AIS: 78% of Ships at Los Angeles Stay Over 48 Hours, Long Beach 53%

The Analysis

Spread the map out and put a finger on San Pedro Bay. Two ports, one harbour, the busiest gateway on the trans-Pacific, and this month a queue that is measured not in weeks of backlog but in the shape of the wait itself. The AIS count through 4 November is stark. Over the previous seven days Los Angeles took in eighteen container ships, and fourteen of them were still in port beyond forty-eight hours, which is seventy-eight per cent. Long Beach took nineteen, ten of them past forty-eight hours, fifty-three per cent. Both ports are flagged as congested. The median ship at Los Angeles sits about a hundred and five hours. The median at Long Beach sits about seventy-two.

Those two numbers are the reason I do not read this as one congestion event with one cure. A thirty-three-hour gap in the median between two ports that share an anchorage, a channel and a labour pool is not noise. It is structure. Before anyone books a diversion or takes a rate on the strength of a headline, that gap is the thing to explain, because the cure depends entirely on what is causing it.

And the first thing to be clear about is that the anchorage is only the front of the queue. A ship waiting to berth is the visible part. Behind it there is a second queue, on land, and the two together are what your customer experiences as a late delivery. The port tells you about the water. Nobody's dashboard tells you about the sum.

Spread the map out and put a finger on San Pedro Bay. Two ports, one harbour, the busiest gateway on the trans-Pacific, and this month a queue that is measured not in weeks of backlog but in the shape of the wait itself. The AIS count through 4 November is stark. Over the previous seven days Los Angeles took in eighteen container ships, and fourteen of them were still in port beyond forty-eight hours, which is seventy-eight per cent. Long Beach took nineteen, ten of them past forty-eight hours, fifty-three per cent. Both ports are flagged as congested. The median ship at Los Angeles sits about a hundred and five hours. The median at Long Beach sits about seventy-two.

Those two numbers are the reason I do not read this as one congestion event with one cure. A thirty-three-hour gap in the median between two ports that share an anchorage, a channel and a labour pool is not noise. It is structure. Before anyone books a diversion or takes a rate on the strength of a headline, that gap is the thing to explain, because the cure depends entirely on what is causing it.

And the first thing to be clear about is that the anchorage is only the front of the queue. A ship waiting to berth is the visible part. Behind it there is a second queue, on land, and the two together are what your customer experiences as a late delivery. The port tells you about the water. Nobody's dashboard tells you about the sum.

The standard reading of a headline like seventy-eight per cent past forty-eight hours is that Los Angeles is congested and Los Angeles should be avoided. I want to slow that down, because past forty-eight hours is a low bar and it flatters nothing. Forty-eight hours is two days. On a trans-Pacific string that calls Los Angeles first, discharges, and then works a second West Coast port before turning back, a two-day spread between first and second call is well within normal for a large vessel with a heavy exchange. Count how many ships exceeded two days and you will always get a high number at a first-call port. That does not mean the number is meaningless. It means it is measuring the wrong thing to sort ports by.

The number that sorts ports is the median, and the medians here are a hundred and five hours at Los Angeles against seventy-two at Long Beach. Put them in units a dispatcher uses. A hundred and five hours is four and a third days. Seventy-two hours is three days even. So the average ship at Los Angeles spends roughly a day and a third longer from arrival to departure than its counterpart a few miles across the harbour. That is the spread to interrogate, and it is almost never about which port is bigger or busier. It is about which ships go where.

Three structural reasons explain most of a gap like that, and all three are visible if you know your service. The first is vessel size and exchange profile. First-call ports on the largest trans-Pacific strings take the deepest discharge lifts, and a heavier lift takes longer at the berth regardless of how well the terminal is run. The second is terminal and berth configuration. Some terminals in the complex run more of their volume over on-dock rail; some are more truck-heavy; some have more or fewer berths working a given ship. The third is what happens after discharge, and this is the one the ship-side numbers hide. A ship is only "in port" for the purpose of this dataset until it departs. If a terminal is quick off the ship but slow onto the truck or the railcar, the ship departs on time and your container still sits, and the port looks fine while your delivery date slides.

That third mechanism is where the real damage to a delivery promise lives, and it is the point the coverage has not made. The AIS numbers are a ship dataset. Your commercial risk is a container dataset. The two diverge exactly at the rail stack. This month the congestion is not being driven by a shortage of ship capacity or even primarily by berth productivity. It is being driven by what happens after the container leaves the vessel: railcar supply into Southern California has been tight, on-dock rail stacks have been building, and empty-container repositioning is competing for the same chassis and the same yard slots as loaded imports. A ship can be discharged briskly and still hand you a container that cannot move inland, and nothing in the seven-day AIS count will have warned you.

Let me put a number on the door-to-door reality, with the assumptions written out, because this is where a schedule quietly becomes fiction. Take a lane where the carrier's proforma says fourteen days port to door. Assume a fourteen-day voyage, which is fair for South China direct. Add the anchorage wait at the median, which at Los Angeles is a hundred and five hours, about four and four-tenths days. Add the landside leg: berth to availability, then truck or railcar. Assume two days for pickup under normal conditions, and add a further two days this month because the rail stack and the chassis shortage are real constraints rather than cyclical grumbles. Fourteen plus four and four-tenths plus four is about twenty-two and a half days. That is not a fourteen-day lane with a small delay on it. It is a twenty-two-day lane being sold as a fourteen-day one, and the eight-and-a-half-day gap is the same figure in every quote you sign until you change what you put on paper.

The landside queue also has a date attached to it that is not in the AIS data at all. The Vincent Thomas Bridge, the link between San Pedro and Terminal Island and the main route for trucks serving the inner terminals, closes from 7 November for around sixteen months, with traffic diverted to the I-710 and Queensway Drive. Read that against the seven-day AIS window. The trucks that serve those terminals are about to take a longer route, which cuts the number of turns a drayage tractor can make in a day, which lengthens the pickup leg precisely when the rail stack is already making it longer. Two independent constraints, both on the same node, both landing in the same fortnight. The ship median may not move much. Your container dwell and your cartage rate will.

Then the tail risk, which is the part I would flag to anyone signing a service commitment this week. ILWU Local 26, the roughly two hundred and fifty security watchmen who control gate entry, patrol the yard and escort high-value cargo, have been negotiating a new contract, and the strike deadline has been pushed out to 5 November with federal mediation at the table. The arithmetic of that dispute is what makes it serious out of proportion to its size. Local 26 is about two per cent of the San Pedro Bay workforce, but its members work terminals that handle a large share of both ports' volume, and the wider ILWU body is not expected to cross a picket line.

So a walkout by two hundred and fifty people can stop vessel, yard and gate operations at those terminals, which turns a two-hour wait into a stopped queue. A settlement on the fifth does not clear the stack either. The rail backlog did not appear because of the labour dispute, and it will not disappear when the dispute ends, which means the recovery tail runs past the deadline whether or not the deadline is met.

I want to hold the two ports apart in one more way, because it changes the advice. The gap between a hundred and five hours and seventy-two hours is not a reason to prefer Long Beach over Los Angeles for every shipment. It is a reason to know which of your containers are routed where, and why. If your Los Angeles volume sits on a first-call string with a heavy exchange and a truck-heavy terminal, your exposure is systematically higher than a Long Beach box on a lighter string, and a blanket "shift to Long Beach" is as crude as a blanket "avoid Los Angeles". The structural question is which node in your own routing is the weak one, and that is answerable per lane, not per port.

There is also the eastern alternative, and it is the counterfactual worth testing. The Panama Canal has been restored to full operation, which means the all-water route to the US East and Gulf is no longer capacity-constrained the way it was, and the water-and-draft bottleneck there is cleared. But the canal restoration and the ongoing no-booking transit of three to six days are two different facts, and only the first one is a reason to relax. A shipper weighing a US West call against a US East one this month is comparing a West Coast ship queue that medians at four-plus days against a canal transit that can still cost three to six days without a slot, plus the longer all-water voyage. On the arithmetic alone, neither route is a free lunch, and the right question is which one you can plan around with a slot and a schedule rather than which one looks less congested in a headline.

Step back to the decade and the thing that is actually changing is not the queue. Queues come and go; this harbour has been congested before and will be again. What is changing is where the bottleneck sits in the network. For years the constraint on the trans-Pacific was on the water, capacity and sailings. This month the water is fine and the constraint has moved ashore, onto railcars, chassis, truck turns and bridge closures, all of it underneath the ports' own dashboards. When a gateway's binding constraint migrates from the terminal to the last fifty miles, the ships are not the variable you optimise. Your inland plan is. That is the pivot a port data series will never show you, and it is the one that decides whether your delivery promise holds.

And there is a scheduling lesson in the murky middle of this that a lot of importers get wrong. When people hear congestion they think about the ship and reach for a rate. This quarter the rate is not the binding constraint at all. The spot market on the trans-Pacific has been soft, capacity is ample, and blank sailings have been low, which is exactly why a shipper can look at the booking screen and conclude the system is calm. The system is calm on the water and tight on the shore, and those are two different markets with two different signals. The rate tells you about the ocean leg. It tells you nothing about whether your box can get off the dock, which is where this month's cost is actually landing. If you optimise your supply chain on the rate and ignore the shore, you will buy a cheap sail and pay an expensive dwell, and the two do not cancel out.

Put another way, the queue has changed owner but not length. Two years ago the constraint was whether you could get a slot on a ship. Now you can get the slot easily, and the constraint is whether the container can leave the terminal once it is discharged. That migration is the point. It moves the lever from procurement, which negotiates the rate, to operations, which manages the pickup, the chassis and the rail booking. If your organisation still treats a trans-Pacific lane as a rate negotiation with a fixed delivery window attached, you are managing the half of the problem that this month solved itself and ignoring the half that this month created.

There is also a repeat-purchase argument for fixing the inland side first, which is easy to miss in a single-month view. A rate move resets every time the market resets, but a landside process you build, the terminal-level routing map, the on-dock rail flag, the standby drayage, the early-pull discipline, keeps paying across every congestion cycle and every rate cycle. This month's numbers will be gone in a few weeks. The process you build to read them will still be there in the next peak. That is why I would spend the next fortnight on the shore, not on the rate sheet.

There is a further point about the two ports that almost no one puts on a slide, and it is the one I would take to a customer. When the median at one port is a day and a third longer than at the other, the two do not fail in the same way. A port with a heavy first-call exchange fails upstream, at the berth, where the fix is ship-side and largely outside your control. A port with quicker berthing but a clogged inland leg fails downstream, at the gate and the rail ramp, where the fix is yours to make. Reading the two medians as one number hides which failure mode you have bought. The importer who knows he is on the downstream-failure side can act this week, on pickup discipline and transload. The importer who thinks he is on the upstream side waits for the ship and waits for the port, and neither one is going to move for him.

There is also a crude and expensive mistake summarised by the phrase "just go somewhere else." Shifting a box to another gateway does not delete its shore leg. It relocates it. The container still has to clear a terminal, find a chassis, and reach a rail ramp or a truck, and the receiving gateway is now absorbing diverted volume into its own limited gate and rail capacity. A diversion made on the strength of a berth median, without checking the destination's inland stack, can trade a four-day ship wait for a three-day ship wait and a five-day landside wait, and come out worse. The test before any diversion is not whether the other port's ships move faster. It is whether the other port can move the box after the ship is gone.

What I would do, in order. Rebuild the door-to-door estimate for every trans-Pacific lane this month using the median anchorage figure rather than the proforma, and re-quote any customer commitment that still assumes the old number. Identify, terminal by terminal, which of your containers move on on-dock rail and which move by truck, and flag the on-dock rail boxes as the highest-risk cohort while the stack is building. Pull released containers off the terminals early and transload where it is cheaper than the dwell that is coming, and check whether your demurrage and detention clock pauses for labour action, because it does not. Confirm the terminal routing on every booking against the affected list before the fifth, and keep a drayage alternative on standby for anything that would fall into a Local 26 action. And when you weigh a reroute to the East Coast, price the canal's three-to-six-day no-slot transit into the comparison rather than treating the restoration as free capacity.

The map is not changing hands here. Los Angeles and Long Beach are still the gateway on this coast, they handle around forty per cent of American containerised imports between them, and they will for a long time yet. What has moved is the load, off the ships and onto the shore, from the water line to the rail ramp and the truck gate. Anybody still managing this gateway from the water line is watching the wrong half of the queue, and this month the wrong half is the expensive one.

↑ Back to the key points

— By Tomás Ferreira