North Asian Ports Queue Three to Four Days as 3.57m TEU of Capacity Waits

The Analysis

Port updates dated November 5 put Busan New Port's BNCT yard density near 95 percent, with loaded boxes at roughly 100 percent. Ningbo Meishan and Yuandong YZCT report three to four day waits. Yangshan runs about three days, Waigaoqiao Phase Two three and a half days, Phase Five about two. Shekou sits at half a day to one day with density above 90 percent, Yantian's congestion worsened in week 45, Nansha is up to about a day and a half, Qingdao one to one and a half days. Globally about 3.57 million TEU, or 10.3 percent of the fleet, is tied up by congestion, and North Asia accounts for roughly 40 percent of that.

The headline number in that pile is the 3.57 million. The number that actually sets your freight bill is a different one, and it is not printed anywhere.

Congestion is not charged as a delay. It is charged as capacity that has been removed from the market and cannot be sold, and capacity is priced in days.

Start with why a port can be "congested" at 90 percent density and still move cargo perfectly well, because the word hides the mechanism.

A container yard is a buffer, not a warehouse. It exists to absorb the mismatch between when a ship discharges and when a truck or a train can take the box away. When density is at 60 or 70 percent, the yard does its job: boxes land, get sorted, get picked up, and the yard breathes. When density gets to 90 percent, the yard is close to full, and the operational consequences are not gradual. A yard at 90 percent cannot freely restack, because restacking requires empty slots to move boxes into. So a high-density yard stops being able to re-optimise, and every new discharge has to go wherever there is space rather than where it should go. That is the point at which a yard stops being a buffer and becomes a queue, and a queue at a port is measured in ship waiting time, which is exactly what the Ningbo and Yangshan numbers show.

So the 3.57 million TEU figure is not a snapshot of ships sitting still. It is the volume of vessel capacity that has been absorbed by waiting, and the only way to see what it costs is to multiply it by time.

Take the arithmetic. Three-point-five-seven million TEU of capacity tied up. A large container vessel is nominally in the range of, say, 18,000 to 24,000 TEU, so the number of ship-equivalents in the queue is the capacity divided by the ship size. Call it roughly 150 to 200 large ships' worth of capacity, sitting idle across the network. Now multiply by days. If the average box in that figure is tied up for three days, the capacity-days lost are 3.57 million multiplied by three, which is about 10.7 million TEU-days. Every one of those days is a day that capacity was not available to move revenue cargo. That is the true size of the congestion bill, and it never appears on an invoice; it appears as tighter supply, firmer rates, and blanked sailings.

Which is the pivot the release does not make. A vessel waiting three days outside Yangshan is not only late; it is unavailable for the next leg of its rotation for those three days. So congestion removes capacity twice: once by absorbing the waiting time, and again by forcing the carrier to pad the schedule to protect reliability, which absorbs even more effective capacity. That second effect is the one that changes the market. A carrier that has to build buffer days into a rotation to keep its on-time performance is a carrier with less effective capacity than its fleet list suggests, and the resulting tightness is what pushes spot rates in a season nobody expected to be tight.

Now the first thing the trade press skipped, and it is about where the 40 percent sits. North Asia taking 40 percent of the global congested capacity is not evenly spread through the region; it is concentrated in the two or three port clusters with the worst yard density. Busan at 95 percent density and 100 percent loaded-box density is a near-saturated yard, and a yard that saturated has lost its ability to sequence. Ningbo Meishan and Yuandong at three to four days of waiting are the operational expression of that saturation. So the practical read is that the congestion is not a region-wide fog; it is a set of specific chokepoints, and a shipper who is exposed to those chokepoints is in a very different position from one who is not, even though both read the same headline.

There is a second skipped point, and it is the one that should worry a planner most. Look at the dispersion inside the region. Shekou is at half a day to one day, Yantian worsened in week 45, Nansha is up to about a day and a half, Qingdao one to one and a half days, while Ningbo and Yangshan are at three days and Meishan is at three to four. The gap between the best and worst ports in the same region is four to five days. That gap is not noise; it is the shape of the problem. It means a shipper with routable flexibility can arbitrage the congestion by shifting the port of loading or the port of transhipment, and a shipper without that flexibility is fully exposed to whichever port their service happens to call. So the real risk here is not the average wait; it is your own concentration in a single congested cluster.

Then take the empty-box yard density, because it is the part of the picture that is genuinely new. Busan's loaded boxes are at about 100 percent density. That means the yard is full of loaded containers, not empties. A yard full of loaded boxes is a yard that cannot accept new export boxes efficiently, because export boxes need to be staged for loading and there is no room to stage them. So the same saturation that slows imports also throttles exports, and the two effects feed each other: full yards delay the discharge, delayed discharge delays the pickup, delayed pickup keeps the yard full. That is a closed loop, and once a yard is in it, the loop does not unwind on its own. It unwinds only when a carrier drops a call or blanks a sailing, which is the market's way of forcibly reducing the inflow.

Now put the number against the orderbook, because there is a fact in the release that changes the medium-term read. The orderbook is equal to about 47.8 percent of the existing fleet. That is an enormous pipeline of new capacity. So the congestion is happening against a backdrop of record newbuild deliveries still to come. That combination is unusual: near-term congestion with medium-term capacity abundance. Which means the congestion is a transient, driven by operations rather than by a structural shortage of ships, and the strategic question is not "should I pay more for capacity" but "how long does this transient last, and can I bridge it". A shipper who buys long-term capacity at a congestion premium is buying at the top of a transient. A shipper who bridges the transient with routing flexibility and schedule padding buys nothing at the top.

Now widen the frame to the seasonal driver, because the timing of the congestion matters. Week 45 is early November, and the pattern of the region's port delays is being set on the eve of the winter weather window. Northeast Asia in winter brings fog and high winds to the Yangtze and Bohai approaches, both of which slow pilotage, berthing and pilot boarding, and both of which turn a three-day wait into a five-day one without any change in cargo volume. So the congestion figures in this release are a fair-weather reading of a system that is about to enter its bad-weather season. The capacity tied up in the next two months is likely to be worse than the capacity tied up now, not because demand rises, but because the operational efficiency of the ports falls. That is the forecast nobody in the coverage put in writing.

Now take the carrier's response, because the response is what reaches your rate. A carrier facing three-to-four-day waits at its main load ports has three levers: blank a sailing to rebalance the rotation, add buffer days to the schedule to protect on-time delivery, or raise rates to price the loss. In practice, carriers do all three, in that order, at different points in the cycle. So the congestion shows up first as blanked sailings, then as longer published transit times, and later as a firmer spot market. A shipper watching for the rate increase is watching the last lever; a shipper watching the blanking and the schedule padding sees the pressure before it hits the price.

Which brings the argument to a boundary, because the exposure is not the same for every shipper. If your cargo moves on a service that calls one of the saturated clusters, you are directly exposed and your first action is to test whether an alternative port of loading or transhipment, in the same region but at a lower-density cluster, is available. If your cargo moves on a service that calls the less congested ports, you are partly insulated and your action is to keep your routing away from the saturated clusters as the weather season arrives. If you are a buyer of space on a spot basis, you are exposed to the firming market and should consider locking a share of your requirement before the weather window, not after. If you are committed to a fixed rotation for the season, your exposure is capped but your flexibility is gone, and the correct mitigation is to build a delay allowance into your customer promises rather than to hope the port recovers.

Add a short note on transhipment, because the region's hubs are not all origin and destination ports. A large share of the boxes moving through Yangshan, Busan and Ningbo are transhipment boxes, connecting a feeder service to an ocean service. For those boxes, congestion at the hub is doubly damaging, because the box has to be discharged from the feeder, held, and then loaded onto the ocean vessel, and every one of those steps needs yard capacity and a free berth. So a saturated hub does not just delay the ocean leg; it delays the feeder connection, which pushes the delay back up the feeder network to smaller ports far from the congestion. That is how a problem at one big hub becomes a problem across a whole region of secondary ports, and it is the reason a shipper at a small feeder port should read the Yangshan number as if it were their own.

Now detail the congestion-cost model for a single container, because the port-level numbers are the least useful way to think about your exposure. A container caught in a three-to-four-day wait does not merely arrive late; it sits without moving, which for the shipper means demurrage or detention exposure, missed trucking appointments, and in some cases a missed customer delivery. So the cost of congestion for one box is the delay cost plus the storage cost plus the downstream re-planning cost, and the delay cost is usually the smallest of the three. That is why a shipper who manages congestion by negotiating for more free time is solving the wrong part of the problem; the free time helps with storage, but it does nothing for the missed delivery. The lever that matters is routing, and routing is decided before the container is loaded, not after it arrives.

There is one more mechanism worth naming, because it is the reason congestion is self-reinforcing at the network level rather than just at one port. Vessels are deployed in strings, and a string is a fixed rotation of ports served by a fixed set of ships. When one port on the string slows down, the delay propagates to every subsequent port on the string, so the whole string runs late even though only one port was congested. To recover, the carrier either speeds up on the sea legs, which costs fuel, or cuts a port call, which strands the cargo booked to that port. Both responses are expensive, and both are paid for eventually by the shipper. So congestion at one saturated cluster is not a local problem; it is a problem that a fixed string rotation spreads across the whole service, which is why the numbers at Yangshan and Ningbo matter to a shipper whose boxes never touch either port.

Let me also put the orderbook fact to work in the opposite direction, because a 47.8 percent orderbook is a double-edged number. It tells you that capacity is coming, which caps how long a congestion premium can last. But it also tells you that the market is going to be oversupplied in a year or two, which will push carriers toward larger vessels and consortium consolidation, both of which tend to concentrate calls on fewer, larger hubs. And larger hubs are exactly the ports that saturate first. So the structural trend out of this orderbook is toward fewer, bigger port calls, which makes congestion at the surviving hubs more likely, not less. The near-term congestion is transient; the medium-term concentration of calls is not. A shipper building a five-year route strategy should plan for the second even while hedging the first.

Now the boundary condition as an operational rule, because a planner needs a trigger, not a lecture. Set your own density threshold and your own wait threshold for each port you depend on, and pre-decide what you will do when a port crosses it: reroute to the alternative cluster, shift a share of volume to a different service, or extend the customer promise. A port at 90 percent density is a port that will be at 95 percent within a week during the surge, and a shipper who decides the response in advance avoids the worst of the scramble. The value of the threshold is that it moves the decision out of the moment of stress and into a calm planning conversation.

Then lay out who wins and who does not, because congestion redistributes as well as costs. Ports in the region that are running at lower density win, because they can absorb diverted volume and charge for the privilege. Carriers that can blank sailings and re-route win, because they get to price a tighter market. Shippers with routing flexibility win, because they can move away from the chokepoints. Shippers locked into a saturated rotation lose, and they lose the most if their delays fall on a seasonal delivery. And the end consumer loses last and least visibly, through higher prices and longer lead times that nobody bothers to attribute to a yard that was 95 percent full in early November.

How I would act on it.

Map your exposure to the specific saturated clusters within 30 days: list each of your active services against the port density and wait numbers in this release, and flag every service that calls Busan New Port, Ningbo Meishan or Yangshan as a high-exposure service.

Then qualify at least one alternate port of loading or transhipment at a lower-density cluster in the same region, and test it with a live routing rather than a quote, so that the alternative is proven before you need it.

Then pull forward your winter planning, because the fog and wind window will worsen the waits, and lock the capacity you need for December deliveries before the weather season, not during it.

Then build the congestion delay into your customer promises for any cargo moving through the saturated clusters, and price the delay allowance into your quotes rather than absorbing it later.

And watch the blanking and the schedule padding, not the rate, because those are the two levers carriers pull first, and they give you several weeks of warning before the congestion reaches your invoice.

One last thing about how to read congestion data. A wait figure is a symptom, and yard density is the cause. The number to hold on to is the density number, because density predicts the wait. Once a yard is near 95 percent, the wait is not a matter of if but of how long, and the shipper who is watching density rather than waits is watching the illness rather than the fever. That is the difference between reacting to congestion and planning around it, and in a market with a 47.8 percent orderbook, planning around it is the only move that pays.

↑ Back to the key points

— By Tomás Ferreira