China's commerce ministry opposes reported US plan for extra 7.5% tariff on Chinese goods

The Analysis

Every time a tariff headline breaks, my phone lights up inside the hour, and nobody asks me about the headline. They ask me whether their container is safe. My answer almost always starts with two other questions: what paper exists for that shipment right now, and which of those documents can still be changed? Those are two separate questions, and only one of them leaves you something to do before lunch.

What landed on 27 August came from the Ministry of Commerce's regular press conference, with spokesperson Huang Ling on the podium. Beijing said it firmly opposes the additional 7.5 per cent tariff on Chinese imports that Washington is reported to be considering, and it reserves the right to take all necessary measures. The object of the complaint is the Section 301 investigation launched against sixteen economies over alleged overcapacity, described as politicising trade and as textbook unilateralism and protectionism. A month earlier, on 28 July, Beijing had already published a position paper on the so-called overcapacity question, arguing that exporting a lot and running a large surplus are not the same thing as having excess capacity. That is the whole factual ground under our feet today: a statement of opposition, a reserve-the-right sentence, sixteen economies named in a proceeding, and a paper issued on 28 July.

Notice what is not on that list. There is no tariff line register in front of us, no application date, no implementing text we can quote line and verse. That matters more than the number itself, because documents used to clear goods are not written against intentions. Whether any of this carries binding force from a given date is a question Dr. Ingrid Voss should answer with the source open in front of her; I only deal with what happens after that is settled. So this piece is going to walk one shipment's file folder end to end, from the proforma invoice to the corrections you can still squeeze out after the goods have landed, marking who holds each page, when it stops being yours to touch, and how long each fix actually takes.

Here is the shipment I will use as the working example, and I want to be blunt that it is invented for the arithmetic, not lifted from a client file. Assume a forty-foot container of metal-framed chairs, 1,240 cartons, leaving Ningbo, invoice value USD 46,800 on FOB terms, booked by a seller who quotes DDP to a US distributor because that is what the buyer demanded three years ago and nobody has reopened the question since. Assume thirty CFR-style days on the water to the US west coast. Assume the buyer paid nothing extra at the border last month, which means whatever customs took out of the shipment came out of the seller. Everything below hangs off those assumptions, and you should redo it with your own numbers before it means anything to you.

Follow the paper and you find there are actually four moments when a shipment's data hardens, and they are weeks apart. The first is order placement, when somebody in sales writes a commodity description onto a proforma invoice, often in whatever language came out of the factory's quotation, often copied down from the previous order without anybody looking at it. The second is bookings, when that description gets keyed into the shipping instruction and then onto the bill of lading draft. The third is the entry itself, when classification, valuation, origin and party data are submitted to customs by whoever files for you.

The fourth is whatever happens after arrival, once someone decides a line was wrong and you are asking permission to change it instead of just changing it. The interesting thing is not the moments themselves but how much each one costs to undo. The first is free. The fourth can run to months, and the fact that duties may have been assessed already does not make correction faster.

Consider how much flexibility actually exists on the water. Assume the carrier's documentation cut-off sits forty-eight hours before container yard closing, and assume each round of corrections takes half a working day if the file has a single owner and nobody is on holiday. Under those assumptions you have maybe two or three rounds before the bill of lading is released, which means practical control over consignee, notify party, marks and numbers, piece count and description for something in the range of thirty-six to seventy-two hours. After release, every one of those becomes a formal amendment with a fee attached, and some of them become impossible without the consignee's cooperation, which takes you into another calendar entirely and into a different set of signatures. I have watched people lose a week chasing a consignee change on a released bill because three parties had to agree and one of them was in a different timezone and answering email twice a day.

The label on the box says rate shock, but the documents underneath it say classification and origin, and those two decide almost everything about what a 7.5 per cent figure actually means to a given importer. Tariff classification is the index key into the entire rate structure. Get the heading wrong and any prediction about duty is noise, whatever the headline percentage is. Assume our chairs sit in a heading carrying a zero general rate for illustration; assume a competing reading puts them in a heading with a small ad valorem rate; the difference between those two answers changes the base that any additional levy is calculated on, changes which exemption or exclusion path is even arguable, and changes the answer to a question nobody asks in advance, namely which specific sub-line any future measure actually names. This is why I keep repeating the same four words to my own team: classification first, duty treatment second. The reverse order wastes a lot of afternoons.

Origin is the other half, and it is the half nobody can fix at the last minute. An investigation framed around capacity across sixteen economies pushes examiners and importers toward the same question, which is where the thing was made and out of what. The evidence for that answer does not live with the exporter who signs the invoice; it lives with whoever ran the production line, in bills of materials, routing sheets, purchase records for input materials, sometimes in photographs of a process nobody thought to photograph.

Ask a factory for that package and the honest answer from experience is five to ten working days if the relevant people are at their desks, longer if they are not, and no amount of urgency on your side changes it, because you are not the one who has to dig through last quarter's records. That is the single longest lead time in the whole file, longer than the voyage in some cases, and it is also the one that never gets started until somebody panics.

Now the number. Assume the reported additional 7.5 per cent eventually applies to this heading, ad valorem on customs value, with no exclusion available to us. Assume the FOB invoice value of USD 46,800 is our customs value, and that there are no assists, royalties or additions nobody has bothered to put into the declared price. Assume no proceeds of subsequent resale applies, no rebate, no commission sitting outside the invoice. Under all of that, USD 46,800 times 7.5 per cent is USD 3,510 on this container. Per unit across 1,240 chairs that is USD 2.83. It does not sound like much until you put it next to the margin that paid for the container to exist.

Assume the seller earns an eleven per cent gross margin on this line. Eleven per cent of USD 46,800 is USD 5,148, which is the entire reward for manufacturing, packing, financing and shipping those chairs. The additional duty, if the assumptions hold, would consume USD 3,510 of that USD 5,148, which is roughly sixty-eight per cent of the margin in a single stroke. Another way to hold the same fact: preserving the same absolute profit would require landing this cargo for USD 3,510 less, or selling it for that much more, and both options have their own paperwork attached. Cutting cost means renegotiating with a factory that probably has less than eleven per cent to give. Raising price means going to a buyer who has read the same headline and is already practising the face he will make when you ask.

Scale the same arithmetic across a year and it stops being an anecdote. Assume three containers a month on this program, thirty-six a year, same assumptions otherwise. Thirty-six times USD 3,510 is USD 126,360 of additional duty in a twelve-month period. Because duty is cash out at the border, that is also a working capital event, not just a profit event. Assume your cash cycle leaves roughly sixty days of inventory on the water and in the warehouse at any time, so about two months of additional duty is outstanding at once, call it around USD 21,060 tied up. Assume your cost of funds is six per cent a year. That is roughly USD 1,264 annually in financing cost on top of the duty itself, which is small next to USD 126,360 but large next to any of the individual fixes being discussed internally.

Then there is the cost of waiting, and it cuts both ways, which is why nobody should make the waiting decision by instinct. Assume the terminal gives four free days and charges seventy-five dollars a day per box after that, and assume nothing about when any measure takes effect, because we genuinely do not have that date. Ten days of dithering over a policy signal that has not arrived yet costs seven hundred and fifty dollars in demurrage under those assumptions, which is still well short of USD 3,510, so a week of deliberate waiting is not irrational. Thirty days of waiting costs nearly two thousand dollars and gets you no closer to certainty. My own rule of thumb at that point is to put a number of days in writing before starting to wait, and to stop waiting when that number runs out, because open-ended waiting is just paying rent on a box you cannot use while telling yourself you are being prudent.

Where people get into real trouble is the third category, which is not paying the duty but papering around it. When a rate headline breaks, somebody always suggests reissuing the commercial invoice with a lower value, or asking the factory for a certificate of origin naming a different country, or splitting one shipment into several smaller ones. I am not going to give you a legal judgment on any of that, because that is not my lane and I have watched too many people hurt themselves doing it. What I can tell you is that those moves create a second problem that is entirely independent of whether any tariff lands: once the documents in a file contradict each other, every future encounter with customs involves explaining the contradiction, not just the original entry.

A wrong entry can be corrected. A file that shows inconsistent prices, inconsistent origins, inconsistent dates is much harder to walk back, because once a version of reality is in writing and submitted, every later version has to explain why it differs. The date particularly. Do not backdate anything. There is no scenario I have seen where a backdated document improved a bad situation, and plenty where it turned a computable problem into a credibility problem.

The more boring work is also the more useful work, and it starts with knowing exactly who can touch each piece of the file. Alphabetically it looks like a list of documents; operationally it is a list of people with different incentives. The factory controls production evidence and the certificate of origin source documents, and it responds to whoever pays it. The forwarder controls the bill of lading draft and the shipping instructions, and it responds before its own cut-off. The buyer controls the purchase order wording and, under DDP terms, has very little reason to help you tidy up a description they never read. Whoever files the entry controls what actually goes to customs, and they will go with whatever data you sent unless you told them otherwise. Everything useful I have ever done on a shipment happened because one person owned that mapping and could chase three other parties in an afternoon, not because anybody had better information.

Ports make this messier rather than tidier, which is the part that never makes it into calculators. The same correction that takes three days to process in one place can take three weeks somewhere else, and the same description that sails through one entry point can draw a document request at another. Which port you clear at changes your examination exposure, how long a post-entry amendment sits, whether someone will accept a scanned set or insists on paper with wet stamps, and how much of the communication happens by telephone. I am deliberately not telling you this port is fast and that one is slow, because rates and staffing move, and my experience is one person's experience. Ask whoever files your entries what their recent turnaround was at your own port, in writing, and hold them to the answer. There is no substitute for that conversation, and no tool sells it.

One thing I would do today, before any of the modelling: close the exposure on quotes that are not signed yet. If you are quoting DDP into the US and the quote does not say who absorbs a duty change taking effect after the contract date, then by silence you have said it is you. Assume that across West Coast programs you have eleven open quotations on the table at any time. That is eleven unpriced risks, each one turning into a three-thousand-dollar argument nine or ten weeks later. Adding a clause takes half an hour of someone's morning and costs nothing when the feared outcome never arrives. Freezing new DDP quotes for a stated window costs you a day's delay in negotiations. Losing those eleven arguments costs most of the above arithmetic, twice over, because you will also lose the ones you win.

The other thing I would start this week is the boring audit that everyone postpones. Pull twelve months of shipments to the US at line level, and for every line record the tariff classification actually used, the country of origin declared, the factory of record, the name of whoever reviewed the classification, and the file path to the evidence that supports both answers. Do not do this as a spreadsheet exercise done by someone who has never seen the products; do it with the person who signs the invoices in the room, because half the lines will raise a question only that person can answer. Expect two things: a pile of lines where nobody can produce the file path, and a handful where the description on the invoice has drifted for three years away from what is actually in the box. Both are more valuable than another round of news reading, because both are fixable at zero freight cost.

Alongside that, write down your own triggers, because triggers are what stop a headline from becoming a panic. A trigger is a sentence with a number in it that somebody downstream can execute without calling you. Something like: when a formal measure is published with an application date, rerun landed cost for the top twenty lines within twenty-four hours. Something like: if the additional duty on a line exceeds five per cent of its customs value, stop accepting new bookings above USD 50,000 ex-works until pricing has been redone. Something like: if any shipment would arrive less than fifteen days after a published application date, get the entry data frozen seventy-two hours before container yard closing so nothing is still being debated at the cut-off. Nothing clever there, but it converts a hundred emails into three decisions, and decisions are what you pay your own people for.

Inventory and routing options deserve a sober look, especially while everyone is excited. Front-loading looks smart until you price it. Assume pulling two extra months of stock forward ties up some USD 93,600 of inventory value in this example; at six per cent funds cost with a sixty-day extension that is maybe a little over three thousand dollars of carrying cost, which buys you the merchandise while the old rules apply and nothing else. It does not buy you a lower rate, and if the measure ends up carving out your heading, you have spent real money avoiding a cost that never existed.

Bonded storage or a foreign trade zone arrangement lets you defer and sometimes re-export without paying, and it can be genuinely useful, but it adds its own record-keeping: admissions, manipulations, weekly or monthly reconciliations of quantities, another set of numbers that must agree with another set. If your record discipline on ordinary entries is shaky, the zone will find that out quickly. Changing the mode buys days, not a rate: air freight moves arrival earlier relative to a possible application date, and for high-value low-weight cargo that can be worth the premium, while for chairs it plainly is not. Switching sourcing countries is the slowest lever of all and the one with the longest shadow, because new factories bring new classification questions, new origin evidence gaps and new first-article inspections, and nobody should do that after reading one news story.

There are also three specific traps whose cost lands on the desk of whoever is trying to be careful. The first is the surcharge basis: when carriers or intermediaries pass a cost through, check what the percentage is applied to, because it is rarely the customs value you have in your example, and freight components are frequently included. The second is the transit commitment: a promise of thirty days on the water is not a promise about when revenue is recognised or when duty is assessed, and the two calendars drifting apart by a week has killed more budgets than any rate change. The third is document delay turning into rent: every day the file is incomplete at destination is another day of storage, and for a value-dense cargo the storage bill can exceed the per-day cost you were willing to pay to avoid the exposure in the first place. None of those three require a rate to change by a single percentage point to hurt you.

On accountability, keep it boring and specific. One named owner per shipment family, one place where the final, submitted data is stored as sent rather than as intended, and a monthly reconciliation between what was filed and what was invoiced to somebody. If the person who files the entry and the person who pays the duty are never in the same reconciliation meeting, mistakes stay invisible for exactly as long as your audit cycle allows, which is often a year. Retention matters too. Origin evidence, classification rulings or opinions, supplier declarations and any correspondence asking for changes should live somewhere findable by someone who was not there, because the useful life of that paper easily outlasts the tenure of the person who filed it. When a reviewer two years from now asks why a heading was chosen, the answer should be a file, not a memory.

So, can this shipment be saved? Sometimes. It depends less on the policy than on the calendar, and mostly on which of the four hardening moments have already passed for that particular box. If it has not sailed, you have real room: tidy the description, verify the classification, get the origin package moving, sort out who pays, and you are mostly fine. If the bill is already released, your room is down to what the entry data says and how quickly you can get corrections accepted at your port, and every hour costs something. If the goods have arrived and been entered, then it is an amendment, a timeframe measured in weeks, and a conversation with the carrier about rent while you wait. The honest answer I give when someone phones me about a container is usually a time of day rather than a strategy. This one I can save, but you have to have the shipping instruction to me before four this afternoon.

↑ Back to the key points

— By Derek Xu