Domestic Substitution versus Regional Diversification: A Strictly Matched Counterfactual Analysis of China's Manufacturing Supply Chains

In the modern world, the things we buy are rarely made in a single place. A smartphone, a car, or even a simple shirt is the result of a vast, invisible web where parts travel across borders, often passing through several countries before assembly is complete. This global network makes production efficient and affordable, but it also creates a hidden vulnerability: if one link in the chain breaks due to a natural disaster, a trade dispute, or a political conflict, the entire system can stall. When a factory in one country cannot get a specific component from abroad, it stops producing, and that stoppage ripples outward, affecting industries and economies far removed from the original problem.

To protect against these disruptions, governments and companies are currently debating two main strategies. The first is to bring production back home, replacing foreign parts with those made domestically. The second is to spread the risk by buying the same parts from many different countries instead of relying on just one or two big suppliers. Intuitively, both approaches seem like they should make supply chains safer. However, measuring which one actually works better is difficult because the web of connections is so complex. A change in one place can trigger unexpected reactions elsewhere in the network, meaning that simply looking at direct imports tells only part of the story.

A new study by researchers Yujie Guo and Junwen Cao from the Jiangxi University of Finance and Economics tackles this question by simulating what would happen if China's manufacturing industries switched their sourcing strategies. The researchers focused on nineteen different manufacturing sectors, ranging from chemicals and electronics to motor vehicles and machinery. They used detailed data from 2016 to 2022 to map out the entire web of production requirements for each industry, tracing not just the parts bought directly, but also the materials those parts needed, and the materials needed to make those materials, going back through the entire supply chain.

To ensure a fair comparison, the researchers designed a strict test. They did not simply compare two different industries or two different years. Instead, they took a specific situation where a Chinese industry was buying a certain amount of a specific part from a specific foreign country. They then created two identical scenarios for that exact same situation. In the first scenario, they simulated what would happen if that foreign supply were replaced by a domestic Chinese supplier. In the second scenario, they simulated what would happen if that same foreign supply were replaced by suppliers from other countries in the same region. In both cases, the amount of the part being replaced was exactly the same, and the original source was exactly the same. The only difference was where the new supply came from. This method allowed them to isolate the effect of the destination of the replacement supply, removing other variables that often confuse such comparisons.

The results of these simulations revealed a clear and somewhat surprising distinction between the two strategies. When the researchers looked at the total amount of foreign production required to make a unit of a product, domestic substitution consistently reduced this total exposure. In every single one of the 399 comparisons they ran, replacing foreign inputs with domestic ones lowered the overall reliance on foreign production. This makes sense, as bringing the work home naturally reduces the need for cross-border materials.

However, the story changed when they looked at how the remaining foreign supply was distributed. The goal of regional diversification is to spread purchases across many different countries so that no single country holds too much power. The simulations showed that regional diversification did succeed in making the distribution of sources more balanced in many cases. In 242 out of the 399 comparisons, spreading the purchases among regional neighbors reduced the concentration of supply better than bringing it home did. But this advantage was not absolute; in the other 157 cases, domestic substitution actually did a better job of balancing the sources. Furthermore, as the researchers increased the intensity of the change—simulating a larger shift in sourcing—the advantage of diversification became less frequent.

Perhaps the most critical finding was that making the distribution of sources more balanced does not necessarily mean reducing the total amount of foreign dependence. In fact, under the regional diversification strategy, the total exposure to foreign production actually increased in 122 out of 133 specific industry and year observations, even though the concentration of sources went down. This happened because the new foreign suppliers introduced their own dependencies. When a Chinese company switched from buying a part from Country A to buying it from Country B, Country B might have needed to import materials from Country A or other foreign nations to make that part. These indirect requirements traveled through the network, adding new layers of foreign dependence that were not visible if one only looked at the direct purchase.

The study also examined the risk of relying on the single largest foreign source. Domestic substitution consistently lowered the exposure to the largest single source in almost every case. Regional diversification, while successful in lowering the overall concentration index in many instances, did not always lower the exposure to the biggest single source. In some simulations, reducing dependence on the original largest supplier simply shifted the heavy reliance to a different country. For example, in the computer and electronics sector, diversifying away from the original top supplier resulted in a new country becoming the largest source, with a level of dependence that was actually higher than the original baseline, even though the overall spread of suppliers looked more balanced.

The researchers emphasized that their work measures the structural shape of the supply chain, not the likelihood of a disaster or the cost of making these changes. They did not calculate how much it would cost a company to switch suppliers or how quickly a factory could ramp up production. They also did not predict whether a specific crisis would happen. Instead, they provided a clear map of how different sourcing choices reshape the web of dependencies. Their work suggests that there is no single strategy that improves every aspect of supply chain security at once. If the primary goal is to reduce the total volume of foreign production required, bringing production home appears more effective in this simulation. If the goal is to avoid relying too heavily on any one country, spreading purchases around can help, but it may come with the cost of increasing the total amount of foreign involvement in the production process.

Ultimately, the study highlights that supply chain resilience is not a simple choice between two options. It is a trade-off between different types of risk. Reducing the total amount of foreign input is different from spreading that input across many sources, and one does not automatically guarantee the other. Policymakers and business leaders looking to strengthen their supply chains must decide which dimension of risk matters most to them: the sheer volume of foreign dependence, the concentration of that dependence, or the exposure to the largest single source. The simulations show that these goals can pull in opposite directions, and a strategy that looks good on one measure might look worse on another. The most effective approach likely requires a careful, industry-specific balance rather than a one-size-fits-all solution.