Global Supply Chains and Productivity (Summary)

Opening Remarks

TOMIURA Eiichi (President, Chief Research Officer (CRO), Chief EBPM Officer, RIETI / Professor, Otsuma Women's University)

The World KLEMS Initiative, started by the late Professor Dale JORGENSON, aims to build a global productivity database for international comparisons. Every two years, it hosts an international conference, and RIETI is proud to host the 8th World KLEMS Tokyo Conference, with the last one held in 2014. We also thank Gakushuin University, Hitotsubashi University, and the Japan Productivity Center for their support.

This panel session, focusing on global supply chains and productivity, is open to the public. It will include a keynote by Professor Richard BALDWIN and presentations by Professors Shujiro URATA and Kyoji FUKAO, and Dr. Satoshi INOMATA. Thank you again for your participation.

Keynote Speech

Richard BALDWIN (Non-Resident Fellow, RIETI / Professor of International Economics, IMD Lausanne)

What I want to do in this talk was explain how supply chains are evolving—not just due to geopolitics, but primarily because of technology. The basic economics of trade and productivity are well understood: global value chains (GVCs) improve productivity by allowing specialization, exploiting scale, and sharing know-how. But the story now is about how those same GVCs are changing shape.

The Rise—and Role—of China in Global Supply Chains

Over the past two decades, we saw a massive shift in manufacturing from the G7 to a few emerging economies—most notably China. China now dominates global manufacturing by every metric: gross output, exports, intermediates, and more. You simply cannot talk about global supply chains without special-casing China. And it is not just final goods; it intermediates that matter. All major manufacturing economies have grown increasingly reliant on Chinese industrial inputs, while China has reduced its reliance on them.

From Fragmentation to Relocalization: Technology’s Real Impact

But manufacturing trade peaked around 2008. People often attribute that to geopolitics, but I argue it is mostly technological. First, ICT enabled fragmentation—it lowered coordination costs, so it made sense to split production across many locations. But now, automation is reducing the benefits of fragmentation. It puts multiple tasks into a single machine, shifting production back into fewer stages, and often back home. So, we are seeing both defragmentation and relocalization—a retrenchment of supply chains that began around 2012, well before Brexit, Trump, or US-China tensions.

Relocalization is not just reshoring—it is also regionalization. The data shows a clear trend: countries like Germany, Japan, and the U.S. are sourcing fewer intermediates globally and more from themselves or nearby regions. China, meanwhile, has massively repatriated its own supply chains.

Services: The Future of Trade and Development

While manufacturing trade is flattening, services trade is booming. Digitally enabled services—what I call international remote work—are growing rapidly. Services now account for about half of export-related jobs in the G7. The barriers to exporting services are technological, not regulatory, and those are falling fast.

This “Globotics Upheaval,” as I call it, will bring a tidal wave of talent from emerging markets. Emerging economies are already seeing fast growth in service exports, and the era of manufacturing‑led development is giving way to a new, service‑led structural transformation.

Panel Discussion

Panelists:

Richard BALDWIN (Non-Resident Fellow, RIETI / Professor of International Economics, IMD Lausanne)

INOMATA Satoshi (Chief Senior Researcher, Institute of Developing Economies, JETRO)

URATA Shujiro (Distinguished Senior Fellow (specially appointed) and Chairman Emeritus, RIETI / Professor Emeritus, Waseda University / Senior Research Fellow, Economic Research Institute for ASEAN and East Asia (ERIA))

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

Moderator:

INUI Tomohiko (Faculty Fellow, RIETI / Faculty of international Social Sciences, Gakushuin University)

Presentation 1

INOMATA Satoshi (Chief Senior Researcher, Institute of Developing Economies, JETRO)

Recent global supply chain dynamics highlight the tension between efficiency and risk, particularly as supply chains become more concentrated in certain regions. This concentration can create vulnerabilities, turning production hubs into chokepoints during disruptions. Events such as the global financial crisis and the Great East Japan Earthquake illustrate how global interdependencies make systems susceptible to a single point of failure.

To assess these risks, two indicators are introduced: trade-in value-added (TiVA) and pass-through frequency (PTF). TiVA measures concentration risks in terms of volume by showing how much value a product sources from specific countries, effectively indicating supply chain dependence on the country of concern. PTF measures supply chain concentration in terms of frequency, tracking how often a supply chain passes through high-risk countries along production processes.

A concentration analysis of the U.S. and China’s supply chains reveals a dramatic shift in their bilateral production structure from 1995 to 2018. In 1995, U.S. supply chains were only concentrated in China for the textile and leather products sector. However, by 2018, this concentration had increased across nearly all industries, including critical sectors like ICT and automobiles, while China’s supply chain concentration in the US remained almost unchanged. The US became asymmetrically dependent on China during this period.

The new indicator, pass-through frequency (PTF), was recently integrated into the OECD official statistics, allowing analyses across 45 industries and 77 economies over 26 years, providing a deeper understanding of global supply chain vulnerabilities.

Presentation 2

URATA Shujiro (Distinguished Senior Fellow (specially appointed) and Chairman Emeritus, RIETI / Professor Emeritus, Waseda University / Senior Research Fellow, Economic Research Institute for ASEAN, and East Asia (ERIA))

Today, I am here to introduce my work, which examines whether firms involved in global value chains can improve productivity, using Japanese manufacturing firm data. The share of overseas affiliate sales in parent firm sales grew significantly from 8.7% in 1985 to 37.2% in 2019, indicating increasing GVC participation by Japanese firms. GVC firms, defined as those engaged in both importing and exporting, can potentially benefit from gaining knowledge, technology, and access to high-quality intermediate inputs, which may lead to higher productivity.

Using propensity score matching and difference-in-differences (DID) methodologies, the study compares GVC-involved firms with non-GVC firms. Results show that GVC firms have higher productivity, measured by total factor productivity (TFP), compared to non-GVC firms. Moreover, the productivity benefits from GVC participation tend to increase over time, suggesting that firms gradually learn and benefit from their involvement in GVCs.

The analysis found that firms with higher initial productivity are more likely to participate in GVCs, and once involved, their productivity tends to improve. This "learning by participating in GVC" effect strengthens over time. A robustness check confirmed these results for firms consistently engaged in GVCs over five years.

The study suggests that policies supporting non-GVC participating firms to improve productivity and enter GVCs could help those non-GVC participating firms to realize productivity gains through GVC involvement.

Presentation 3

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

Our research, conducted at the Policy Research Institute within Japan's Ministry of Finance, quantifies offshoring bias in productivity estimates using firm-level micro-customs data. The study addresses the impact of GVC participation, especially in relation to imports of intermediate goods, and its potential to overestimate productivity due to the decline in prices of imported inputs, which can bias traditional TFP measures.

The study builds on previous work by DIEWERT and NAKAMURA, as well as HOUSEMAN and others, who highlighted the issue of offshoring bias. However, this research is the first to apply micro-customs data to directly estimate this bias, revealing that firms that import cheap intermediate inputs from countries like China may appear to be more productive than they truly are if the price decline of those inputs is not accounted for.

To estimate offshoring bias, the study uses two deflators for intermediate inputs: one for total intermediate goods and another separating imported from domestic inputs. The difference in TFP growth rates between these two estimates is defined as offshoring bias. Results show that firms with higher import shares tend to experience greater offshoring bias, though this effect is smaller for firms importing from affiliates.

The findings suggest that offshoring bias increases with the import share of intermediate goods but decreases when imports come from affiliates. Additionally, imports from China appear to slightly magnify the bias. To our knowledge, this study is the first to quantify offshoring bias using firm-level customs data.

Discussion

BALDWIN:
Professor Fukao, can the bias in TFP be interpreted as China's contribution to Japan’s TFP, given that importing cheaper intermediates from China boosts Japan’s productivity? Mr. Inomata, your method uses input-output tables, which do not define production sequences. How can you address frequency or sequence in this context?

FUKAO:
At the firm level, China’s TFP growth lowers production costs, which can be mistaken for Japanese TFP growth. At the macro level, input deflators already reflect these price changes in terms of trade and welfare improvements, and not productivity growth.

INOMATA:
The Leontief inverse matrix, which is derived from an input-output table, can be decomposed using a Taylor series, where each layer represents a different stage of production propagation sequence. Therefore, the input-output analysis does allow us to trace production sequences. My question to Prof. Baldwin is that, with increasing automation in manufacturing sector, do less-developed economies now have little chance to industrialize by offering cheap labor to the global economy, as China has done in the past?

BALDWIN:
The idea of developing through labor-intensive manufacturing is no longer viable. China’s dominance and recent automation have closed that path. The shift has fundamentally changed how economies can develop, with no alternative but to focus on higher-tech sectors.

URATA:
Why did the IT revolution and defragmentation occur around 2008? I believe it was tied to China’s shift from an outward-oriented to inward-oriented strategy around the financial crisis, but please provide any further information. Also, do data protection and national security policies slow digital trade and service-oriented economic growth?

BALDWIN:
China’s shift was pivotal, changing global value chains. The 2008 global financial crisis marked a peak in world trade-to-GDP ratios, but the picture varies by country and sector. On digital trade, privacy and security policies can have an impact, but digital trade is less vulnerable to geopolitical disruptions than goods trade. Geostrategic conflicts, like U.S.-China tensions, focus more on manufacturing than services.

FUKAO:
Professor Baldwin, you argue that IT causes defragmentation, but how does this occur? Technologies like GPT make global collaboration easier. Why does IT reduce fragmentation in manufacturing?

Secondly, to both Mr. Inomata and Professor Baldwin: with random tariffs like those being implemented under the TRUMP administration, fragmentation and offshoring seem to be declining. Will the pass-through frequency that Mr. Inomata mentioned drop? Professor Baldwin, how might this affect your outlook?

INOMATA:
It depends on the industry. ICT equipment sector may be quite vulnerable because its supply chain is extremely complex and crosses national borders multiple times, cumulatively affected by the imposition of tariffs.

BALDWIN:
Trade policy is more chaotic than I have ever seen. Even U.S. officials are not sure of what is coming. During the first Trump administration, tariffs ended quietly and were forgotten, but this round may have lasting effects. Firms are already reacting, not just to tariffs but also to COVID, geopolitics, and other shocks. We are seeing reshoring and regionalization. If this continues, fragmentation will reverse—leading to less variety, lower quality, and higher prices.

Q1:
For Professor Baldwin, if you were president of the U.S. with limited resources, should the U.S. focus on developing services or reshoring manufacturing?

BALDWIN:
Most countries, including the U.S. and China, overemphasize manufacturing. It is a long-term trend that manufacturing is becoming less important in terms of employment and value added. While manufacturing is critical for national security, it is not the future of the economy, especially for advanced nations like the U.S. Developing countries betting on manufacturing infrastructure may be wrong unless they are already integral parts of the GVCs. The U.S. already has a comparative advantage in services and runs a surplus with nearly every country, so the focus should be on services, and not reshoring manufacturing.

Q 2:
How will President Trump’s tariff measures affect the benefits of global value chains, based on your theories or assumptions?

INOMATA:
I believe I already addressed this earlier.

URATA:
Professor Baldwin already responded to this, noting that Trump’s tariffs promote reshoring and onshoring. From my perspective, this policy is highly inefficient and is having and will have negative impacts on not only the U.S. but the entire world. While there may be some short-term gains for the U.S. through increased investment, the medium- to long-term effects will be detrimental—even for the U.S.

FUKAO:
I completely agree with the previous points. GVCs provide benefits through lower product prices and improved terms of trade. If Trump’s tariff policies continue, we will likely see a global rise in inefficiency and output prices. The U.S. itself could face severe inflation if this path continues.

URATA:
Should Japan retaliate against Trump's tariffs, considering Europe and China have done so?

BALDWIN:
Retaliation is important to uphold WTO rules and avoid escalation. However, responses should be measured and de-escalatory to avoid being overly disruptive. Japan did not retaliate in the first Trump administration, and retaliation should be avoided if possible..

URATA:
Japan cannot retaliate without changing foreign exchange laws, though there is some support for revising the law to allow for retaliation.

BALDWIN:
If it is illegal, then retaliation should obviously not be pursued.

INOMATA:
Retaliation is a short-term response; long-term focus should be on the development of deterrence powers, like enhancing Japanese economy’s strategic indispensability and collective economic security measures.

Q3:
Regarding the production of automation and other IT technologies that promote localized manufacturing, do you think the producer market will be monopolized by U.S. tech giants, or fragmented among various nations, like GVC manufacturing?

BALDWIN:
Unless labor is a key factor in specific products, production will likely be spread out, reducing the extreme centralization that we see in semiconductors or similar supply chains. Fragmentation, rather than monopolization, seems to be the trend.

Q4:
Given that services have less economies of scale than manufacturing, to what extent could growth in service exports lead to productivity growth?

FUKAO:
I think with the improvement in digital technology and AI, the productivity growth in the service sector will be quite large. Whether it will occur within the service industries themselves or in ICT supplier companies is hard to say.

INUI:
I would like to close the session and thank the panelists and participants. On behalf of the organizing committee, we are grateful for your active participation and the excellent papers presented. A special thank you to Professor Bart van Ark for your strong leadership in making this a successful conference.