Port of Los Angeles Projects Over 900,000 TEUs in June and July, Calling It a Window of Stability for Import Flows

The Analysis

Put the map on the table before anything else. On the eastern shore of San Pedro Bay sits a harbour that has spent a century learning how to be the place where American imports arrive, and on 18 June its executive director, Gene Seroka, said it expects to handle more than 900,000 container units in June and more than 900,000 in July, and called the period a window of stability for import flows. May finished at nearly 840,000 TEU, a figure the port says has been surpassed only in the pandemic surge years. Year-to-date volume has passed 4.1 million TEU, about 1% ahead of last year and about 1% ahead of the five-year average. Port Optimizer's import booking data pointed to another strong June, no significant congestion was identified, and the harbour maintenance and seismic resiliency work funded by the US Army Corps of Engineers goes on.

I want to argue with the word window. A port can be perfectly stable at thirty thousand boxes a day and your fifty can still sit for four days, because stability is a property of an aggregate and sitting is a property of a particular box. Stability at the harbour gate is not the same promise as smoothness at your dock door, and the gap between those two sentences is where a second-half plan either holds or does not.

What follows is an attempt to convert a harbour-level number into a shipper-level one. How much headroom that one per cent really is, what has to break for the window to close, and what you can agree with your own partners this month that will still be standing in September.

Start with what was said, because the sentence has travelled further than the numbers behind it. More than 900,000 container units in June, more than 900,000 in July, described as a window of stability for import flows. Behind it sits a May of nearly 840,000 TEU that the port says has been beaten only in the pandemic surge years, a year-to-date total past 4.1 million TEU running about 1% ahead of both last year and the five-year average, import booking data from Port Optimizer pointing to a strong June, coordination among terminal operators, truckers and rail partners, no significant congestion or operational slowdown identified, and continuing investment in harbour maintenance and seismic resiliency paid for by the US Army Corps of Engineers. That is the whole of it, and it is worth reading as a whole rather than as a headline, because the parts of it do not all say the same thing.

Look at the shape of the sentence rather than its size. It is a projection with a hard edge on it: June, July, and then nothing. Nobody has guided August, and nobody appears to have been asked. A window is only as useful as its frame, and this frame is two months wide. A shipper building a second-half plan on it is building on a number that stops talking at the end of July, at exactly the point in the calendar when import programmes for the autumn begin to arrive in volume. The port is not being evasive. It is telling you what it can see, and what it can see ends two months out.

Nine hundred thousand containers in a month is about thirty thousand a day, and if the harbour works something like twenty-six days out of thirty, that is close to thirty-four thousand six hundred on a working day. Hold that figure for a moment, because it explains both the comfort and the danger in the word stability. Thirty thousand boxes a day is a throughput the system knows how to deliver; it is also a throughput at which any individual box is a rounding error. Whatever this harbour promises about congestion is a promise about the behaviour of thirty thousand, and it carries almost no information about the behaviour of your fifty.

So take the one per cent apart, because it is the only figure in the report that speaks to headroom. Year-to-date volume has passed 4.1 million TEU and is about 1% above the five-year average. Divide 4.1 million by 1.01 and the baseline the port is being measured against is roughly 4.06 million, so the gap is about forty thousand boxes accumulated since January. Stretch that across the five and a half months from the start of the year to the middle of June and the port has been running somewhere near 7,400 boxes a month above what it has handled routinely for five years. Seven thousand four hundred boxes a month, on a base of nine hundred thousand: the harbour is running at about one hundred and one per cent of a level it has demonstrated it can hold. That is not slack. It is a demonstrated capability with almost nothing on top of it, and the difference between those two descriptions is the whole question.

Now look at the step rather than the level, because steps are what break systems. May came in at nearly 840,000 TEU and June is projected above 900,000, which is about sixty thousand boxes more in a single month, close to a 7% jump. July is projected at the same level again. Berths absorb a step like that reasonably well; a crane is a crane and a yard is a yard, and a terminal that has handled thirty thousand a day can usually handle thirty-two. The things that absorb it last are the things that are not bolted down: chassis, rail cars, warehouse space, driver hours, appointment slots. Those are shared pools, and a shared pool does not care what the average is. It cares what the peak is. Which is why the honest translation of this report for a shipper is not that the port is stable. It is that the port is stable at the berth and the pressure, if there is any, will show up inland first.

Let me put my own numbers against it, with the assumptions written out so you can argue with them. Assume you are an importer bringing 150 TEU a month through this harbour. Assume your distribution centre can receive, unload and put away ten containers a day, five days a week, which is twenty-one receiving days in a month. Assume your vessel schedule delivers that 150 in three clusters a month rather than evenly, because that is what vessel schedules do. Assume no diversion, no transloading and no change in your mode split.

Spread evenly, 150 over twenty-one days is about seven containers a day, and you have three doors a day of slack. That is the version of your operation that a harbour-level report describes. Now run the clustered version: three clusters of roughly fifty boxes each, and a trucker who pulls each cluster over three days because that is when the appointments are available. Fifty over three days is about seventeen a day, which is seven a day beyond what the building can take. Three days of that is roughly twenty boxes of backlog carried straight into the next cluster, which arrives before the last one has cleared. By the third cluster you are not receiving cargo, you are queueing it. Nothing about the port caused this. The port never touched it. Your boxes will sit in a yard or on a chassis while the harbour reports that everything is flowing smoothly, and both statements will be true.

Here is the arithmetic that puts the two scales in the same room. Your 150 boxes a month is 0.0167% of the port's month, and 1,800 boxes a year against a harbour that moves more than four million in half a year. A one per cent wobble in the port's monthly volume is about nine thousand boxes. Nine thousand is five times everything you will import in a year. The port's number can move by an amount that would be invisible in its own reporting and still be five times your annual programme. That is not an argument for ignoring the port's data. It is an argument for measuring yourself, because the aggregate you are being handed is five orders of magnitude away from the thing you are trying to manage.

So measure yourself, and put thresholds on it, because a number nobody owns is a number nobody acts on. Three figures are worth tracking weekly from this month on. Discharge to gate-out: I would want ninety per cent of my boxes out of the terminal within three days of discharge, and I would want to know within a week if that slips to four. Gate-out to arrival at my dock: ninety-five per cent within forty-eight hours, because that is the leg where chassis and appointment failures hide. Dry runs, meaning a truck that arrives without a chassis or leaves without a box: one per cent of pulls or less, and every one of them logged. If you use rail, add a fourth: ninety-five per cent of boxes off the ramp within twenty-four hours of availability. Pick your own numbers if mine do not fit, but pick them now, write them down, and give each one a name next to it.

Now zoom out, because the reason this harbour matters is older than this report and larger than this number. Ten years ago this was a different kind of place, and I do not mean that romantically. The water is the same, the channels are the same, and the cranes are newer versions of the same machines. What actually changed is the network behind the water: the dedicated rail corridor that runs from the harbour up to the downtown rail yards, the warehouse sheds an hour inland where a box becomes inventory, the chassis pools, the appointment systems, the booking data that lets a port see arrivals weeks before they land. Throughput numbers are the visible part. The network is the part that decides whether a throughput number is comfortable or unbearable, and the network is what gets rebuilt slowly, in decade-sized pieces, usually while the harbour keeps working.

Which is why the last line of this report deserves more attention than it usually gets. Harbour maintenance and seismic resiliency upgrades, funded by the US Army Corps of Engineers, continue. Dredging and seismic retrofit are not seasonal events; they are the sort of work measured in years, and they happen inside a working harbour. A berth taken out for a season, a yard block closed for a year of piling work, a channel that is a metre shallower than the drawings say while the dredge is somewhere else: none of that appears as congestion, because nobody counts planned work as a disruption. It appears as less room. A harbour running at one hundred and one per cent of its five-year baseline has less room to give. If you want to know where that work is and when, ask now, in June, while there is no crisis and somebody has time to answer a letter.

There is one instrument in this report that points forward rather than backward, and it is the one I would build on if I were in your chair. Port Optimizer tracks import bookings destined for Los Angeles, and that is what the port used to forecast a strong June. A booking is made weeks before a box is discharged, so the booking series leads the throughput series by roughly a voyage, and it is the only genuine early warning in this story. You do not need the port's feed to use the idea. Ask your carrier and your forwarder for booking or forecast numbers on your own lanes, weekly, in containers rather than in percentages, and compare them against what actually gates out. Three or four weeks of that gives you your own lead time, and your lead time is what decides how early the warehouse and the chassis have to be booked. Everything else here describes something that already happened.

Then there is the cooperation sentence, which I read differently from most people. Terminal operators, truckers and rail partners working together is given as support for the forecast, and it should be. Cooperation is also a property of a system that is not under stress. Everyone coordinates beautifully at one hundred and one per cent of a five-year average. The question that matters is what coordination looks like at one hundred and fifteen, and nobody can tell you that in advance because nobody has run the experiment this year. The practical response is not cynicism, it is pre-commitment: get the commitments you need while cooperation is cheap and the system has room, which is now, rather than in the week when everybody's phone is busy.

So what would I actually agree this month. Start with the calendar, because arrival cadence is the one variable here you fully control. Cap daily pulls at eight containers, which is eighty per cent of my assumed ten-a-day capacity, and hold that cap for the rest of the summer. On 150 boxes a month that needs roughly nineteen receiving days out of twenty-one, which fits; on three clusters it means pulling each cluster over about six days instead of three, which is the entire difference between a building that keeps up and a building that falls behind. Book the appointments that make that possible rather than hoping the schedule cooperates. Confirm chassis arrangements in writing for the weeks you know will be heavy, and ask specifically what happens if the pool runs short, because the answer to that question is different in July than it will be in October.

Then decide the mode question at booking rather than at arrival. If your dock can receive a box within twenty-four hours, local dray is right. If it cannot, the box should be sitting somewhere that is not your dock and not a chassis: on rail, or in a yard, and you need to have agreed in advance how many free days you have in each of those places and what each additional day costs you. Those numbers are yours to collect, not mine to guess. Write them on one page and give it to whoever books, so the decision is made by rule and not by whoever happens to answer the phone that morning.

Then pre-commit the warehouse. If the autumn programme lands above the June and July pace, the constraint in your operation will be floor space before it is anything else, and floor space is booked in weeks, not days. Reserve the overflow space now for the two months you expect to be heaviest, on the assumption that they will be heavier than June. And pre-qualify a second gateway: not because this harbour has given you a reason to leave it, but because an option that has never been priced cannot be exercised in a week. Ten per cent of your volume costed through another West Coast gate, or through an all-water routing to an Eastern or Gulf port, with transit times and inland costs written down, is an insurance policy that costs you an afternoon.

When does the window close. I can name the conditions without predicting them, which is the only honest way to do it. It closes if a seasonal peak arrives on top of a base that is already at one hundred and one per cent, because the shared pools I described have no reserve in them. It closes if rail car supply or the inland ramp tightens while volume is at this level, because the rail leg is the one with the fewest substitutes. It closes if the maintenance and seismic work takes a berth or a yard block out during a heavy month, because planned work and peak demand are scheduled by different people who do not talk to each other. It closes if chassis pools thin out, which historically happens not because boxes disappear but because they sit. And it closes on 1 August in a purely administrative sense, because that is where the projection stops and nothing has been said about what comes after.

The ways this goes wrong are mostly boring, which is why they get missed. Cadence bunching: three vessel clusters a month turning a comfortable seven a day into an impossible seventeen a day. Appointment drift: slots taken for a Tuesday and used on a Thursday, which quietly destroys whatever discipline you built. Chassis assumed rather than confirmed, discovered at seven in the morning by a driver who then bills you for the hour. Rail dwell tracked by nobody, so the box that should have left the ramp on Friday is found on Tuesday. Warehouse space treated as infinite until the week it is not. And the biggest one: reading no congestion identified as meaning no problem, when what it means is that the berth and the gate were moving, which is a statement about a harbour and not about your dock.

One last piece of arithmetic, because somebody will ask for a number to plan against. Take the year-to-date total of 4.1 million at mid-June, add roughly half of June's projected 900,000 to reach the end of the month, and you have about 4.55 million at 30 June. Add July at 900,000 and you have about 5.45 million. Assume the last five months run at 800,000 a month, below the projected pace, since the port has guided nothing past July, and the full year lands near 9.45 million. Assume instead that they hold at 900,000, and it lands near 9.95 million. Those two answers are half a million boxes apart, and nothing in this report tells you which one to use. Which is really the point. The window is a sentence about a harbour. Your budget is a sentence about your own dock doors, and the only person who can write it is you.

Harbours are patient things. They take decades to build, they get rebuilt a piece at a time while the ships keep coming, and they outlast every forecast made about them, including this one. What they cannot do is tell you what your next fifty boxes will do after they clear the gate. That part of the map is yours, and it is worth drawing properly while the water is calm.

↑ Back to the key points

— By Tomás Ferreira