Sea-Intelligence: Asia-North Europe Golden Week Capacity Up 27%, 12 Double Sailings

Sea-Intelligence counted 1.50 million TEU of scheduled capacity on Asia-North Europe across the four weeks around Golden Week. That is 27% more than the same weeks last year and 60% above the 2017 to 2019 average. Inside those four weeks the holiday week itself is up 63% year on year and the week after is up 39%. Twelve double sailings fall in that fortnight and not one blank sailing is planned. On Asia-Mediterranean the same exercise gives 1.14 million TEU, up 49% year on year and 141% above pre-pandemic.

My first reaction to a number like that is not space is loose. It is: what is in the numerator, and what is it standing on. Show me the number, not the feeling. So we take it apart before anybody books anything against it.

Start with the shape of the 27%, because an average over four weeks hides everything. One and a half million TEU across four weeks is 375,000 TEU a week. Now hold the two weekly figures against the four-week total. If the four weeks average 27% and two of those weeks are 63% and 39%, the remaining two carry four times 27 minus 63 minus 39, which is 6, or roughly 3% each. That arithmetic rests on one assumption, which I will state rather than bury: last year's weekly bases were roughly comparable. If they were not, the split moves. But on a like-for-like base, capacity up 27% across Golden Week actually means capacity up 63% and 39% in two weeks and almost flat in the other two.

That changes what the number is for. The average is carried entirely by a fortnight, and that fortnight is exactly when your October cargo is trying to sail. A shipper reading 27% spread evenly over a month and planning bookings on that basis has read a number that does not exist anywhere in the data. What exists is two very fat weeks and two normal ones.

Now the base. Last year's comparable weeks work out at 1.50 million divided by 1.27, about 1.18 million TEU. The 2017 to 2019 average works out at 1.50 million divided by 1.60, about 938,000 TEU. So last year was already about 26% above the pre-2019 average before this year added its 27% on top. Compound them, 1.26 times 1.27, and you land on 1.60, which is the 60% in the headline. The point of that small exercise is that the 60% is not one event. Some of it was already there last year. Only the last 319,000 TEU or so arrived this year, and that is the part we have to explain.

The report explains it, and the explanation is the interesting part. Capacity is up not because carriers added ships, but because delayed vessels have bunched into the ordinary weekly windows. A ship that should have sailed in one week sails in the next, so the next week counts it and the week after counts whatever follows it, and for a fortnight every affected service looks like it is running two vessels instead of one. Twelve double sailings in two weeks, six a week, is what that looks like when you count it.

Now put the reliability data next to it, and I want you to look at these two sets of numbers side by side rather than in sequence. Schedule reliability on Far East-Europe in August was 6%. Average delay for late vessels was 8.2 days. About 3 million TEU of capacity was sitting in congestion, which the report puts at 8.5% of the global fleet. Run that last one backwards: if 3 million is 8.5%, the fleet is roughly 35 million TEU. And note the scale against the lane you ship on. Three million TEU stuck still is twice the entire 1.50 million TEU of Asia-North Europe capacity scheduled across these four weeks. The capacity tied up doing nothing is double everything the trade is planning to sail.

One detail in the delay figure closes the loop, and once you see it the 27% stops being mysterious. The average late vessel is 8.2 days late. A week is seven days. An average lateness longer than a week means the average late vessel does not simply arrive late inside its own week: it lands in the following week's window. Do that across a fleet running at 6% reliability, which is to say across almost the entire fleet, and the following week collects not only its own sailings but most of the previous week's as well. That is a double sailing. It is not a decision anybody made. It is arithmetic performed by a schedule that stopped working.

Here is the part nobody has said about this report, and it is the one that matters most in practice. The 27% and the 6% are not two facts. They are one fact, counted twice. The same delayed vessels that pulled schedule reliability down to 6% and pushed average delay out to 8.2 days are the vessels that inflate the scheduled capacity count, because a late ship is counted in the window it actually sails rather than the window it was planned for. Nominal capacity goes up because the schedule broke.

Read that in the other direction and it gets worse. A capacity number that is high because the schedule is broken is not capacity you can book. It is capacity that looks bookable in the system and disappears at cut-off. If you have ever had a booking confirmed on a Monday and rolled on a Thursday without anyone being able to explain why, this is the mechanics of it: the slot you were sold was a late vessel wearing a scheduled vessel's name.

There is a second tell in the same report, and I would weight it almost as heavily. Zero blank sailings. Golden Week is the week carriers normally blank hardest, because Chinese export volume drops and there is nothing to fill the ships. Zero blank sailings in a holiday fortnight is not a sign of strength. It is the fingerprint of a disrupted schedule: the vessels that would have been blanked are the vessels running late, and they are now needed to cover the slots they missed. Ask yourself why nobody cancelled anything, and the 27% answers itself.

So who gets hurt, and by how much.

If you are a contract shipper with a named allocation on a service, your space is protected to the extent your contract protects it, which means read your contract this week rather than next month. The bunching does not take your box. What it takes is your transit time, because you are on one of two vessels in the window and you do not control which one sails.

If you are a spot shipper, which in practice covers most small and mid-sized exporters, you are at the back of that queue. When a double sailing collapses into a single sailing, the cargo that rolls is the cargo with the weakest commitment behind it. Twelve double sailings in a fortnight is twelve windows in which that can happen to you, and no amount of capacity in the headline changes the fact that only one of the two vessels needs to survive.

If you ship to the Mediterranean rather than North Europe, the exposure is larger, and the numbers say so without any interpretation required. North Europe: 1.50 million TEU, up 27% year on year, 60% above the 2017 to 2019 average. Mediterranean: 1.14 million TEU, up 49% year on year, 141% above pre-pandemic. Every one of those figures is the more extreme version of the same artefact, and extreme versions unwind harder. The Mediterranean is not a milder North Europe. It is the same problem with the volume turned up.

We can also see how far the map has moved under these numbers. North Europe at 60% above the 2017 to 2019 average implies those weeks used to run at about 938,000 TEU. The Mediterranean at 141% above pre-pandemic implies about 473,000 TEU. So the Mediterranean used to run at roughly half of North Europe's volume in these weeks, and is now running at about three quarters of it. A lane whose scheduled capacity has grown that much faster than its neighbour is a lane where the capacity number is least anchored to anything you would still call normal, which is another way of saying the Mediterranean figure deserves more suspicion, not less.

There is a mechanical consequence of double sailings that catches people out every time, and it has nothing to do with space. Two vessels in one week means two cut-offs. If you close your documentation against the later cut-off, which is what most people do because it buys them a day or two, you have quietly put yourself on the second vessel. And the second vessel is the one that disappears when the pair collapses. So the ordinary human preference for the later deadline is, in this particular fortnight, a preference for the weaker slot. Ask which vessel your cut-off corresponds to, in writing, before you let anyone relax a deadline on you.

When does any of this reach you. The report gives the window and it is short: once the backlog clears and Chinese volumes drop after the holiday, expect late cancellations and rolled cargo through late October. The mechanism runs inside the same fortnight. Right now the system has twelve double sailings and zero blank sailings, because nobody has cancelled anything yet. Once the delayed vessels have caught up, the second vessel in each pair has no reason to exist. At that point the blank sailings appear, and they appear late, which is the worst possible timing: announced after your cargo has closed, with your box already sitting at the terminal.

That puts the practical risk window in the second half of October and the decision window in the first half. Anything you book in the next seven to ten days is a booking made into the bunching, when the system looks fullest and is least reliable. Anything you book after the cancellations begin is a booking made into a thinner market, when the space is real but scarcer. Neither is comfortable. The first is the one where you get rolled, the second is the one where you pay for the space you actually get.

Let us put the reliability number on your own volume, with the assumption written down, because that is the rule here. Suppose you ship 40 high cubes a month from Asia to North Europe. At 6% schedule reliability, 40 boxes times 0.06 is 2.4 boxes. That is how many of your monthly boxes arrive on the date the schedule promised. The other 37.6 arrive late, and the average lateness is 8.2 days.

Now the part that reaches your balance sheet. Those 8.2 days are not an inconvenience, they are inventory that has to exist somewhere else. Take a nominal door-to-door cycle of 30 days, and I am flagging that as an assumption because yours will differ. Add 8.2 days and the pipeline runs 38.2 days, which is 27% longer. Pipeline inventory scales with pipeline length, so a pipeline that is 27% longer carries roughly 27% more stock in transit. If your safety stock is set at 15 days against a 30-day cycle, the same service level against a 38.2-day cycle wants 15 times 38.2 divided by 30, which is 19.1 days. Four extra days of safety stock, on every line you buy out of Asia, funded by you.

Multiply 8.2 days by 40 boxes and you get 328 box-days of lateness a month. That is the number to hand to whoever owns working capital, because it is the same money described in a unit they already work in. It is also the number that tells you this is not a freight rate problem at all. There is no rate in this story. Nothing here moves the price of the box. What moves is whether the box sails, and when.

That last point deserves its own paragraph, because it is where most planning goes wrong. A replenishment plan built on a nominal schedule assumes arrivals are periodic: roughly the same quantity, roughly the same interval, with variance around a mean you can plan against. At 6% reliability with an average delay of 8.2 days, arrivals are not periodic with variance. They are aperiodic with a tail. Two of your forty boxes behave like the plan, and the rest arrive somewhere along an eight-day tail whose shape nobody publishes. Planning against the mean of that distribution is how you end up with a warehouse that is either full or empty in alternating weeks, and the fix is not a better forecast. The fix is carrying the tail, which is what the four extra days of safety stock in the calculation above actually buy.

On to the other side, because a one-sided read is how people get hurt.

It is entirely possible that some of the 27% is real. Carriers do add ships, and a service can genuinely be upgraded. If part of this capacity is real supply, the bunching story explains only part of it, and the outcome flips: real extra capacity arriving into falling post-holiday Chinese volumes means soft rates, no rolling, and a booking market that gets easier rather than tighter. In that world the right move is not book early. It is do not over-commit, because you will be able to buy space cheaper in two weeks.

There is a clean way to tell the two apart and it costs nothing. Watch the blank sailings. Vessels that were merely delayed catch up, and once they have caught up the bunching disappears on its own. Vessels that were added stay in the rotation. So if the double sailings persist beyond the holiday fortnight and no blank sailings appear, at least some of the 27% was added capacity. If the double sailings vanish and blank sailings start showing up in the second half of October, it was bunching all along, and the warning about rolled cargo is exactly right. That is a two-week test with a free answer, and it is a better basis for a booking decision than any forecast.

The other thing that flips this is volume. The whole warning rests on Chinese export volumes dropping after the holiday. If volumes hold up instead, the backlog clears into a market that still needs the slots, the cancellations never come, and what you get is congestion at the European end, because 1.50 million TEU arriving inside four weeks has to be discharged somewhere. Your transit time suffers either way. Only the mechanism changes: rolled at origin, or queued at destination. If you are planning for one of those, you should have a view on the other.

A few ways to get this wrong, all of them common.

Treating 27% as slack. It is the opposite of slack. It is the number you get when a schedule is broken and the same ships get counted twice. Anyone who reads capacity up and concludes they can book later has the sign wrong, and the sign is the whole trade here.

Booking without a vessel name. If your confirmation says space confirmed and does not name a vessel and a voyage, you do not have a booking, you have an intention. In a fortnight carrying twelve double sailings, the gap between those two words is the gap between sailing and rolling.

Taking the second vessel in a double sailing. When a pair collapses into one, the second usually goes. If you are offered a choice, take the first, and if you are not offered a choice, ask which one you are on.

Giving customers the schedule date. At 6% reliability the schedule date is a fiction with a 94% chance of being wrong. Publish the schedule date plus 8.2 days, and if the customer relationship can carry it, publish a range rather than a date. A promise you miss is more expensive than a promise you set low.

Treating the Mediterranean as North Europe with a smaller number. It carries 49% year on year against 27%, and 141% above pre-pandemic against 60%. Book Mediterranean cargo earlier than North Europe cargo, not on the same day.

And the one I would put at the top of the list: doing nothing for ten days because there is plenty of capacity. There is not plenty of capacity. There is plenty of counted capacity, which is a different animal entirely, and it stops being counted the moment the delayed ships catch up.

My line on this one is about dates, not rates. There is nothing in this story to lock on price, so do not go looking for a rate to lock. What you lock is the vessel: get a name and a voyage on every October booking by 8 October, recalculate safety stock against a 38.2-day pipeline by the end of this week, and put a weekly blank-sailing check on the calendar from 20 October. Three dates, none of which requires guessing where the market goes next.

Sea-Intelligence did the counting and the count is right. What I would add is that a capacity number built out of delayed vessels is a schedule problem wearing a capacity costume. The costume is what shows up in the headline. The schedule problem is what rolls your box.

• 10 月 8 日前把 10 月中下旬出运的亚欧订舱确认到具体船名航次，不接受只写已配舱没有船名的确认件。
• 遇上双班次的那一周优先订同一服务的第一条船，并要求书面写明被甩后的下一班船期与免收改单费。
• 本周内把安全库存按在途多 8.2 天重算：以 30 天周期为例，15 天安全库存提到 19 天后再下发参数。
• 10 月 20 日起每周一次查亚欧空班公告，连续两周出现空班即判定撤船开始，未订舱的货立刻提前到当周出运。
• 地中海线的货比北欧线提前 5 到 7 天订舱，同一窗口内地中海运力同比增 49%、较疫情前增 141%，被甩概率更高。
• 对客户与销售发出的预计到港时间一律按船期表加 8.2 天，不再发送船期表上的名义日期。