Sea-Intelligence: Double Sailings Pushed Golden Week Capacity to a Record 393,000 TEU

The Analysis

The number is 393,000 TEU. That is what Sea-Intelligence counted as weekly deployed capacity on Asia to US west coast during the Golden Week window, against a normal figure of under 300,000 TEU for the holiday, and it is the highest on record for the transpacific and Asia-Europe trades taken together. Asia-Mediterranean ran 18% above the third quarter, Asia-US east coast 15%, Asia-North Europe 4%. Between 2012 and 2019 the same window did the opposite: capacity was cut by 8% on the US west coast, 6% on the east coast, 16% on North Europe and 21% on the Mediterranean.

Every one of those numbers is real and every one of them can be misread in the same direction. Deployed capacity is a schedule number, counted in slots on vessels that were on the water or booked to be. It is not a cargo number, and it is not a demand number. The reason it set a record is written in the same report: typhoon congestion bunched vessels together, and the delayed ships cascaded into October as double sailings.

Give me the number, not the feeling. So we take the 393,000 apart before we let anyone tell us what it means. The record is not a story about too much shipping capacity. It is a story about ships arriving late, and those two stories tell a shipper to do opposite things.

Start with the arithmetic, because the headline hides most of it. Take the holiday normal at 300,000 TEU, which is the top of the under-300,000 range the report gives. Against 393,000 that is 93,000 TEU of extra nominal space in a single week, or 31% more. Now set that beside the 8% cut that used to happen in the same window between 2012 and 2019. The swing between cutting 8% and adding 31% is roughly 39 percentage points on one lane in one fortnight. Run the same comparison on the other three and the shape repeats: the Mediterranean moved from a 21% cut to an 18% increase, a 39 point swing; the US east coast from a 6% cut to a 15% increase, 21 points; North Europe from a 16% cut to a 4% increase, 20 points. Golden Week sits in weeks 40 to 42, and this year those three weeks were the fullest on record.

One more caliber question, and it is the boring one that decides whether a record is a record. Highest on record means highest since the series starts. If that series starts in 2012, then the 2012 to 2019 comparison and the record are the same dataset and everything is consistent. If it starts later or earlier, the two statements are measured over different windows and the 39 point swing calculated above is not comparing like with like. A press release never prints the series start date. It is one line in the methodology note. Ask for it, because a record measured over seven years and a record measured over fifteen are not the same claim, and you are about to make a booking decision on the strength of it.

Now convert the number into things you can actually book, with the assumptions written down where anyone can argue with them. Assume 70% of the TEU sits in 40ft equipment, so 1 TEU for a 20ft and 2 TEU for a 40ft, which gives 1.7 TEU per average box. At 393,000 TEU that is about 231,000 boxes of nominal space in one week. Assume an average Asia-US west coast vessel of 14,000 TEU and the week is roughly 28 sailings against about 21 at 300,000 TEU. Seven extra sailings in a week is what double sailings look like once you stop calling them a phrase and start counting them. And the 93,000 TEU of extra space is about 54,700 boxes that had to be filled by cargo nobody had in the plan.

Why it happened is stated plainly in the report and it is the part that gets lost. Carriers did not abandon capacity discipline. Typhoon congestion bunched vessels, the delayed ships cascaded into October, and two sailings that should have been a fortnight apart turned up in the same week. So the increase in deployed capacity is not a decision to add service. It is a schedule failure showing up on the supply side of the ledger. For eight years the industry took capacity out of this window because factories were shut and there was nothing to carry. This year the capacity arrived anyway, late, in a lump, and it arrived carrying cargo that was supposed to have sailed in September.

Set the four old cuts side by side and one more thing falls out. Between 2012 and 2019 the same window took 21% out of the Mediterranean, 16% out of North Europe, 8% out of the US west coast and 6% out of the east coast. Were blanking a uniform response to a Chinese public holiday, you would expect a similar percentage on every lane. The spread says carriers have always blanked hardest where the service is thinnest and the cargo most seasonal, and least where the network is dense and boxes keep moving through the shutdown. That reading matters this year because the same ranking reappears in the increases: the Mediterranean up 18% against the third quarter, North Europe up 4%. The lane that used to be cut hardest is now the lane running fullest. Either the seasonality changed, or the delay hit the thin lanes hardest, and a thin lane has far less slack to absorb a bunching than a dense one does.

Which brings us to the most important caliber question of the lot, and it is the one nobody asks. Deployed capacity counts the nominal slots on the vessel. It does not count whether those slots could be sold. A vessel that arrives six days late has nominal capacity in week 41 and almost no usable capacity in week 41, because the cargo for that sailing was booked three weeks earlier, has already been rolled once, and is now sitting on a berth somewhere behind schedule. Somewhere inside the 393,000 TEU is a block of space that is counted but not sellable in the week it is counted in. The number is correct. The denominator is wrong. Effective capacity is lower than deployed capacity whenever the schedule is broken, and the gap between the two is roughly the size of the delay.

Here is the piece that appears in neither the Sea-Intelligence note nor the headline written from it. Record deployed capacity is being read as abundant space, and abundant space is being read as cheaper freight, and cheaper freight is being read as a reason to book late. That chain has a broken link in the middle. What actually reaches your accounts is not the freight rate on a fuller lane. It is schedule reliability and arrival bunching. A lane running 18% above its third quarter is a lane where every box on it carries a wider arrival window, where three vessels that should have landed across a fortnight land in the same week, and where the yard, the truck appointment and the free time clock are all working against you at the same moment. The rate you might save by waiting is a line item. The demurrage, detention, missed delivery window and extra safety stock you take on by waiting is a different line item, and it is bigger.

Read the record as a record of lateness and the decisions invert. You book earlier rather than later, because the space you need is on a specific vessel rather than in a lane total. You book the sailing that is actually on time rather than the one with room, because in a double sailing week the second vessel is usually the one that was rolled and the one most likely to slip again. You negotiate free time before the bunching lands rather than after the invoice arrives. And you stop treating a three week window in weeks 40 to 42 as a trend you can extrapolate into the first quarter.

When does it land, and in what order. The bunching happened in weeks 40 to 42. The delayed ships cascaded into October, which means the double sailings land through the second half of October and into November, and the arrivals bunch at the destination end roughly a transit later, so late October through mid November on the transpacific and into early December on Asia-Europe and the Mediterranean. The rate effect is a pulse rather than a step. In the week the two sailings are both open for sale, the spot rate on that lane softens because there is genuinely more space chasing the same cargo. The week after, the delayed vessel has sailed, the schedule is short again until the next rotation catches up, and the rate rebounds. A shipper who sees record capacity, waits for the rate to fall, and books in the rebound week has taken all the schedule risk for none of the price benefit.

Who carries what, and how much. A pure spot shipper on the transpacific gets the softest rates of the quarter for as long as both sailings are open, and gets nothing after that: a two week benefit stuck onto a twelve month problem. A contract shipper sees no rate change at all and takes the entire hit on schedule reliability, and that group is usually the last to hear about it, because nobody telephones a customer to say a vessel will be six days late until it already is. The consignee carries the arrival bunching: yard density, truck appointment lead times stretching from a day to three, and the demurrage clock starting earlier than the plan assumed. Anyone running a distribution centre on a fixed labour plan carries the overtime, because three vessels landing in one week does not rearrange itself around the number of shifts you scheduled.

Put a delay bill next to a freight saving and see which one is bigger. Assume you move 20 40ft a month into a US west coast destination. Assume the bunching puts three vessels that should have arrived across two weeks into the same week. Assume your free time is four days of demurrage and five days of detention, that yard congestion adds three days before your boxes can be gated out, and demurrage runs at $150 a box a day with detention at $100 a box a day.

Every one of those is an assumption you must replace with your own contract terms, and most people never look them up. Three days of demurrage across 20 boxes is 20 x 3 x 150, which is $9,000. Now take the freight side: if the extra capacity pushed the spot rate down by $200 a 40ft on a month of 20 boxes, you saved $4,000. The saving is less than half the cost, and only one of the two numbers was in the headline.

The capacity side has its own arithmetic and it is just as unflattering. Filling 93,000 TEU of extra nominal space, about 54,700 boxes, requires week 40 to 42 cargo volumes to run 31% above the holiday normal in the quietest fortnight of the Chinese export year. Either utilisation fell, or the backlog filled the ships, and backlog cargo is delayed cargo rather than new cargo. That is why I would not read a 31% capacity increase as a 31% demand increase, and why the only way to settle it is to watch the load factor and the spot rate together over the next two weeks rather than to argue about the capacity number on its own.

What would make me wrong, and how you would know. If volumes genuinely rose, because importers front-loaded ahead of a tariff change or because the year end restock started six weeks early, then record deployed capacity met record demand, utilisation held, and rates do not fall at all. There is a clean test for that. Over the next two consecutive weeks, look at the spot rate on your lane next to the capacity number. Capacity at a record with the rate flat or up means demand absorbed the supply, and you should stop waiting and book.

Capacity at a record with the rate down more than 10% means it was bunching, and the soft window is real but short. The second thing to watch is blank sailings: once carriers start announcing blanks for November and December to reabsorb the capacity that arrived late, the glut is over and the direction reverses. Typhoon is a one-off exogenous shock. Schedules work the bunching through in something like four to six weeks, and what follows a lump of extra capacity is usually a hole.

Where is the line, and what do you do when it is crossed. For the rate side, the tripwire I would write down is this: if deployed capacity on your lane runs more than 10% above the third quarter average for two consecutive weeks while the all-in spot rate falls by less than 3% over the same fortnight, demand has absorbed the supply, so stop waiting and lock your November and December volume. If the rate falls more than 10% over a fortnight on the same capacity signal, take another week, but set a hard date of 20 November and lock whatever happens, because the year end restock and the January contract reset both sit on the other side of it. For the schedule side, the tripwire is simpler: if three or more vessels that should have arrived across two weeks are now arriving in the same week at your discharge port, apply for extended free time in writing before the vessels berth rather than after the charges start.

What to do this week, in order of how much money it touches. Check whether the vessel you are being offered is the on-time sailing or the rolled one, and ask for the actual berth history of the previous three calls, because in a double sailing week the second vessel is the one carrying the delay. Ask your carrier or NVOCC in writing how much free time you have on demurrage and detention at your discharge port and what it costs per day after that, since most shippers discover both numbers on an invoice. Restate your delivery promise to your customer now, in writing, with the extra days that bunching costs you, rather than explaining a missed window in December.

Stagger arrivals: if a month of volume is currently planned to land in one week, split it across two, because the constraint in a bunching week is not the vessel, it is the yard and the truck appointment on the other side. And separate the two signals running in the same week: record capacity pushing rates down on the transpacific and a fourteen day FAK increase being tested on Asia-Mediterranean are not contradictory, they are two lanes at two different points in the same cycle, and you should treat them lane by lane rather than as one market view.

The traps, since this is where the money usually goes. Treating deployed capacity as a demand signal is the first one and it drives every bad decision after it. Assuming double sailings mean spare space is the second, because the space you need is on one named vessel and the lane total has nothing to do with whether that vessel has room. Comparing freight rates without the demurrage and detention exposure attached is the third, and in a bunching month it is easily the most expensive. Booking a full month of arrivals into a single week because the rate looked good is the fourth, and it turns a freight saving into a warehousing bill.

Extrapolating a three week Golden Week window into a full year view is the fifth, and it is how people end up signing a twelve month contract off a fortnight of data. The sixth is the one I see most often: using record capacity as the argument in a rate negotiation while ignoring the fact that the same week's schedule reliability data is the thing that will actually cost the customer money.

A record is a record of something, and this one is a record of ships being late. The capacity did not arrive because the market wanted it, it arrived because the weather took it out of the week it belonged to and put it back two weeks later. Take the freight saving where it is genuinely on offer, on flexible volume in the week both sailings are open. On everything that has a delivery date attached, buy the sailing that is on time and pay for it, because the discount on the late one is not a discount, it is a delayed arrival with a price tag on it.

↑ Back to the key points

— By Vivian Zhao