Nearshoring Is Not Relocation: The Real Ledger of Mexico, Vietnam, and India

The Analysis

A folder lands on my desk with four sheets in it: an invoice, a packing list, a bill of lading and a certificate of origin. The client wants to know whether moving the last assembly step to Vietnam takes the tariff off. My answer is that the address on the invoice is the least interesting sheet in the folder. The one that decides the case is the costing sheet behind it, and in most of the files I see, that sheet does not exist in a form anybody can hand to an officer at a counter.

The article lays out the shift with numbers that are hard to argue with: 65% of firms changed sourcing, 51% nearshored, Vietnam's apparel exports to the US passed China's for the first time in 2025, Mexico shipped $535 billion to the US against China's $308 billion, and the US-China corridor contracted by roughly 30%, about $165 billion rerouted. It also names the mechanism that decides who survives the shift: substantial transformation, judged on value added rather than on the assembly address.

Where I sit, that mechanism is not an argument. It is a stack of paper with dates on it. What follows is which sheet, in what order, and by when.

Start with how a shipment actually travels through the paperwork. Classification first: the finished good needs one code, and every bought-in component needs its own. Then the declaration elements hanging off the code, meaning what it is, what it is made of, what it is for. Then origin evidence: a bill of materials with unit prices, a process route with hours per step, invoices for locally purchased inputs, and a written explanation of why the operations performed in the new country amount to a substantial transformation. Then examination, then duty, then release. Six steps, and the file breaks at whichever step was built on an assumption.

The numbers in the article describe a factory that moved. They do not describe a filing system that moved with it. A plant can be stood up abroad in months; the evidence chain for its output takes longer, because the evidence has to be generated while the work is being done, by people who are not thinking about customs at the time. That mismatch is why I keep meeting companies that genuinely relocated production and still cannot answer the one question an examiner asks: how much of this thing's value was created where you say it was created.

Why now is not mysterious. Tariffs hardened from a bargaining chip into a permanent line item at a combined rate around 37.3%, and the article is blunt that the motive was exposure rather than wage cost, since Mexican wages are not low and firms went anyway. Meanwhile 39% of firms are absorbing the cost silently. Silent absorption is not a strategy; it is what happens when nobody kept the paper that would let you claim something later. Duty drawback, exclusions, contract renegotiation, every one of them needs a clean entry record with a date, a code and a value on it. If the record is wrong, the remedy exists in the regulation and not in your file.

The exposure is not spread evenly. A large manufacturer with a trade compliance function already has a costed bill of materials, a process route and a classification file, and can walk into an examination without a scramble. A mid-size exporter usually has the data, sitting in the ERP and in the costing team's spreadsheets, but not in the shape a declaration needs, which means it can be reassembled in three weeks and not in four hours. A trading company that buys finished goods and resells them has nothing but an invoice and a packing list. For that company, nearshoring can only ever be a change of address, and a change of address with no manufacturing behind it is exactly the fact pattern that gets read as transshipment.

Trade direction decides who carries the argument. The importer of record carries the liability, so a US buyer facing an origin question pushes it straight back up the chain to the supplier. In practice that turns origin evidence from a customs document into a quotation term. Buyers are starting to ask for a costed component breakdown before they place the order, and a supplier who cannot produce one gets filtered out at the RFQ stage, long before anybody discusses price. That is a sales problem wearing a compliance costume.

When you feel it is not at the border on the first shipment, because the first shipment usually clears. You feel it the first time a container is selected for examination, and the first time a buyer runs a supplier audit. Both arrive without warning and both ask the same question, which is not where the factory is but what happened inside it. That is also the moment when the file you should have built two years ago becomes the only thing standing between you and a very expensive conversation.

Here is what the coverage misses: origin is a costed chain, not an address, and the cheapest piece of proof in that chain is usually the code transition. If your imported components sit under one set of tariff codes and your finished good sits under another, and the operations in between are real, then the code change is the easiest thing in the world to document, because it falls straight out of the bill of materials. Most companies I audit have never lined those two lists up against each other. They will argue about substantial transformation for an hour and then discover that their own finished-good code has been wrong for three years, which is a separate and much more expensive problem. Look at the codes before you look at the map.

The other thing nobody says out loud is that this evidence has a vintage. A certificate can be reissued. An invoice can be reprinted. A process route from two years ago cannot be reconstructed, because the people who ran those steps have moved on and the line has been changed twice since. Origin files are written in the present tense or they are not written at all. That is why the useful deadline is not the date of an examination. It is the date of your next production change. Build the file before you move the line, not after.

A small ledger, with the assumptions stated. Take an exporter with 40 active SKUs going into the US market, and assume the top 20% carry most of the value, which is my assumption and not a figure from the article. That is eight SKUs for the first pass. Rebuilding one SKU properly means pulling the component list, mapping each component to a code and a country, and writing down the local value added. Call it 90 minutes each when the data is clean, which puts the first pass at roughly 12 hours of work. Now put the exposure on the other side. On a declaration of $100,000 in value, again my assumption, every percentage point of duty is $1,000, and a classification dispute is never about one point. Twelve hours against that is not a close call. The reason it does not get done is not cost. It is that nobody owns the twelve hours.

The counter-case deserves a hearing. If your local value added is real and documentable, moving production is defensible and the file will hold up under questioning. If the value added is a label and forty minutes of packing, no amount of paperwork rescues it, and the correct decision is to leave that stage where it is and move something else instead. What flips the conclusion is substance, measured in cost, hours and code change, not in kilometres. And if your channel never audits you, the risk is latent rather than absent. It surfaces the day your buyer gets bigger, or the day the channel hires a compliance department.

The mistake I see most often is a beautiful origin narrative written by somebody who has never stood on the line, describing a process the plant stopped running last spring. It survives a quotation and dies at an examination. My rule is boring: if the person who runs the step cannot explain the sentence, the sentence comes out of the file. Tonight, pull the ten highest-value SKUs you ship to the US, and for each one write the finished-good code and the code of your single largest imported component on the same line. Where the two codes are identical and nothing real happens in between, you have found the shipment that will cost you the most, and you have found it before anybody asked.

↑ Back to the key points

— By Derek Xu