Debate: “Will the COVID-19 pandemic reinforce preexisting trends that in turn lead to reshoring or other forms of GVC restructuring – and what does it imply for policymakers?”

Dalia Marin & Caroline Freund
May, 2024
#Trade and FDI

Debate between Dalia Marin and Caroline Freund

Dalia Marin is Professor of International Economics at TUM School of Management, Technical University of Munich. Before joining TU Munich she was Professor of International Economics at Ludwig-Maximilians University of Munich, Associate Professor at Humboldt University Berlin, and Assistant Professor at the Institute for Advanced Studies, Vienna. Marin is a Senior Research Fellow at BRUEGEL, Brussels, Fellow at the European Economic Association, and Research Fellow at the Centre for Economic Policy Research (CEPR). She published in leading academic journals and acted as consultant for the European Commission and International Organizations.

Caroline Freund is Dean of the UC San Diego School of Global Policy and Strategy (GPS) and an expert in international trade and economic development. Prior to joining GPS, she served as global director of Trade, Investment and Competitiveness at the World Bank, where she was co-director of the flagship World Development Report 2020 on Global Value Chains. Freund also served as a senior fellow at the Peterson Institute for International Economics. Her work has appeared in leading academic journals.

The COVID-19 pandemic has highlighted the vulnerability of global supply chains. In a survey among high-level executives in 1,181 companies in the US and four European economies, Euler Hermes (2020) found that “almost all (94%) companies surveyed reported a COVID-19 induced disruption to their supply chains.” As a response to such disruptions, companies consider advancing automation to become less dependent on workers; some are rethinking their sourcing strategies, favouring close-by suppliers and diversifying sources. Some researchers therefore expect global production organization to change, whereas others argue that high initial fixed cost of global sourcing and production prevent firms from rigorously adjusting existing sourcing practices. What is the evidence, and what does it imply for policymakers? Dalia Marin argues that the uncertainty shock induced by the COVID-19 pandemic increased cost of GVCs. At the same time, cost of automation decreased and the pandemic is expected to further accelerate this trend. As a result, reshoring and restructuring of global supply chains is likely to occur. In contrast, Caroline Freund argues that the COVID-19 pandemic will not reshape GVCs substantially. Firms are expected to increase resilience through dual sourcing strategies more often, however, reshoring is too costly. Furthermore, she reasons that automation will not be reinforced by the pandemic and will not lead to restructuring, rather it could be trade enhancing.

Dalia Marin: Uncertainty Changes all: How Supply Chains change with COVID

The COVID pandemic is an unprecedented uncertainty shock. We use the rise of uncertainty in the financial crisis to evaluate how global supply chains (GVCs) will evolve with the pandemic. I then ask whether the government needs to step in to help firms to navigate in the crisis.

With the fall of communism in 1989 and the entry of China into the WTO in 2001 major markets with low labor costs entered the world economy. Firms in high income countries started to produce in GVCs, relocating part of production to these regions to save on labor costs. Moreover, a revolution in the transport sector - containerization  - lowered transport costs making offshoring very profitable.  As a result, GVCs exploded in the hyper-globalization period 1990-2008. According to estimates GVCs accounted for 60% of world trade. But since the financial crisis 2008 GVCs have stopped to grow.

Why have GVCs stopped to grow? The financial crisis changed the relative costs of GVCs and robots. The increase in uncertainty in the financial crisis made GVCs more costly with the increased risk of a non-delivery of an input good. Uncertainty rose after the financial crisis until the Euro debt crisis by over 200% as indicated by the World Uncertainty Index (WUI). WUI developed by Ahir et al (2018) counts the frequency of the word uncertain or variants in EIU country reports.

At the same time the cost of financing a robot relative to hourly wages declined sharply (by over 100%) favoring the adoption of robots. As a result, firms in high income countries reshored production back to their home market and invested in robots instead. We find that after the financial crisis GVCs and robots became substitutes. The more robot intensive a sector is the less it engages in GVCs (Kemal and Marin 2020).

COVID accelerates this trend and is likely to lead to deglobalization. We expect that COVID will reduce GVC participation by 35% and increase robot adoption by 76%. The calculation assumes that in the COVID pandemic the WUI increases by 300% (the first SARS1 in 2002 epidemic increased the WUI index by 70%) and lowers the ratio between interest rates to hourly wages by 30%. The estimated growth of robot adoption of 76% is on the high end since it does not take into account that uncertainty also reduces investment and robot adoption.

Rising transport costs are likely to accelerate the shift away from GVCs. During the pandemic the cost of containers used to ship goods from Asia to Europe and the United States has risen nearly tenfold (Drewry 2022), and transport workers, facing increasingly harsh working conditions, have been leaving their jobs. It remains to be seen whether the turmoil in the transport sector with supply-chain bottlenecks is transitory or persists for longer.

Some might argue that rich country governments’ effort to strengthen domestic and regional production networks reflects new form of economic nationalism driven by fear of China. But the crucial question is whether companies really need state help to protect themselves against supply chain turbulence.

There are three ways advanced economy firms can make their input supplies more resilient and only one of them requires government involvement. One option is to take control and reshore production from developing countries. A second way to insure against supply chain shocks is to build inventories and to switch from “just in time” production to a “just in case” model. Third, companies can dual-source or triple-source inputs, relying on suppliers from different continents in order to hedge the risk of natural disasters and other regional disruptions.

But the third strategy, diversification of input sources, has its limits. For example, a highly specialized supplier that invests in research and development in order to provide a specific input is not easily replaceable, and sourcing others can be costly. Heavy regional concentration of suppliers also make diversification difficult. Most producers of chips, battery cells, rare earth materials such as cobalt and lithium, and pharmaceutical ingredients are based in Asia. Geographic clustering of input suppliers can generate upheavals in the rest of the world, as the current global semiconductor shortage illustrate. In a 2012 paper, MIT’s Daron Acemoglu and his coauthors showed that disruptions to an asymmetric supply chain network – in which one or a few suppliers deliver inputs to many producers – can spread throughout the world economy and potentially lead to a global recession. That supply chain disruptions can have economy-wide effects have been recently shown in empirical studies of the 2011 Great East Japan Earthquake (Carvalho et al 2021) and of three decades of major natural disasters in the US (Barrot and Sauvagnat 2016).

In such cases, governments can play a useful role by helping to provide firms with more potential alternative suppliers. Governments in the US and EU can ensure that a sufficient number of suppliers are available in both Europe and North America to hedge against the risk of disruption.
Caroline Freund replies to Dalia Marin: COVID won’t reshape supply chains significantly

In the 1990s and early 2000s, the costs of international trade fell and global value chains multiplied. Trade costs fell because of new mega trade agreements, WTO formation and expansion, EU deepening and expansion, as well as unilateral tariff reduction in developing countries.  The drop in trade costs made it profitable to offshore stages of production and factories began increasingly operating across borders. New communications technologies, such as the internet and e-marketplaces, also supported the formation of global value chains because complex, fragmented production could be seamlessly coordinated and finding new suppliers was easier. This period became known for expanding global value chains (GVCs), and by the mid-2000s more than half of the value of trade was crossing more than one border (World Bank 2020).

After the financial crisis, it was economic fundamentals that held back supply chains, not uncertainty, as Marin argues. Global income growth was tepid which depressed trade and removed some of the incentives to expand supply chains.  Moreover, there were no major liberalization initiatives or changes in technology to spur another round of GVC expansion. Global value chains stagnated; they remained largely intact, with many parts and components crossing borders, but they stopped expanding.

Then COVID happened. Goods trade initially plummeted and then quickly recovered, with exceptionally strong performance in 2021.

As Dalia Marin writes, COVID has been an uncertainty shock, but given the strong trade performance during the crisis, I disagree that it will lead to a major shift in GVCs. COVID triggered both supply and demand uncertainty.  Supply has become less predictable because of periodic, geographically concentrated labor shortages or port closures. But by far the biggest effect has been on demand.  In the initial months of COVID, demand for most goods plummeted, as workers lost income and retreated to their homes. Later, government stimulus programs kicked in and consumers accrued savings from weak spending on services, such as travel, restaurants, and gym memberships, and demand for goods exploded.

The result has been a surge in international trade in goods, as consumers gobbled up electronics, home office furniture, stationary bikes, etc. The surge in consumer demand was unexpected and many firms had mistakenly cancelled orders of inputs, and were now seeking to expand and build inventories, so demand surged even more. But, even in absence of supply stoppages, in the short run, capacity is effectively fixed for many goods and the sudden sharp rise in demand was simply unmeetable. For all the discussion of ships stuck in LA port and Long Beach, container traffic was up 16 percent in 2021 as compared with the previous record haul in 2018.

It is not surprising that supply has been unable to keep up with demand, even as we embark on the third year of the pandemic. Economic theory is clear that the optimal investment response to a temporary and highly variable demand shock is to wait for more information. Indeed, firm surveys show that the vast majority of firms are largely sitting on the sidelines, waiting to make major investment decisions--though on balance they have shifted from a negative investment outlook at the onset of the crisis to a positive one now (Abhishek, Kusek, Albertson 2021).  Given that firms do not know if goods demand will remain strong, it does not make sense to invest in greater capacity that could become excess capacity in future.

Like Marin suggests firms can moderate supply problems with dual sourcing.  I agree that this is likely to happen more often and will help resilience, but dual sourcing is unlikely to cause a major shift in supply chains.  Expanding diversification beyond dual sourcing, however, is too costly because of the need to develop relationships, meet customization standards requirements, and benefit from scale economies.

What about reshoring or nearshoring? The benefits from sourcing in low cost countries is simply too great to lead to much reshoring or nearshoring. Moreover, reshoring is a terrible way to reduce risk since it reduces the scope to maintain production when shocks are local.  The old adage “don’t put all your eggs in one basket” and all that.

One way to examine how firms behave in response to increased risk is to look at what happened following the earthquake in Japan in 2011.  In a recent paper, we examine the change in sourcing over the long run owing to the supply shock (Freund et al. 2021). We find that importers did seek alternate suppliers, but they tended to find other large, low cost suppliers that could produce at scale. They did not diversify, reshore or nearshore.  In other words, that shock (which unlike COVID, destroyed capital) simply accelerated shifts that were already underway (Freund et al. 2021).

Contrary to Marin, I am not convinced that automation/adoption of robots and supply chains are substitutes.  As shown in the World Bank (2020), the most automated industry—automobiles is also the most intense user of cross-border supply chains.  In fact, automation and GVC production tend to go hand in hand. The reason is that automating one part of production lowers costs and allows firms to produce and sell more output. This scale effect results in more demand for imported inputs that are not automatable.  In related work, we examine the adoption of 3D printing technology for hearing aid production (Freund, Mulabdic, Ruta 2019). Similarly, we find that the shift to the new technology expanded trade, as the process remained complex and subject to returns to scale.  Unlike conventional wisdom, which predicted a shift away from imports, trade actually surged following the adoption of 3D printing!

I think any future reorganization of supply chains is more closely related to geopolitics than to COVID-induced resilience planning or the adoption of new technologies. US-China trade tensions could lead to a world where supply chains are carved out by political alliances. Reorganization along these axes can happen because of export controls, import protection, sanctions, and discriminatory investment practices—all of which are currently in place for some goods.  If government policy continues to encourage such “allied” supply chains, trade will not decline but will be reshaped.  The medium-term economic gains from trade and innovation will be lower, in exchange for the well-intentioned, complex goal of expanding long-run economic and political security.
Dalia Marin: Reply to Caroline Freund

I would like to focus my response to Caroline Freund on one point: Are supply chains and robots substitutes or complements.

Whether supply chains and robots are substitutes or complements is an empirical question. If the adoption of robots lowers costs, firms become more competitive and produce more. They will thus import more intermediate inputs from developing countries as is pointed out by Caroline Freund. However, we find that firms in high income countries which are already sufficiently endowed with robots reshore production from the developing countries. For these firms the share of labor costs in total costs is already low. They can then relocate production back to high wage countries and employ robots instead of expensive workers. This is a profitable option in particular when the delivery of the input from the developing country becomes uncertain due to a pandemic, a natural disaster or due to geopolitical risks.  This is the reason why we find that supply chains and robots became substitutes after the financial crisis (when uncertainty increased substantially) while they were complements (although hardly statistically significant) before the financial crisis.  Our analyses covers all high income countries (except Japan) and most developing countries including China. Automobiles are no exception. When we run robustness checks in which we exclude individual industries from the analysis, our reshoring results do not change when we exclude or include the car sector.
Caroline Freund: Reply to Dalia Marin

The period before the financial crisis and the period after it were very different.  Before the crisis, trade costs fell sharply, global value chains expanded, and income growth was robust in most of the world.  The period after the crisis was characterized by stagnation on all fronts. In contrast, robot adoption has been increasing steadily throughout both periods. The correlation between robot adoption and GVC formation, which Dahlia Marin argues changes over time, is not necessarily the result of a causal relationship.  An alternative explanation is that the cost of robots fell throughout the period, raising their use; while trade costs first fell and later stagnated, leading to a slowdown in GVC formation.

I have not seen any compelling evidence of extensive reshoring in advanced countries, nor of reshoring related to automation.  Although there is a lot of hype around reshoring, the evidence tends to be anecdotal. Kearney’s reshoring index has been positive for the US in only two of the last 12 years—implying that offshoring remained dominant. A studyof Spanish firms finds that automating tends to precede more intense importing from, or opening affiliates in, lower income countries, suggesting robots and GVCs are complements.

By definition, robots directly replace some workers.  The three pertinent questions are: (i) Do robots in advanced countries primarily replace domestic workers or replace workers in developing countries? (ii) What types of new jobs are created to complement the automation? And (iii) Where are these new jobs created?  One thing Dahlia and I can probably both agree on is that more research is needed to answer these questions.
Abhishek S., Kusek, P. & Albertson, M. (2021). World Bank Investor Confidence Survey: Evidence from the Quarterly Global Multinational Enterprises Pulse Survey for the Second Quarter of 2021, World Bank, Washington, DC.

Acemoglu, D., V. Carvalho, A. Ozdaglar, A. Tahbaz-Salehi (2012). The Network Origins of Aggregate Fluctuations, Econometrica, 80(5), p. 1977-2016.

Ahir, H., N. Bloom, D. Furcery (2018). World Uncertainty Index, Stanford University.

Barrot, J.N. and Sauvagnat, J. (2016). Input Specificity and the Propagation of Idiosyncratic Shocks in Production Networks, Quarterly Journal of Economics, 131(3), p 1543-1592.

Carvalho, V.M. et al. (2021). Supply Chains Disruptions: Evidence from the Great East Japan Earthquake, Quarterly Journal o Economics, 136(2), p 1255-1321.

Drewry (2022). World Container Index - 03 Feb 2022,

Faber, M., Kilic, K. and Marin, D. (2022). Uncertainty, Robots, and Supply Chains, Technical University of Munich, Mimeo.

Freund, C., Mulabdic, A. & Ruta, M. (2019). Is 3D Printing a Threat to Global Trade? The Trade Effects You Didn’t Hear About, Policy Research Working Paper, No. 9024. World Bank, Washington, DC.

Freund, C., Mattoo, A.; Mulabdic, A. & Ruta, M. (2021). Natural Disasters and the Reshaping of Global Value Chains, Policy Research Working Paper, No. 9719. World Bank, Washington, DC. License: CC BY 3.0 IGO.

Kilic, K., and Marin, D. (2020). How COVID-19 is transforming the World Economy, VoxEU, May.

World Bank (2020). World Development Report 2020 – Trading for Development in the Age of Global Value Chains,
Scroll to Top