Asia-Pacific Investment Report 2018 — Supply Chains Become Political

2018's trade measures affected Asia-Pacific more directly than any other region, since the region's economies are the manufacturing base of global supply chains.

The immediate effects were the obvious ones: tariffs raised the cost of affected goods, exporters faced reduced competitiveness, and companies with cross-border production saw margins compress.

The larger effect was uncertainty, and it operated through a mechanism worth understanding because it is frequently underestimated.

A company deciding where to locate production is making a decision with a horizon of a decade or more. Factories are expensive, take years to build, and cannot be moved. That decision requires assumptions about trade rules over that horizon.

When those rules become uncertain, the rational response is frequently to delay. Building in the wrong location is expensive; waiting is cheaper. So investment is deferred — not because tariffs made a location uneconomic, but because nobody could determine which location would be economic.

The cost of that delay is incurred regardless of how the uncertainty resolves. Investment that does not happen in 2018 does not happen, whatever the rules turn out to be. This is why trade uncertainty measures correlate more strongly with investment behaviour than tariff levels do.

The second finding of 2018 was that supply chains move more slowly than commentary assumed. The binding constraint is not factory construction — that is comparatively fast. It is the supplier ecosystem: the hundreds of component suppliers, tooling firms, logistics providers and skilled workers that accumulate around a manufacturing centre over decades.

Third, and most durably: technology restriction emerged as a distinct category from trade policy. Restrictions on specific companies' access to specific technologies operate through different mechanisms than tariffs, are less reversible, and have proved far more persistent.

Why uncertainty costs more than the measure

The mechanism is a general one in investment under uncertainty, and 2018 is its clearest recent demonstration.

An irreversible investment under uncertainty has an option value in waiting. If a company builds a plant now and conditions change, the investment is stranded. If it waits, it learns more before committing. That option has value, and the value rises with the uncertainty.

The consequence is counterintuitive but robust: investment falls when uncertainty rises, even if the expected outcome is unchanged. A 50% chance of favourable rules and 50% of unfavourable produces less investment than certainty of the average, because the option to wait is worth more when outcomes are dispersed.

Applied to 2018:

• Companies deferred decisions about where to locate future capacity.
• The deferral cost was real and immediate — capacity not built, hiring not done, suppliers not qualified.
• It was incurred regardless of resolution. Even a complete resolution in favour of the status quo would not recover the deferred year.
• Resolution was slow, so the deferral extended.

The tariff is a cost you can calculate. The uncertainty is a cost you incur by not deciding, and it accrues whether or not the tariff ever arrives.

This is why measures of trade policy uncertainty track investment behaviour better than tariff rates do, and it is the strongest argument for policy predictability independent of policy content.

Why supply chains move slowly

The assumption that manufacturing relocates readily in response to trade measures substantially underestimates what a supply chain is.

A modern manufacturing centre is an ecosystem, not a factory:

• Component suppliers. A device contains hundreds of parts from dozens of suppliers, most located near the assembler because proximity reduces cost and lead time.
• Tooling and equipment. Specialised machinery, and the firms that build, maintain and repair it.
• Skilled labour. Not just assembly workers but process engineers, quality specialists and production managers — skills accumulated over years.
• Logistics. Ports, freight, customs handling and warehousing scaled to the volume.
• Institutional knowledge. Accumulated understanding of how to run a specific process at scale, which is largely tacit and does not transfer with documentation.

Building a factory is the fast part. Building the ecosystem takes years to decades.

The consequences observed from 2018 onward:

• Relocation happened in stages, starting with final assembly — the least ecosystem-dependent step — while component production remained.
• "Made in" designations changed faster than production did. A product assembled in a new location from components made in the original one has changed origin without changing much economically.
• Costs rose. Relocation to a less developed ecosystem means higher costs, at least initially, which is why it does not happen without a policy reason.
• Duplication rather than relocation became the common pattern — maintaining existing capacity while adding new capacity elsewhere, for resilience rather than cost.

That last point is the durable one. The change was not that production moved. It was that companies stopped optimising purely for cost and began paying for redundancy. That is a permanent increase in the cost of manufacturing, and it persists.

Who benefited, and how slowly

Southeast Asia and India were the principal beneficiaries of supply chain diversification, and the pattern of benefit is instructive.

Vietnam was the clearest beneficiary, with existing manufacturing capability, proximity to the incumbent ecosystem, competitive labour costs and an accommodating policy environment. Electronics assembly grew substantially.

India attracted assembly investment supported by domestic policy incentives, and benefited from a large domestic market that made local production attractive independent of export considerations.

Thailand, Malaysia and Indonesia benefited in specific sectors where they had existing capability.

The pattern of benefit is the analytically useful part:

• Final assembly moved first, being least dependent on the local supplier ecosystem.
• Component production followed slowly, and much did not follow at all — components continued to be sourced from the original ecosystem and shipped.
• The value added locally was initially small. Assembly is the lowest-margin stage; the value is in components and design.
• Deepening took years. Local supplier ecosystems developed gradually, and where they did, the local value added rose.

The investment implication is a warning about timing. An investor allocating to beneficiary markets in 2018 on the expectation of rapid gains was working on the wrong timescale. The benefit was real and it accrued over five to ten years, with the early years capturing the least valuable stage of production.

Technology restriction as a distinct category

The most durable development of 2018 was the emergence of technology restriction as a separate instrument from trade policy.

Tariffs are a tax on goods. They raise cost, are quantifiable, and are relatively easy to reverse.

Technology restrictions limit access to specific technologies, components or software for specific entities. They operate differently in every respect:

• They are not primarily about cost. A restricted company cannot buy the input at any price.
• They are motivated by security rather than economics, which means economic arguments do not resolve them.
• They apply extraterritorially where the technology involves controlled intellectual property, so a non-US company using US-origin technology may be affected.
• They are far less reversible. A tariff can be removed by a decision. A restriction reflects a security assessment that persists across administrations.

The consequences for the region proved structural and long-lasting:

• Companies dependent on restricted inputs faced existential rather than margin risk.
• Domestic substitution efforts accelerated, with substantial state support directed at reducing dependence.
• Supply chains bifurcated, with some companies building parallel arrangements to serve different markets under different rules.
• Technology investment acquired a political dimension, since a company's access to inputs and markets depended on regulatory determinations rather than commercial performance.

For investors the implication is a diligence requirement that did not previously exist. Assessing a technology company in the region requires assessing its exposure to restriction — which inputs it depends on, whether they have controlled origin, and whether substitution is feasible. That is a regulatory and geopolitical assessment, not a commercial one, and it has become a standard part of diligence across the region since.

Reading trade data when assembly moves and value does not

A measurement problem became acute in this period and is worth setting out, because gross trade statistics systematically mislead about where supply chains have actually moved.

Gross exports record the full value of a shipment, attributed to the country it was exported from. A device assembled in one country from components made in several others is recorded at its full value as an export of the assembling country.

Value added records what each country actually contributed. For a device assembled from imported components, the assembling country's contribution is the assembly — which is typically the lowest-margin stage.

The divergence between the two is what supply chain relocation looks like in its early years:

• Gross exports from the new location rise sharply, because full device values are being recorded there.
• Value added rises far less, because the components are still made in the original ecosystem and are being imported for assembly.
• Gross imports into the new location also rise, since the components have to arrive — which means the trade balance may barely change even as gross exports surge.

The consequence for interpretation is direct. A headline showing a large increase in electronics exports from a beneficiary economy is consistent with a genuine deepening of local manufacturing capability and equally consistent with final assembly having relocated while everything valuable stayed where it was. Gross data cannot distinguish them; value-added data can.

The OECD's Trade in Value Added database exists precisely for this and is free. It is the correct source for assessing whether a supply chain has moved or whether only its last step has, and the answer determines whether the beneficiary economy is capturing significant economic value or performing a low-margin service.

"Made in" is a customs designation. It is not a statement about where the value was created, and in a fragmented supply chain the two can diverge almost completely.

What an allocator could act on

Use value-added rather than gross trade data. This is the single most useful correction available for assessing supply chain shifts and it costs nothing. Gross export growth in a beneficiary economy overstates the economic gain, frequently by a large margin in the early years.

Expect a five-year lag before a thesis becomes measurable. The ecosystem constraint means relocation proceeds in stages: assembly first, components later, local value added rising slowly. An investor allocating to supply chain beneficiaries in 2018 was working on a timescale that would not produce visible results until roughly 2023, which the 2023 regional report confirms. That is a normal timescale for industrial change and an uncomfortable one for a fund with a defined life.

Price uncertainty separately from policy. The investment-suppressing effect of unresolved rules operates regardless of how the rules settle, and it is measurable — the Baker/Bloom/Davis trade policy uncertainty sub-index tracks it free. An analysis focused on which outcome would occur misses the cost being incurred while the question is open.

Assess technology restriction exposure as a distinct diligence item. Which inputs a company depends on, whether those have controlled origin, which markets it sells into, and whether substitution is feasible. This is a regulatory and geopolitical assessment rather than a commercial one, and it became a standard requirement across the region after 2018 for good reason.

Recognise that duplication, not relocation, was the durable change. Companies largely maintained existing capacity while adding new capacity elsewhere — paying for redundancy rather than optimising for cost. That is a permanent increase in the cost of manufacturing, borne across the whole system, and it persists regardless of how any specific trade dispute resolves.

What 2018 established for Asia-Pacific

• Uncertainty was shown to cost more than the measures, through deferred irreversible investment.
• Supply chain relocation was shown to be ecosystem-constrained, not factory-constrained, and therefore slow.
• Duplication replaced optimisation, permanently raising manufacturing cost in exchange for resilience.
• Beneficiary markets gained slowly and at the lowest-value stage first.
• Technology restriction emerged as a distinct, durable and less reversible instrument, adding a geopolitical dimension to technology diligence.