDWIN (Non-Resident Fellow, RIETI / Professor of International Business, International Institute for Management Development (IMD))

There are two broad ways to analyze supply chains. One is the business approach associated with PORTER, which follows a single firm through its value chain. This approach is essential for highly specific products such as advanced semiconductors, rare earth magnets, or Russian gas, where firm identity, ownership, and product specificity matter. Its limitation is scale. As tiers expand, the number of suppliers grows exponentially, voracity of the information becomes questionable with conflicting incentives, and confidentiality constraints limit coverage beyond two or three tiers.

The second approach is the economic perspective using input-output tables, which represent the economy as a supply network rather than a single chain. This allows analysis of how all firms buy from and sell to one another and reveals vulnerabilities that are invisible when firms are asked only about their direct suppliers. Input-output analysis can be conducted on a value-added basis to show where work is performed, or on a gross production basis to show where physical production takes place. For disruption analysis, gross production measures are often more relevant.

The main limitations of input-output tables are data delays, aggregation, and simplified assumptions about substitutability. Despite these limits, they provide a necessary first-pass view of systemic exposure that detailed micro studies cannot deliver at scale. Customs data alone are insufficient because they blur intermediate and final goods and cannot identify true origins of production. Pharmaceutical supply chains illustrate this clearly, where surface trade data hide extreme dependence on Chinese inputs routed through third countries.

Using input-output tables, look-through measures show that exposure to China is far larger than face value trade data suggest. For the United States, look-through exposure to China is almost four times higher than direct exposure. Although face value dependence on China has declined since the Trump trade war, look-through dependence has continued to rise through rerouting via other countries.

The key message is that full decoupling is unrealistic. Global production structures cannot be unwound without enormous cost. Effective economic security requires identifying where vulnerabilities truly lie using look-through, network-based analysis, and then making selective, strategic choices rather than assuming broad disengagement is occurring.

Presentation 2

Angela GLOWACKI (Policy Analyst, Research Institute for Democracy, Society, and Emerging Technology (DSET))

China has long leveraged control over critical minerals and rare earths, but this position is the result of deliberate industrial strategy shifts since 2014 to 2015. With Made in China 2025 and subsequent five-year plans, China moved from infrastructure-led growth toward advanced technologies such as semiconductors, electric vehicles, and solar panels, while explicitly treating mineral supply chains as a national security issue.

This shift sharply increased demand for critical minerals. Although China lacks sufficient domestic supply, it dominates refining capacity, particularly for battery materials. Nickel illustrates this dynamic. China produces little nickel domestically, yet controls a disproportionate share of global refining. To secure inputs, China expanded foreign direct investment in resource-rich countries, especially Indonesia, following the launch of the Belt and Road Initiative.

Indonesia pursued its own strategy through export bans and downstream processing mandates, using incentives alongside restrictions. In response, Chinese firms invested directly in Indonesian industrial parks, embedding themselves across mining, smelting, and battery material production to circumnavigate such restrictions. Financing evolved from policy banks to state-owned commercial banks, enabling broader participation by battery and materials firms. The Morowali Industrial Park exemplifies this model, where localized processing aligns with Indonesian policy while maintaining structural dependence on Chinese capital, technology, and markets.

Despite gains in domestic processing, Indonesia remains heavily tied to China. Chinese firms hold roughly 75% of refining capacity, and over 90% of nickel products are exported to China. This creates asymmetric dependence with economic and political consequences, underscoring the need for coordinated action among like-minded countries.

Downstream, similar vulnerabilities appear in lithium-ion batteries for drones and defense applications. Taiwan has strong manufacturing capabilities but faces severe constraints sourcing non-Chinese battery-grade materials. Alternatives exist but are three to four times more expensive, and limited domestic demand discourages private investment.

The policy implication is clear. Securing supply chains requires upstream intervention, state-backed financing, clear critical minerals strategies, and international coordination. Japan’s experience through JOGMEC shows that targeted public investment can reduce dependence, but competition with state-backed rivals remains intense. Effective economic security will depend on coordinated standards, joint upstream investment, recycling, and stockpiling, supported by sustained public private cooperation.

Presentation 3

ISONO Ikumo (Head of the Economic Integration Research Group, Development Research Center, Institute of Developing Economies, Japan External Trade Organization (JETRO))

Geopolitical shocks are already delaying shipments, disrupting intermediate inputs, and sometimes reconfiguring supply chains. Simulation analysis helps in three ways. First, it translates theoretical assumptions into numerical predictions. Second, it estimates the magnitude of damage or policy impacts and allows comparison across options and priorities. Third, it can be applied quickly, without waiting for new trade statistics.

Our tool is the geographical simulation model, IDE-GSM, which combines a general equilibrium framework with geographic and transport cost data. It can generate prefecture level results and incorporates structural change, capturing how shocks shift procurement and sales patterns over time. We model tariffs, transport costs, and certain conflict related frictions, and we identify where disruptions occur and what alternative routes exist. The model covers many regions but only eight industries, reflecting data constraints. It also has limits: exchange rates are fixed, bilateral investment patterns are not fully captured, expectations and uncertainty effects are excluded, and Chinese overproduction dynamics are only partially represented.

Across past simulations of decoupling, sanctions, and trade wars, participants in sanctions generally incur costs, while neutral countries often gain through trade diversion. Stronger sanctions raise both costs for blocs and gains for neutral countries, making full isolation difficult.

For a reciprocal tariff scenario, our results show a GDP impact of minus 3.0% for the United States, minus 0.7% for China, and minus 0.8% globally. Countries facing relatively lower tariffs, such as Singapore, can gain, especially in electronics. The US loss reflects higher import prices for consumers, higher input costs for producers, and only partial offset from increased domestic sourcing. Impacts elsewhere depend on export exposure to the US and relative tariffs versus China, with some countries gaining through rerouting of intermediate inputs and final exports.

Border and route disruption simulations show concentrated losses where dependence is high, but potential gains where alternatives exist. The policy implication is to support switching, route diversification, and market diversification, including through trade agreements and frameworks that reduce adjustment costs.

Comment

KIM Byung-Yeon (Distinguished Professor in the Department of Economics, Seoul National University)

The first paper reaches a clear conclusion. Full decoupling from China by the United States is not possible. Using look-through trade measures rather than face-value data, the United States is effectively exposed to Chinese inputs at 3.8 times the level implied by direct trade statistics. This dependence is highly asymmetric, with US reliance on Chinese industrial outputs far exceeding China’s reliance on US inputs. Broad trade decoupling is therefore unrealistic.

This raises two key questions. First, how does this diagnosis change once substitutability is taken into account? If China is not dominant in physical production, can inputs be replaced by sourcing from other countries? Second, what does this imply for policy strategy? The United States has adopted a small yard, high fence approach, targeting only strategically important sectors. Evidence on emerging industries suggests that in areas such as semiconductors and artificial intelligence (AI), the combined capacity of the United States, Japan, Germany, and South Korea may exceed that of China, implying that targeted strategies may be more feasible than economy-wide decoupling. The central issue is what economic leverage the United States can realistically exercise in a context of hegemonic rivalry.

The second paper highlights China’s dominance in critical minerals and battery supply chains. The challenge is how like-minded countries can reduce asymmetric dependence, including whether alternative technologies or different sourcing strategies can mitigate risks in areas such as electric vehicle (EV) batteries and drones.

The third paper uses simulation analysis to quantify economic security risks. Its finding that tariffs are self-harming is important, but the estimated 3% GDP loss for the United States appears large. The model excludes exchange rates, interest rates, and investment responses, and may overstate effects by limiting trade diversion and relocation dynamics. Taken together, the papers are complementary, combining diagnosis of past dependence, analysis of critical constraints, and projections of future risks.

BALDWIN:
China produces about 35% of total world manufacturing output, more than the next eight countries combined. It is the world’s sole manufacturing superpower, and this dominance also applies to intermediate goods. Roughly half of China’s exports are now intermediate inputs, making it nearly impossible to manufacture anything globally without relying on Chinese inputs. Reducing China’s share from 35% to 25% would require an almost unimaginable economic effort in the short term. Full decoupling is therefore unrealistic. Governments have long protected farms and arms from pure market forces, and a limited number of additional sectors, such as semiconductors and medical supplies, may belong in this category. Targeted reduction of dependence in critical industries is feasible, but broad trade conflict is counterproductive and risks alienating allies.

GLOWACKI:
Korea and Japan hold strong positions in NMC batteries, which makes excessive Chinese investment influence a growing concern. Reducing asymmetric dependence requires increased investment from non-Chinese sources, particularly in countries that remain open to foreign direct investment. Battery technologies are still evolving, and it is too early to rule out alternatives such as solid-state or quasi-solid-state batteries. A two-tiered approach is needed, combining diversification of investment with continued exploration of new technologies, while recognizing that most battery systems still depend on critical minerals dominated by China.

ISONO:
The estimated 3% decline in United States GDP reflects real-side effects only and may be overstated. Exchange rate and monetary adjustments could offset part of the impact, as seen during the 2018 trade war. The model includes substitution toward lower-tariff imports and domestic production, which cushions the United States due to its large economy. China’s negative impact reflects reduced United States demand and trade diversion. Additional factors include cost absorption by firms and unusually strong investment conditions that the baseline scenario does not capture.

Session 4: China and East Asia

Session Chair:
Richard BALDWIN (Non-Resident Fellow, RIETI / Professor of International Business, International Institute for Management Development (IMD))

Presentation 1

ITO Asei (Associate Professor, Institute of Social Sciences, The University of Tokyo)

Chinese AI capabilities have advanced rapidly, and Chinese large language models (LLM) are now widely used outside China. These models perform well and are increasingly competitive with frontier systems developed in the United States.

China is now one of the two leading countries in AI model development, and Chinese firms such as Alibaba, Baidu, Tencent, Huawei, and several startups have released large language models at scale. A prominent example is Alibaba’s Qwen, an open-source model that can be freely downloaded. By midyear, more than 300 Qwen variants were available, with over 400 million downloads and more than 5,000 derivative models created through fine-tuning. This indicates global diffusion of Chinese AI technology.

At the same time, Chinese LLMs are subject to strict state regulation. Since August 2023, the Interim Measures for the Management of Generative AI require model outputs to adhere to socialist core values. In practice, this results in systematic censorship. Testing Chinese models with sensitive questions shows refusal or avoidance rates of around 40%, and in some advanced models over 60%. These patterns reflect how the models are trained.

The key issue is spillover beyond China. Open-source Chinese models can be used abroad, including in Japan. More importantly, derivative models created by foreign developers may inherit censorship features. In Japan, some derivative models based on Chinese foundations show refusal rates of up to 30%, indicating residual political filtering. Although current download numbers are limited, the presence of such models raises concerns about information control and narrative influence.

This matters because empirical research suggests that Japanese citizens can be influenced by illiberal narratives. If large language models subtly embed political messaging, they may shape opinions over time. These dynamics may also expand through initiatives such as an AI-enabled Digital Belt and Road.

The policy challenge is not limited to China. It concerns how political values travel through machine intelligence. Japan has begun responding through AI safety guidelines, assessment frameworks, and research initiatives. However, domestic measures may not be sufficient. International coordination will be necessary to address the cross-border implications of AI diffusion in an era of geopolitical competition.

Presentation 2

Joris TEER (Research Analyst for Economic Security and Technology, European Union Institute for Security Studies (EUISS), and Senior Advisor at the Chips Diplomacy Support Initiative (CHIPDIPLO))

In a globalized world, we preferred not to mine or refine rare earths because of environmental and social costs. Under great power competition, this logic no longer holds. Control over these inputs can be misused through refusal to supply, or lost entirely through conflict.

Critical raw materials form the skeleton of the global economy. Remove them, and pacemakers, drones, fighter jets, wind turbines, and energy systems fail. Semiconductors are the central nervous system, and chemicals are the connective tissue. For decades, countries focused on economic value added, keeping design while outsourcing production. With rising military risk around Taiwan and China’s weaponization of supply chains, strategic value added now matters more than economic value added.

Two threats dominate. The first is conflict in East Asia. China’s military expansion, especially its navy and missile forces, combined with its dominance in shipbuilding and manufacturing, raises the risk that war or blockade would halt production across Japan, Korea, Taiwan, and China. This region produces 75% to 80% of global semiconductors and dominates many critical raw materials. Taiwan imports over 95% of its energy. If energy flows stop, semiconductor fabs stop. Cascading effects would halt defense production, health care, digital infrastructure, and energy transition worldwide.

The second threat is unwillingness to supply. China has already imposed creative export controls on critical minerals. European firms report production disruptions even without military conflict. These controls target military end use, deny inputs to defense supply chains, extract sensitive information, and function as economic leverage to demand policy concessions. This is coercion, not trade.

China dominates the geoeconomic escalation ladder. Securing one input simply shifts pressure to the next, from minerals to batteries to manufacturing. De-risking alone is insufficient. Allies must combine supply-side diversification with demand-side measures and economic deterrence. Japan, the Netherlands, Korea, and the United States all control choke points. Without coordinated, credible countermeasures, coercion will continue.

Comment

Alex BRISTOW (Senior Analyst, Cyber, Technology and Security Program, Australian Strategic Policy Institute (ASPI))

I want to respond to Dr. ITO’s presentation by emphasizing that AI cannot be separated from China’s broader data extraction and propaganda system. Research shows that data scraped from smart cities, IoT devices, and online activity can be recycled into propaganda, coercion, and potentially hostile interventions. This makes AI exports a strategic issue, not just a technological one. I would also caution against complacency about Western leadership in AI. While some indices suggest the United States still leads, research trend data from ASPI’s Critical Technology Tracker indicate that China is already ahead across most AI subfields and is likely to overtake remaining Western advantages within five to ten years.

On Taiwan, I would shift focus away from invasion scenarios toward coercive pressure short of war. Taiwan is already under sustained subversive pressures through disinformation, economic pressure, and political interference. Research based on fieldwork and tabletop exercises, including ASPI’s Pressure Points initiative, suggests that Beijing would prioritize misinformation, social division, and targeted disruption to force accommodation without invasion. From Taiwan’s perspective, this is already a form of conflict.

Economic deterrence matters here. Taiwanese contingency planning may involve proactively shutting down semiconductor production during extreme coercion, not only for resilience but to globalize the crisis early and deny Beijing control over timing. Retaining critical capabilities in Taiwan preserves this deterrent effect.

Australia offers a cautionary case. While often cited as having resisted Chinese trade coercion, the real lesson is different. Pressure was mitigated largely by a functioning global trading system and China’s continued dependence on Australian iron ore. Beijing is likely to close those gaps next time. The key takeaway is that economic security depends not only on resilience and diversification, but also on preserving an open international trading system. Without it, resistance to coercion becomes far harder.

ITO:
China’s digital governance model creates a feedback loop in which authorities supply data and platform firms innovate domestically, but this same system constrains Chinese firms’ expansion into G7 markets. The larger concern lies in spillovers to the Global South. Officials from developing countries increasingly visit Chinese cities to study technology and governance solutions, raising the risk of China’s digital governance model diffusing globally. On technological leadership, China’s strength reflects a large pool of highly skilled young engineers and a system that performs efficiently when goals are clearly defined. On decoupling, this is not feasible for either side. China’s production capacity is strong, but domestic consumption is weak, and external demand remains essential. With the real estate bubble deflated, full decoupling would impose significant macroeconomic costs on China itself.

TEER:
Recent experience shows that coercion through control of critical inputs is more powerful than tariffs or import bans. China’s ability to restrict access to rare earths and related inputs has proven more disruptive than trade penalties. This episode revealed the limits of the “small yard, high fence” strategy, as firms faced production shutdowns and offshoring pressures. Even limited, incremental restrictions on critical materials have already caused major disruption. China is deliberately deepening global dependence on its manufacturing base, viewing this as economic deterrence, particularly in extreme scenarios such as a Taiwan crisis. It is stockpiling inputs, expanding energy and port infrastructure, and prioritizing self-reliance. The cost of inaction is continued concentration of manufacturing power in China. Partial, piecemeal responses help but are insufficient. What is required is allied scale: coordinated demand and policy action across trusted economies to rebuild supply across critical layers. Disruption can catalyze action, as Trump 2.0 has done, but durable solutions require cooperation rather than unilateral pressure.

Closing Remarks

TODO Yasuyuki (Program Director and Faculty Fellow, RIETI / Professor, Faculty of Political Science and Economics, Waseda University)

Today’s discussions reconfirmed that supply chain disruptions driven by geopolitical risks and protectionism are now the norm and must be managed continuously. We examined supply chain vulnerabilities from three perspectives: new analytical methods using granular data, networks, and simulation; the risks created by heavy dependence on China and the need for diversification and substitutability; and the importance of public-private and international cooperation. Strengthening economic security requires continuous analysis, clear messaging for policymakers, and sustained collaboration among governments, businesses, and researchers.

TOMIURA Eiichi (President, RIETI / Dean, Faculty of Data Science, Otsuma Women's University)

Thank you very much for participating in the RIETI, JOGMEC, IDE-JETRO co-hosted symposium. On behalf of the hosts, I extend my sincere gratitude. Today’s discussions highlighted three shared foundations between economics and economic security: the central importance of high-quality data, the necessity of international collaboration among like-minded countries, and the critical role of public-private cooperation. These elements are essential for strengthening supply chain resilience and national security. I thank all presenters, participants, and organizers, and hope today’s dialogue contributes to future policymaking and corporate action.