Shanghai Backlog Runs Into November as Shekou and Vung Tau Yard Density Tops 90%

The Analysis

The weekly says this. Kuehne+Nagel's 27 October port update reports that Ningbo and Shanghai lifted their typhoon emergency responses on 26 October, yet Shanghai's 7-day average vessel wait is still about 1.65 days, with severe backlogs and congestion expected into November as bunching and blank sailings bite. Shekou averages 1.34 days with yard density above 90%. Vung Tau runs 1.5 days with off-window berth delays of up to four days and yard density above 90%. Rotterdam's MVII yard is 99% full and ECT 85%. And the line I care about most: confirm cut-offs by vessel, not by sailing week.

I read port weeklies for a living, and I never accept a number until I know its base, its scope and its effective range. This update is a better class of evidence than an organiser's press release — one named source, one dated snapshot, field observations — but it is still full of small traps. A 7-day average is not a snapshot. 'Above 90%' is not the same as 99%. 'Up to four days' is a tail, not a mean. None of these digits answers the question I actually get paid to answer, which is not 'how congested is Shanghai'.

The question I get paid to answer is how many days before cut-off I must stage my cargo. So I'm going to turn every one of these wait days and yard-density readings into that single number, port by port, with the assumptions stated. Then I'll give you the line past which you lock November space, and the hour you lock it. No 'freight is rising, stay tuned' — that sentence is empty and I won't write it.

Start by sorting the readings into what they actually measure, because mixing them is where most planning errors begin. 'Typhoon emergency response lifted on 26 October' is an administrative fact — a status flag changed on a certain date. It costs nothing and moves nothing. '7-day average vessel wait of 1.65 days' is a market observation, a rolling mean of how long ships sat before berthing. 'Yard density above 90%' is a stock reading, a fullness ratio at a point in time. 'Blank sailings and bunching into November' is a forecast. Four different kinds of statement, four different reliabilities, all stacked in one paragraph. Before I use any of them I label the type, because an administrative flag is not a throughput number and a forecast is not an observation.

There is a problem with this number, and it starts with the word average. Shanghai's 1.65 days is explicitly a 7-day average. An average hides the tail. If half the week's ships berthed in half a day and a handful waited four days, the mean still reads 1.65 and tells you nothing about the sailing you actually booked. Your container does not sail on the mean; it sails on one specific vessel with one specific berth window. For planning, the mean is a description of the past week, not a promise about your cut-off. So I never stage cargo to a 7-day average. I stage to the worst case the same report admits, and this report admits it plainly: off-window berth delays of up to four days at Vung Tau. Mean 1.5, tail 4, a gap of 2.5 days. That gap is the entire planning problem in one subtraction.

The next cut is on comparability, and it is the one people skip. Shanghai is labelled as a 7-day average; Shekou's 1.34 days and Vung Tau's 1.5 days are not labelled as anything. They may be single-day snapshots, they may be period averages, the update does not say. So lining them up and declaring 'Shanghai is worst, then Vung Tau, then Shekou' is arithmetic performed on three numbers that may not share a denominator. There is a problem with this number — not that it is wrong, but that its unit is undisclosed. A rate desk does not rank ports off unverified units. I treat 1.34 and 1.5 as rough readings with a wide error band and I plan off the tail, which is the only figure stated as a bound. When the unit is missing, the honest move is to widen your buffer, not to fine-tune a ranking.

Yard density deserves its own paragraph because it is the number that quietly governs everything else. 'Above 90%' — above 90% of what. Design slot capacity, usable ground slots, reefer plug positions, or available empty slots after boxes are stacked. Those are different denominators and they produce different operational behaviour. What I know from the field, and what the 90-versus-99 gap hints at, is that yard efficiency is not linear in density. Below roughly 80% a terminal finds and moves a box in a predictable number of moves. Past 90% the yard has to dig: containers are buried, a single retrieval turns into several restacks, gate-in queues lengthen, and the time from gate-in to a confirmed slot stops being proportional. Rotterdam MVII at 99% is not 9% worse than a yard at 90%; it is a different regime, one where a truck appointment and a retrieval window both become scarce. So density is not a score, it is a cliff, and 90% is the edge of it.

Why now, and not last week. The causal chain is the part the headline compresses. The typhoon response was lifted on 26 October, but a lifted flag does not un-berth the ships that waited at anchorage, does not un-stack the boxes already in the yard, and does not cancel the blank sailings carriers scheduled around the storm. During the closure vessels bunched up offshore; when the port reopened they arrived together, which is the bunching the update names. Carriers had already blanked sailings to protect schedules, which removes capacity exactly when a surge of delayed boxes needs it. Add peak-season export volume on top and you have four forces pushing the same direction: a physical backlog, simultaneous arrivals, withdrawn capacity, and seasonal demand. That is why a clean administrative headline — response lifted — sits next to a dirty forecast — severe congestion into November. The flag changed on the 26th; the boxes did not.

Who gets hit, ranked, not 'the industry is affected'. Shanghai export tail bookings on Europe and US lanes take the 1.65-day wait plus the November backlog, which is a schedule slip measured in days. Shekou shipments face the gate problem rather than the berth problem: with density above 90%, getting a box into the yard and confirmed before cut-off takes longer even when the ship is on time. Vung Tau is the sharpest exposure because a vessel landing outside its berth window can wait up to four days, and there is no buffer in a plan built on a 1.5-day mean. And then there is Rotterdam, which is not an origin problem at all: MVII at 99% and ECT at 85% are destination yards.

Importers there cannot pull boxes out and cannot return empties in. That matters to an Asian exporter two to three weeks later, when the empties that should have cycled back to Shanghai, Shekou or Vung Tau do not, and a confirmed booking suddenly has no box to release. The weekly prints 99% and 85% in the same paragraph as the Asian wait days but never connects them. I connect them, because a full destination yard is an origin equipment shortage with a fortnight of lag.

One more cut at the destination numbers, because 99% and 85% are not the same kind of reading either. MVII at 99% is a single yard effectively full; ECT at 85% is a different terminal in the same port with room to spare. Average them and you get a meaningless 92% that describes neither yard. Equipment does not release from an average, it releases from the specific terminal your specific box is sitting in. So I do not plan off a port-wide fullness figure; I ask which terminal the import box is dwelling at, and I treat the 99% yard as the binding constraint, not the average of the two. The tightest number in a set is the one that governs the queue, the same way the four-day off-window tail governs Vung Tau and not its 1.5-day mean. Plan to the binding number, never to the averaged spread, and half the surprises in this update disappear.

Follow the money, because congestion does not move the base rate and that keeps surprising people. Your contracted base ocean freight is unchanged on 1 November. What moves is the layer beneath it: congestion surcharges, peak-season and general rate increases, the premium on scarce space after blank sailings, rebooking spreads when a box is rolled, and demurrage that grows precisely because you staged cargo earlier to be safe. Carriers pass withdrawn capacity and berth-wait cost to shippers as surcharges. Yards pass dwell to shippers as demurrage. Staging earlier to protect the cut-off is the correct operational move and it also lengthens your dwell, so the safe choice has a bill attached. Read every quote as two columns — base and surcharge — and refuse to approve a revised all-in number until the changed line is named. A rate that went up without a named line is a rounding up, not a rate.

When it reaches you, in days. At T+0 the emergency flag is lifted but the physical queue is intact. Through the following week bunching and blank sailings push Shanghai's wait into the November backlog the update forecasts. Inside two weeks the cut-offs start moving, because a yard above 90% protects itself by pulling cut-offs forward. Two to four weeks out, two things arrive together: demurrage invoices from your early staging, and the equipment shortage feeding back from a 99%-full Rotterdam. The decision window is short and the cost window is long, again. You have the first 24 to 48 hours after this update to reset your staging plan; the bills from those decisions show up a month later.

Now the part worth paying for — turning these digits into the one number that runs a shipping desk: how many days before cut-off I stage cargo. Not by sailing date. By cut-off plus buffer. Start from the principle that cargo must be in the yard and confirmed before cut-off, so the buffer has to absorb three things — the berth wait that may shift the cut-off, the yard time that grows with density, and the tail if the vessel misses its window.

Shanghai: 1.65 days is a 7-day mean, so I refuse to plan to 1.65. I take the conservative 2 days, then add 1 to 2 days for the November backlog the same report forecasts. That puts cargo in the yard 3 to 4 days before cut-off. Shekou: 1.34 days of wait, but density above 90% means the gate-in-to-slot step roughly doubles against a normal yard, so I add 1 to 2 days of yard time on top of the wait. Again 3 to 4 days before cut-off. Vung Tau: 1.5 days of wait plus an off-window tail of up to 4 days, so if my vessel is not in its berth window I plan to the tail, not the mean — 5 to 6 days before cut-off, the widest buffer of the three. Those are the thresholds, and they are derived from the report's own numbers, not invented.

Turn the thresholds into two rules a planner can actually hold. Density is the first line: above 90%, add roughly half a day of staging buffer for every extra point of density you observe, because restacks grow nonlinearly past that edge. Wait time is the second line: if a port's wait moves from 1.5 days to 3 days, shift the whole staging window two days earlier rather than shaving a day here and there. And never mix the two — a berth-wait buffer and a yard-density buffer are separate additions. Past this line, lock it: when density holds above 90% for three straight days, or wait crosses two days, or a carrier issues a November GRI or congestion surcharge, stop fine-tuning the plan and lock November space and the surcharge cap that same day.

Here is the point the coverage did not write down, and it is the one that rolls boxes. Everyone watches wait days and yard density and assumes the risk is a late ship. A late ship is manageable; you can wait for it. The lethal thing is a cut-off that moves earlier. A yard above 90% is under pressure to free ground slots, and the standard lever is to pull cut-offs forward, not extend them. The update's own last line — confirm cut-offs by vessel, not by sailing week — is the tell. In a congested week, two vessels in the same sailing week carry two different cut-offs, and even the same vessel can carry a different cut-off at a different terminal, because each yard sets its own gate close against its own density.

Stage by sailing week and one of those bookings will miss a cut-off that quietly advanced by a few hours and get rolled. Your cargo was never delayed by 1.65 days; it was simply outside the gate when the clock moved. That is the exposure nobody headlines, and it is why I want the cut-off in writing per vessel name.

The disagreements are worth keeping because they show where the conclusion flips. The administrative reading says the typhoon response was lifted on 26 October, so things should be improving; the physical reading says severe congestion into November. Both hold, because one is a status flag and the other is a stock of ships and boxes. Flip the conditions and the verdict flips with them: if carriers restore the blanked sailings and bunching clears, 1.65 days can fall back under a day quickly. If another storm, or another peak-season surge, lands on a 90%-plus yard, then 1.65 days is the floor, not the peak. And the comparability problem stays live the whole time — if 1.34 and 1.5 are single-day readings rather than averages, then ranking the three ports off these three numbers is not analysis, it is three unverified units in a row. When the unit is doubtful, widen the buffer instead of sharpening the ranking.

Let's run the numbers on a real book, assumptions stated. Assume I ship 40 containers a month on Europe and US lanes, and assume three November first-week sailings, one each out of Shanghai, Shekou and Vung Tau. Applying the thresholds above, the Shanghai booking stages 3 to 4 days before cut-off, the Shekou booking 3 to 4 days, the Vung Tau booking 5 to 6 days if its vessel is outside the berth window. That is a spread of up to 3 days between the tightest and the widest booking, in the same week, for the same shipper.

Handle all three off one sailing-week rule and at least one of them stages too late. Now add the destination leg: MVII at 99% means empties cycle back slowly, so two to three weeks out my equipment release, not my berth wait, becomes the binding constraint. The honest plan stages to the widest of the three buffers, confirms each cut-off by vessel name, and re-checks equipment release at the fortnight mark. Every input there is a number from this update; the only things I supplied were my own 40 containers and the assumption of three sailings.

The alternatives each carry a price and the cheapest-looking one is the trap. Switching origin port to dodge a 90%-plus yard buys a cleaner gate but costs a different cut-off, extra inland legs and a fresh density reading you have not seen yet. Staging much earlier protects the cut-off and quietly grows demurrage, because dwell lengthened by choice is still dwell the yard bills. Pulling a booking forward to the prior sailing costs whatever spot premium blank sailings created. Air-freighting the urgent tail jumps an order of magnitude and only makes sense on genuinely high-value cargo. The common mistake is planning off the sailing date instead of the cut-off, the second is treating the lifted typhoon flag as the end of the backlog, and the quiet third is watching only the base rate while the surcharge layer does the damage. The most expensive mistake of all is staging by sailing week, because that is exactly the habit a forward-moving cut-off punishes.

So here is what I do, with times and owners. Within one working day of this 27 October update, I pull the cut-off by vessel name for every early-November booking I own and I write the staging date backward from that cut-off using the thresholds above — 3 to 4 days for Shanghai and Shekou, 5 to 6 for Vung Tau off-window. I re-confirm each cut-off 48 hours before gate close, because that is when a crowded yard advances it. On any booking whose yard density is above 90%, I get the gate-in and VGM and SI deadlines in writing, not by phone.

At the fortnight mark I re-check equipment release against the Rotterdam feedback loop, so a confirmed booking does not turn into no box. And before any of it I set the trigger: density above 90% for three straight days, wait above two days, or a November surcharge notice — any one of them, and I lock November space and the surcharge cap that day. Past this line, lock it.

One dry line to close, because the numbers deserve it. A 99%-full yard in Rotterdam and a 1.65-day wait in Shanghai are, two weeks apart, the same boxes queueing at two ends of the same string; the weekly printed both and joined neither. Cold comfort for whoever staged by sailing week. Watch the density, hold the cut-off by vessel name, and when wait crosses two days, lock. That is the whole job, and it is arithmetic, not weather.

↑ Back to the key points

— By Vivian Zhao