Tariffs Can Move Trade. Building Manufacturing Takes Something Else.

1. Guest Background

The central question in this roughly 46-minute episode of Decoder is what must happen between a tariff announcement and the creation of an American factory. Host Nilay Patel presses the political promise: restricting imports is supposed to restore production and employment. His guest, Harvard Business School professor of management practice Willie Shih, examines the production systems that would have to change for that promise to become real.

The supplied guest background identifies Shih as an economist, professor, and author of Where Does My Stuff Come From: The Powerful Economics Behind Global Supply Chains. The transcript renders his name inconsistently; this article uses Willie Shih throughout. His book’s subject gives the conversation its organizing perspective: products arrive through networks whose locations reflect both economic incentives and accumulated manufacturing capabilities.

Shih describes studying science and technology, taking engineering courses, and working in product businesses before discovering how decisively economics could constrain a technical proposition. He also says that serving on supply-chain committees appointed by the commerce secretary exposed him to gaps in officials’ understanding of how products are actually made. Those experiences help explain his preference for examining processes, suppliers, and investment conditions.

Patel supplies the counterpressure, asking what this analysis means for lost jobs and strategic dependence. The resulting interview combines industrial examples with a policy argument. Shih’s experience gives those examples texture, although his observations remain evidence from a conversation rather than a comprehensive measurement of manufacturing outcomes.

2. What the Episode Covers

A factory’s requirements provide a more revealing starting point than its announcement. Recalling Foxconn’s proposed Wisconsin LCD project, Shih says the absence of a nearby glass plant immediately undermined its plausibility. Residents subsequently sent him planning documents, and he concluded that the proposed concrete and structural steel were insufficient for the stability panel production required. These are his retrospective judgments, but they show the method he brings to the interview: identify what a process physically needs, then ask whether the proposal supplies it.

That method also explains supply chains that look geographically wasteful. Shih describes cashews grown in Ivory Coast or Ghana being processed in Vietnam, and fish caught off Alaska traveling to Qingdao for processing before returning. Distance alone does not determine the economic result. Different stages can remain cheaper in different places, making an indirect route commercially attractive.

A tariff changes that calculation without dictating the response. Shih describes businesses importing ahead of anticipated changes, using what he calls tariff engineering, and relocating production or assembly to Vietnam or Mexico. The interview does not explain the mechanics of tariff engineering, so it provides no basis for supplying a detailed account of how a particular arrangement works.

Televisions illustrate the distinction between changing trade and changing capabilities. Shih connects Mexican television exports with Chinese panel production and new direct shipping routes from China to Mexico. His interpretation is that components travel to Mexico for assembly before the finished products enter the United States. That explanation does not establish the origin eligibility of individual shipments. Its analytical value is that final assembly can move while an important upstream capability remains where it was.

The conversation also acknowledges why the previous trading arrangement lost support. Shih says inexpensive imports benefited consumers while displaced production left furniture, casting, and forging workers without employment. Those workers could themselves shop at Walmart, but cheaper goods did not replace their earnings. He criticizes the failure to adequately arrange retraining and also objects to persistent trade imbalances. His argument therefore does not require treating every existing pattern of globalization as desirable.

His preferred direction involves specialization among partners, including within North America and across a broader group of countries with complementary labor, resources, and technical strengths. He presents balanced exchange as a condition of that cooperation. Against this background, Patel and Shih discuss changing tariffs, China’s manufacturing capabilities, and AI-related pressure on computing resources. The episode’s breadth serves a focused question: which part of a production system must change to deliver the capacity policymakers actually want?

3. Core Views: Reasoning, Examples, and Limits

Shih’s argument begins with the economics of a high-cost location. In his account, durable American manufacturing needs high labor productivity, sophisticated processes, valuable skills, or products that competitors cannot readily reproduce. Raising the cost of an import may improve a domestic producer’s relative position. It does not itself provide the process knowledge or productivity needed to sustain production and wages. This is why he resists imagining reshoring as a straightforward restoration of an earlier employment model.

His example of high-temperature superalloy castings for jet-engine hot sections makes the differentiation argument concrete. He says only three companies possess the relevant capability, two in the United States and one in the United Kingdom. That count is a speaker-cited claim, not an independently established finding in the supplied evidence. The reasoning it supports is that a difficult, valuable capability creates a reason for others to buy from you. Maintaining such an advantage requires investment; the example does not demonstrate that every manufacturing activity can acquire comparable scarcity.

The iPhone exposes a different problem: several kinds of capability must work together. Shih describes advanced TSMC chips whose manufacture depends on equipment from ASML in the Netherlands and American companies including Lam Research and Applied Materials. Yet the finished phone also requires manual assembly. In his account, robots cannot currently perform the entire assembly operation. Moving the product therefore means addressing both sophisticated upstream technology and the organization of substantial human labor.

He invokes the challenge of mobilizing 100,000 or 200,000 workers, discussing production sites together with housing, warehouses, and food. These figures illustrate the scale he has in mind; they are not a staffing requirement for every phone plant. He also acknowledges progress in India while describing the replication challenge as difficult. His explanation concerns the conditions a move must satisfy, rather than proving assembly can never relocate or automation can never advance.

The exchange over China’s manufacturing knowledge contains a disagreement that matters. Patel raises subsidies, intellectual property, and the know-how Apple and Tesla brought into China. Shih emphasizes learning through customer relationships: Apple pushes suppliers to improve performance, quality, and cost, while Tesla’s qualification of local suppliers raises their capabilities. That mechanism offers a concrete explanation of improvement. It does not independently settle the intellectual-property allegations Patel raises, and the conversation provides no basis for treating either speaker’s framing as a complete verdict.

Shih adds regional competition and investment to his explanation. He recounts a student whose father answered competitive threats with more capital expenditure, and a factory replacing equipment only two or three years old. These observations help explain how low prices can coexist with automation and technical upgrading. They cannot establish an economy-wide investment pattern by themselves. Likewise, his favorable assessment of some long-term technology planning, technically trained officials, and the 863 plan must be read alongside his acknowledgment of severe historical planning failures. The interview does not demonstrate that planning necessarily succeeds.

Investment timing connects these capabilities to tariff policy. Shih says overseas businesses considering the American market have delayed decisions because rules keep changing. A factory needs confidence in demand over several years; he describes semiconductor investment in terms of ten- or twenty-year use and return expectations. The implication is that protection can increase a project’s appeal today while uncertainty weakens confidence in tomorrow’s returns. He explicitly qualifies this account: responses vary, and some automotive plants do enter the United States when tariff pressure becomes sufficiently strong.

Automotive production also shows why a national import measure can disrupt a multinational operating model. Shih describes an efficient typical plant as producing roughly 200,000 to 250,000 units of a model or model family annually. Companies concentrate different models in different countries, then exchange them to offer a complete range. Those figures are his description, not a universal engineering rule. Tariffs can disturb the exchange that supports the specialization. Discussing the light-truck chicken tax, he acknowledges domestic investment benefits while arguing that protected profitability can also mean higher prices. Investment, producer profit, and consumer benefit require separate assessment.

Finally, adaptation has limits. Shih argues that inventories, rerouting, and relocation can absorb trade shocks, but prices eventually rise when alternative channels close. He distinguishes memory chips, where he says demand has far exceeded capacity: another shipping route cannot necessarily produce additional chips. The distinction helps explain why commercial flexibility resolves some problems but leaves others intact. Taken together, these mechanisms form a useful account of industrial constraints, without constituting a complete empirical evaluation of tariffs’ overall net effect.

4. Learning and Application

A practical use of this interview is to define the desired capability before judging a reshoring proposal. Is the objective final assembly, a critical component, a hard-to-replace process, or employment? These goals may overlap, but evidence of one does not automatically establish the others. Shih closes by proposing a more strategic use of the American market’s appeal. Applied carefully, that means specifying what the country needs to know how to produce, identifying why that capability sits elsewhere, and asking which conditions a proposed intervention would actually change.

For an investment announcement, the Foxconn discussion suggests examining dependencies before accepting the headline. An analyst can ask whether essential nearby suppliers exist, whether buildings and equipment suit the process, and whether staffing and delivery arrangements are credible. This is an application of Shih’s reasoning, rather than a universal checklist he supplies. The decisive requirements will differ by industry; the glass and structural conditions of a panel plant should not be transferred mechanically to unrelated factories.

For procurement or policy reporting, track final assembly, important components, and core processes separately. The television example shows why a new export origin can coexist with continued upstream dependence on China. A sourcing change might reduce exposure to a particular trade route without establishing a substitute manufacturing capability. Both outcomes can matter, but they answer different questions. Any assessment of a specific shipment’s trade eligibility would also require evidence beyond this interview.

Supplier management offers another application. At Toyota’s Kentucky plant, Shih says he tried to learn what the company paid for American steel. The manager did not disclose the price, instead emphasizing a strategic long-term supplier relationship. The anecdote supports considering sustained cooperation and more predictable demand. It does not prove that a long contract necessarily lowers costs. A buyer adopting this approach should weigh continuity against flexibility and bargaining options, and specify the capability improvements the relationship is supposed to deliver.

Investment review should similarly make time explicit. If a plant relies on protection, ask how long the company expects to need before earning a return, what competitive capability should develop during that period, and how changes in policy would affect its location decision. Shih’s acknowledgment of tariff-induced automotive investment prevents a blanket dismissal of protection. His discussion of light trucks also prevents treating a factory’s arrival as sufficient proof of public benefit. Productivity, product capability, and the cost borne by buyers remain relevant outcomes.

Finally, diagnose the constraint before selecting a response. Inventories, alternative sourcing, or assembly relocation may buffer a trade barrier. If the shortage instead reflects demand exceeding productive capacity, as Shih describes for memory chips, routing changes may accomplish little. The two conditions can also coexist. The interview is most useful as a discipline for matching an intervention to a specific bottleneck and stating what remains unresolved, rather than as a ready-made answer for every industry.

Source

• Original episode: Why Trump’s tariffs didn’t bring manufacturing back to America | Decoder