China-EU E-Commerce Air Tonnes Drop 50% in September as Dutch Parcel Lines Fall 70%

The Analysis

The number I could not get out of my head this week was not the fifty percent. It was the nineteen. Aevean put low-value e-commerce tonnes flown from China into the European Union down fifty percent year on year in September, and down nineteen percent to every destination put together. Nineteen percent on its own is what a soft patch looks like: demand cools a bit, a platform pulls a promotion, buyers wait for payday. Fifty percent is not a soft patch. Fifty percent is surgery. The thirty-one points between those two figures is not cargo that evaporated, it is cargo that changed costume, and the costume it put on decides which hour of your operation breaks first.

The rest of the data backs that reading. Across August and September together, e-commerce declaration lines in the Netherlands ran about seventy percent below a year earlier, while ordinary trade lines crossing the same border rose forty-three percent. In Belgium, low-value import declarations collapsed from roughly one hundred and sixty-six million in June to about twenty-nine and a half million in September, a fall of eighty-two percent, and between the first of July and the twenty-seventh of September the three-euro duty pulled in three hundred and sixty-four million euros. Same fence, different traffic. A road that used to run at motorway speed has been throttled down to a side street, and the traffic did not stop flowing, it took the next junction.

Take that last figure apart, because nobody else has. Three hundred and sixty-four million euros collected at three euros a parcel works out at around one hundred and twenty-one million parcels over eighty-eight days, which is about one point four million a day and a little over forty million a month. Set that against twenty-nine and a half million declarations in September and the two series stop speaking to each other. Either one declaration line carries several taxable parcels inside it, or collections run weeks behind the declarations they belong to, or the September declarations were never the same population as the duty collected since July. Three different worlds, three completely different plans. Paperwork is not my desk, and Derek reads those files better than I ever will, so before anyone budgets against one of the three, get the answer in writing from your broker.

Let me lay the old direct-mail journey out hour by hour, because it is the only way to see what has been cut. The consumer orders at nine in the evening. The seller prints labels and hands the bag over at the next pickup: call it twelve hours. The consolidation house scans, sorts and manifests the day's bags: six to twenty-four hours depending on how late the van came in. Then comes the export leg and the hunt for space: twelve to thirty-six hours, longer if you missed the afternoon cut-off, and missing the afternoon cut-off is the most expensive single mistake in this trade because everything behind it is scheduled.

Flight time from South China to a European gateway runs about eleven to thirteen hours in the air. Arrival handling and presentation of the declaration take four to twenty-four hours in a good month. Release follows, and the parcel enters the postal or last-mile network, where the first delivery attempt lands twenty-four to seventy-two hours after release, sooner on a dense urban route, much later in the countryside. Add the failure tail on top: failed first attempts, redelivery scheduling, collection-point pickups, and parcels that sit in a customer service queue because nobody told the buyer anything at all. Door to door that was seven to twelve days on an ordinary day, and it was sold to shoppers as five to eight.

Now drop a flat charge into the middle of that chain. The three-euro duty scales with nothing. It does not care whether your parcel weighs ninety grams or two kilos, whether it is going to central Paris or to a village two hours outside Lisbon, or whether the thing inside retails for five euros or sixty. Which means its bite depends entirely on ticket value. On a twenty-six euro item it behaves like a tariff of roughly eleven and a half percent. On a twelve euro item it turns into twenty-five percent. On a six euro item it is half of everything the customer paid.

Run that across a catalogue and the damage has a shape: the expensive lines are barely scratched, and the cheap lines stop being products and start being losses. This is why quoting the fifty percent headline without doing your own arithmetic is dangerous. A seller averaging thirty euros a ticket probably lost far less than half its volume. A seller averaging ten euros a ticket almost certainly lost more than half and is now clearing stock rather than shipping it. One headline, two completely different post-mortems.

There is one obvious reaction to a flat per-parcel charge and everybody in this business has thought of it already: bundle. Put two or three units in one parcel, pay the charge once, ship once. I will say the quiet part out loud. Bundling changes the goods description on the entry and it changes the declared value on the entry, and those two fields are exactly what gets reconciled afterwards, in both directions, by people who are not in a hurry. Bundling is a margin trick with a compliance tail, and in my experience the tail arrives roughly two quarters after the trick becomes fashionable. If your entire answer to a three-euro duty is to bundle everything into multi-packs, you have not escaped the rule. You have deferred an argument with someone who has far more patience than you have.

Then look at where the goods actually went, because that is where your hours move to. Dutch e-commerce declaration lines down about seventy percent while ordinary lines on the same border rise forty-three percent is not a picture of demand leaving Europe. It is the same sellers sending the same products, landing them as ordinary import entries, and feeding them into European warehouses. The physical flow changed from per-parcel to per-pallet. And the instant that swap happens, the scarce resource behind your delivery promise changes identity. Before, the scarce resource was aircraft belly space out of Asia plus the release queue at the border.

After, it is a truck appointment at a warehouse dock somewhere in the Netherlands or Belgium, the putaway window inside that building, and the local carrier's afternoon pickup cut-off. Nobody publishes a single number about those three things. Everything about your customer promise now depends on them. If you have watched a European warehouse in November, you know how it goes: appointments booked out days ahead, trailers idling on the access road, receiving promising twenty-four hours for putaway and delivering seventy-two. Those seventy-two hours cost you more than a missed flight, because you pay them on every single order, not on one shipment.

Timing next, because timing is my job. In week one of the switch nothing looks wrong. Sales keep coming, direct parcels keep shipping, and the first bulk shipment is somewhere on the water doing nothing. The trouble opens in weeks two and three, and it opens in documents rather than in transport. A bulk entry needs a full commercial invoice, a proper packing list, a classification per item, and consistency between what you wrote and what is in the boxes. A simplified parcel manifest tolerated sloppy item descriptions for years; a bulk entry does not. If your item master data is incomplete, your first two bulk shipments sit.

Assume one to three business days per hold, and note that unlike one parcel going missing, this hold is thirty thousand units at once with the storage clock already running. Weeks four to eight are the good weeks: inventory becomes sellable, your promise to the shopper drops from eight to twelve days down to one to three inside Europe, conversion improves, and returns arrive at your own building instead of vanishing into a cross-border void. Weeks eight to twelve are where the trap sits, and the trap is the first replenishment decision. Refill by sea from South China and you are buying roughly thirty-five to forty-five days door to door.

Rail into Europe is around eighteen to twenty-five days. Air bulk gets you seven to ten days to the warehouse. Those are observed ranges, not promises anybody signed. If you set your safety stock back when the model was eight-day parcels, your European shelves are calibrated to a pipeline that no longer exists, and nobody warns you about that until the morning a bestseller goes out of stock and the only tool left is a charter you cannot pay for.

Here is the part I have not read anywhere, and it is the one thing I would want you to carry out of this. This is not really a tariff story. It is a commitment story, and it ratchets. A three-euro duty is a political decision, and political decisions get revised, delayed, restructured into a handling fee, or folded into a larger customs reform. European warehouse capacity is not political. It is concrete, it is staffed, and it is leased. Assume the kind of contract a mid-sized seller actually signs: three to six months of notice, a fixed monthly handling minimum, a storage rate that does not fall when your sales fall.

Buy into that and you have taken on an obligation that survives any change in the duty. Move your stock into Europe this quarter and a reversal of the policy next year does not hand the parcel channel back to you, because the reason you will not use the parcel channel is no longer the duty. It is the building you are paying for. I have watched this same pattern with inland depots and with bonded space: the reversible policy pulls the trigger, the irreversible asset follows the trigger, and two years later everybody insists the asset was always part of the plan.

The second half of the ratchet is accounting, and it bites in month one. When goods moved one parcel at a time, duty and handling dribbled out across thousands of small transactions and nobody noticed the cash. When a month of stock arrives on one entry, the duty and the import tax on that whole entry fall due on one day. Finance people call this working capital and charge you for it; operators call it the month nobody could pay anyone on time. Whatever your treasury looks like today, plan for the first bulk month being the worst cash month of your year, and do not discover that the week after arrival.

The third consequence is that the cost of sloppy data rises by orders of magnitude. A misdeclared line on a parcel costs one customer one complaint and maybe a refund. The same mistake on a bulk entry holds thirty thousand units, starts the storage clock, and drops your own staff into a correction loop with people who measure patience in working days. Before the switch, bad data was a customer service cost. After the switch, bad data is a supply chain outage.

Where could I be wrong, and what would change my mind. If the drop were mostly demand rather than mostly channel, everything I said about warehouses is noise. The nearest thing we have to a control group is Aevean's own nineteen percent for every destination combined: the same sellers, the same month, the same consumers, and Europe fell about thirty-one points harder than everywhere else put together. That gap is the customs measure, not the shopper. If that gap closes over the next two months, with Europe recovering faster than the rest of the world, then I have the story backwards and this really was a demand wobble with a customs overlay, and the right response for a mid-sized seller is a hybrid: keep a small parcel lane open even at lousy economics, because rebuilding the muscle later costs more than carrying it now.

There is also a fourth possibility that nobody wants to say out loud: Europe-bound capacity could get so thin that the flights you would rely on for genuine air-bulk restocking stop being daily. Tonnage falling off a lane rarely shows up first as a price cut. It shows up as frequency, and frequency is what wrecks a weekly replenishment rhythm.

Numbers now, because arguing without them is how people end up shipping at a loss and blaming the market. Assume a seller doing three hundred thousand parcels a month into the European Union, average declared value twelve euros, average chargeable weight half a kilo. That is one hundred and fifty tonnes a month and three point six million euros of declared value. Everything after this line is my assumption, not a published figure: last-mile delivery inside Europe two euros twenty per parcel, direct line-haul four euros a kilo, European pick and pack one euro ten per unit, local two-day delivery two euros sixty per unit, bulk air into a European warehouse three euros twenty a kilo, duty on the bulk entry four percent of factory value, factory value set at half of retail, so six euros a unit, and import value-added tax left out because we are assuming a properly registered importer who recovers it.

Under the parcel model each unit costs three euros of duty, two euros twenty of delivery and two euros of line-haul: seven euros twenty, or two million one hundred and sixty thousand euros a month, sixty percent of declared retail value. Under the bulk model: four hundred and eighty thousand euros of line-haul for one hundred and fifty tonnes, seventy-two thousand euros of duty on one point eight million of factory value, three hundred and thirty thousand of pick and pack, seven hundred and eighty thousand of local delivery, totalling one million six hundred and sixty-two thousand euros, forty-six percent of declared value. Roughly half a million euros a month better, with the shopper getting it in two days instead of nine. That is the engine behind every number in the data, and no amount of complaining about a three-euro charge moves arithmetic like that.

Now put the same example under stress, which is where the arithmetic turns mean. Say the European demand forecast is wrong by twenty percent and you are sitting on one point eight million euros of stock at factory prices. At a carrying assumption of one point two percent a month for storage and money together, that is about twenty-one thousand six hundred euros a month for as long as you keep being wrong. Say the slow twenty percent eventually moves at half price: that gives back nine hundred thousand euros of margin in one season, which is roughly two months of everything you saved on freight.

The bulk model wins, but it wins conditional on the forecast holding. When the forecast stops holding, it is worse than the model you walked away from, because the parcel model let you be wrong in small daily instalments while the bulk model makes you wrong in one crate-sized bet. Which is another way of saying this migration is really a forecasting discipline problem wearing a customs costume.

So turn the numbers into rules rather than opinions. Three thresholds drop straight out of the duty arithmetic. Below roughly nine euros declared value, three euros is more than a third of the ticket and those items should not fly one at a time at all; they belong inside a bulk shipment or nowhere. Between nine and twenty the duty is survivable but there is no margin left for a failed first delivery attempt, so keep those flying with a hard cap on landed cost per unit. Above twenty the old habit still works. On the warehouse side, convert time into cover instead of counting cartons: with a forty-five day sea pipeline and a fifteen day demand buffer your reorder point sits at sixty days of forward cover, and the order fires when projected cover falls under seventy-five days, not when somebody walks past the racking and feels uneasy.

Anything selling faster than forecast keeps an air option with a ten day trigger: once projected cover drops below ten days, ship thirty days of volume by air and let the sea shipment land behind it. Writing those two triggers into whatever system you already pay for takes an afternoon and deletes the most expensive kind of decision there is.

Before any of that, there is this week. Get the reconciliation out of your broker, in writing, for the last eight weeks: rows declared, parcels declared, duty remitted, with dates against each. If those three cannot be tied together, do not scale anything until they can. Open your order book today and count the lines with a delivery address inside the European Union and a ticket value under nine euros, because those are the ones whose economics broke on the first of July and nobody has told those customers anything yet. Send the message before the complaint arrives: a line saying the item now ships from a European warehouse and will be there in three days costs you nothing, and a complaint costs you the next order.

Check the receiving performance at your European warehouse instead of believing the brochure: ask for actual putaway hours by week for the past month, and if the number is above forty-eight hours on more than one day in five, get it into the contract with a credit attached. And do something about returns this month rather than in January. Returns are the bill nobody has costed in this migration, and every inventory decision you take now decides how ugly that bill looks when it lands.

I know how this reads from the inside. Everybody is arguing about a customs rule while the actual failure is going to be a truck that cannot get a dock slot in the second week of November. Fix the dock slot first, argue later.

↑ Back to the key points

— By Nadia Pryce