Shanghai waits 4.7 days, Ningbo 3.6 as post-typhoon backlog lingers at ports

The Analysis

Kuehne+Nagel's 2-8 September port update puts Shanghai at 4.72 days of average vessel wait, with the Waigaoqiao WGQ2 and WGQ5 terminals above 9 days. Ningbo averages 3.58 days after the 78-hour closure forced by Typhoon Saudel, and Yantian sits at 2.45 days. The number that matters is not the peak, it is that the backlog has not cleared: the queue the typhoon left behind is still sitting in the port.

Europe and the United States are on the same curve. Rotterdam's APMT Maasvlakte II yard is running at 85-90% utilisation, some Antwerp terminals lost time to the 4 September strike, and on the US west coast Los Angeles and Long Beach show only one vessel waiting but an 8-day rail dwell, with Savannah at six vessels. Barge and chassis supply stay tight, so inland distribution absorbs the delay that the berth no longer shows.

The five briefs below split the work by decision. Exporters need to move cut-offs and re-price quote validity. Import e-commerce needs safety stock measured in days, not in promises. Factories need a no-break buffer for inbound material and spares. Brands need a delivery promise that survives a five-day slip. Procurement needs a port mix and a force majeure clause that hold up when a typhoon closes a terminal for three days.

For Exporters

• By 18 September: pull the documentation cut-off for Shanghai and Ningbo sailings forward 72 hours and confirm which bookings sit on named-space agreements.
• By 22 September: reissue affected lane quotes at 14-day validity and add named lines for the congestion surcharge and the demurrage free period.
• By 25 September: secure container availability five days ahead of stuffing and run one daily cut-off, gate-in and empty-return sheet for Shanghai and Ningbo.
• By 30 September: cap any single week at 40% of the month's East China volume.
• Shift only the 20% of volume with the softest delivery dates to Yantian or Qingdao; a full switch costs about $55,000 a month on 72 FEU.
• Check the trade term before promising a date: on DDP a five-day slip against a contractual date is your cost.

Shanghai's 4.72 days of average vessel wait and Ningbo's 3.58 days are berth queue numbers, and they decide when your container can be lifted and loaded. Add the gate-in queue behind terminals running above nine days at WGQ2 and WGQ5, and the working assumption for East China is that box availability and sailing dates move back by five days, not by hours. Yantian at 2.45 days is the contrast, and it is the only reason a port switch is still worth costing.

Run your own book. Assume 120 FEU a month, split 72 out of Shanghai, 30 out of Ningbo and 18 out of Yantian, with 60% of the East China volume on a named-space agreement at an assumed $2,800 a FEU and the rest on FAK spot. Assume spot runs $600 a FEU above contract during a backlog. Leave 29 FEU on the spot market and you pay the premium; if volume can be moved under the agreement to hold spot at 10 FEU, you save 19 x $600, or $11,400 a month. Now test a port switch. Assume routing an East China box through Yantian adds $900 of inland transport and barge cost and buys back 2.27 days of waiting against Shanghai. On 72 FEU that is $64,800 of extra cost against roughly $9,800 of recovered terminal and demurrage time at an assumed $60 a FEU per day, so a full switch loses about $55,000 a month. The switch only pays on the boxes carrying the heaviest delivery penalties.

By 18 September, move the documentation cut-off for Shanghai and Ningbo sailings forward by 72 hours and confirm in writing which bookings are covered by named-space agreements. By 22 September, reissue quotes for the affected lanes at 14-day validity instead of 30, and put named lines on the quote for the congestion surcharge and the demurrage free period, because a quote without those two lines is one you pay for twice. By 25 September, book container availability five days ahead of stuffing rather than two, and put one person on a daily cut-off, gate-in and empty-return sheet for Shanghai and Ningbo. By 30 September, split the October programme so no single week carries more than 40% of the month's East China volume.

Three routes if the queue does not clear. Move the least time-critical 20% to Yantian or Qingdao, accepting three to four extra days of total transit plus the inland cost. Stay in the East China port but book space with two carriers, so a roll does not become a cancellation. Or move the hardest SKUs to a higher freight level and pass the cost into the price with the buy side informed. The traps are specific. On FOB terms the buyer owns the sailing risk, but if you have promised a stuffing date you still own the production schedule, so read the term before you promise. On DDP the delay is entirely yours, and a five-day slip against a contractual delivery date can cost more than the freight. And a free-time period counted from discharge rather than from gate-in is a period you will overspend, because the queue starts before the box is available.

For Cross-Border E-commerce

• By 17 September: recalculate days of cover per SKU from 60 days of actual arrival data, not planned transit time.
• By 19 September: lift the top 20 SKUs by value from 21 to 30 days of cover and hold the long tail at 21.
• By 22 September: set the replenishment trigger at 12 days of cover instead of 8.
• By 25 September: split the next two inbound batches across Ningbo and Yantian so one queue cannot remove two deliveries.
• By 30 September: put storage cost per unit into the landed cost sheet and re-price SKUs where it passes 3% of gross margin.
• Air-freight only where a stock-out costs more than the RMB 38 a unit air cost against RMB 6 by sea; not across the catalogue.

For a cross-border e-commerce importer, the port queue is a stock-out clock. Shanghai at 4.72 days and Ningbo at 3.58 days mean inbound boxes reach the bonded warehouse roughly five days later than the booking says, and the 78-hour typhoon closure at Ningbo plus waits above nine days at WGQ2 and WGQ5 mean the recovery is not uniform across terminals. Days of cover is the only number that will tell you where the hole opens.

Assume a bonded warehouse selling 4,000 orders a day with 21 days of cover, which is 84,000 orders. A five-day arrival slip takes the same cover down to 16 days, a 20,000-order gap. At an assumed RMB 180 average order value, that is RMB 3.6 million of revenue standing on a boat. Price the storage side too. Assume 30 FEU a month, 4,000 units a container, and RMB 2,100 a container of extra storage and demurrage because the box waits to be collected. That is RMB 63,000 a month, or about RMB 0.53 a unit, which is 1.5% of a 20% gross margin on a RMB 180 order. The gap is the bigger number, and it lands on the fastest-selling SKUs first, because they are the ones that run out.

By 17 September, recalculate cover in days per SKU using the last 60 days of actual arrival data rather than planned transit time. By 19 September, lift the top 20 SKUs by value from 21 days to 30 and leave the long tail at 21, because holding 30 days on a slow mover turns a stock-out into dead stock. By 22 September, write the replenishment trigger: reorder at 12 days of cover instead of 8. By 25 September, split the next two inbound batches across Ningbo and Yantian, so one terminal queue cannot take two deliveries out at the same time. By 30 September, show storage cost per unit in the landed cost sheet and re-price only the SKUs where it exceeds 3% of gross margin.

Three ways to cover the gap. Air-freight the top 20 SKUs for one cycle, accepting an assumed RMB 38 a unit against RMB 6 by sea, which on 20,000 units is RMB 760,000, so limit it to the SKUs whose stock-out costs more than that. Move the inbound port to Yantian or Qingdao and accept a longer inland leg to the warehouse. Or shorten the cycle by ordering smaller and more often, which raises the order count and the per-unit handling cost. The traps: a replacement batch ordered during the backlog arrives into the same queue, so ordering early only helps if the port changes; the free-time clock at the terminal usually starts at discharge, not at booking, so budget the storage line from day one; and a peak-season surcharge quoted at booking can be renamed at billing, so get the surcharge and demurrage responsibility in writing before the container sails.

For Manufacturing Plants

• By 18 September: map every imported material to its port of discharge and flag everything moving through WGQ2, WGQ5 or Ningbo.
• By 22 September: raise single-source materials from seven to twelve days of buffer and keep dual-source at seven.
• By 25 September: stop any material shipping twice in a row through the same terminal and hold one alternative booking with a named second port.
• By 29 September: release purchase orders five days earlier for all materials on the twelve-day buffer.
• By 10 October: run the two most delivery-critical spare items through a dry-run shipping plan and time them end to end.
• Re-port or air-freight only where the line-down cost beats the premium: re-porting adds an assumed RMB 3,000-4,500 a FEU.

A factory reads the same numbers as a material clock. Shanghai at 4.72 days and Ningbo at 3.58 days mean inbound containers reach the plant roughly five days late, and WGQ2 and WGQ5 above nine days mean material routed through those terminals can be more than a week late. Yantian at 2.45 days is not a substitute where the supplier ships from East China, because the inland leg to the plant erases the gain.

Assume 80 FEU a month of imported material, split 30 through Shanghai, 20 through Ningbo and 30 through other ports, with seven days of safety stock on the line. A five-day slip takes cover to two days, which means one weekend, one port holiday or one rolled booking stops the line. Assume a production line running at RMB 800,000 of output a day. Raising the buffer from seven days to twelve adds five days of cover, which is 80 divided by 30 times five, or about 13 FEU in transit and on site. At an assumed RMB 800,000 of goods value a container, that is RMB 10.6 million of working capital, and at an assumed 5% annual funding cost it is about RMB 530,000 a year, which is cheaper than a single day of a stopped line.

By 18 September, list every imported material by port of discharge and flag the ones running through WGQ2, WGQ5 or Ningbo, because that is where the nine-day tail sits. By 22 September, raise the buffer on single-source imported materials from seven days to twelve and keep dual-source items at seven. By 25 September, split the October inbound plan so no material ships twice in a row through the same terminal, and hold one alternative booking per critical material with the second port named. By 29 September, move the purchase order release date forward by five days for every material on the twelve-day buffer. By 10 October, run a spare-parts drill: take the two items that hurt delivery most and work the shipping plan end to end with the clock running.

Alternatives carry a price. Routing East China material through Qingdao, Xiamen or Yantian works, but it adds an assumed RMB 3,000 to 4,500 a FEU of inland transport plus three days, so it belongs on the materials with the highest line-down cost, not on everything. Air freight works for one or two critical components and not for tonnage. Holding more stock works until the cash cost is measured. The traps: a material classified as dual-source is often single-source in practice, because the second supplier cannot hold the tolerance; a spares pipeline that runs through the same terminal as production material is not a backup; and a supplier confirming a sailing date during a backlog is confirming a booking, not a departure.

For Brand Owners

• By 18 September: re-tier the delivery promise by fulfilment channel and label each channel with its port and current wait.
• By 22 September: give customer service the exact wording and start date of every changed promise.
• By 25 September: write the late-shipment allocation rule, with margin and repeat rate setting priority.
• By 30 September: publish one daily delay bulletin from a single owner covering port, vessel and arrival date.
• By 7 October: name which markets move to an eight-day promise and which keep five days.
• Compare the fix against the penalty: five extra days of cover on the fastest 50 SKUs costs about RMB 60,000 a year against a RMB 1.8 million a year slip penalty.

A five-day port slip is a five-day delivery slip unless the promise is built to absorb it. Shanghai at 4.72 days and Ningbo at 3.58 days, with WGQ2 and WGQ5 above nine days, mean an East China stock position reaches the fulfilment point about five days late in the base case and more than a week in the WGQ2 and WGQ5 case. If the promise to the customer carries a date, that date is the first thing the queue touches.

Assume 30,000 orders a month on a five-day delivery promise. A six-day arrival slip pushes about 20% of the month past the promised date, which is 6,000 orders. At an assumed RMB 25 of compensation, re-delivery and return handling an order, that is RMB 150,000 a month and RMB 1.8 million a year, all of it from a queue the customer never sees. Now price the cheapest fix. Holding five extra days of cover on the 50 fastest SKUs, assumed RMB 1.2 million of stock at a 5% funding cost, is RMB 60,000 a year, which is 3% of the penalty. Widening the promise from five days to eight removes most of the 6,000 orders but is a visible change to a competitive promise, so the decision is which is worth more: the tighter date or the smaller penalty bill.

By 18 September, re-tier the delivery promise by fulfilment channel rather than by product, and label each channel with the port it depends on and the current wait, so the promise matches the route. By 22 September, brief customer service with the exact wording of every promise that changed and the date it changes, because a customer told in advance does not file a complaint. By 25 September, set a stock allocation rule: the highest-margin and highest-repeat channels take priority when a shipment is late and the long tail takes the delay. By 30 September, publish a single delay bulletin from one owner covering the port, the vessel and the arrival date, so sales, service and the 3PL read the same number. By 7 October, decide which markets move to the eight-day promise and which keep five days, and put the split in writing.

Alternatives, and what each costs. Move the fastest SKUs into local stock so the promise no longer depends on a berth, which converts a variable delay into a fixed holding cost. Air-freight one cycle of the top SKUs and keep the promise for the customers who are watching the date. Or make the longer promise explicit and compete on price instead of speed, which is honest and loses the customers who buy on the date. The traps: a promise measured order-to-delivery hides the port leg, so the slip surfaces as a service failure rather than a supply failure; a stock allocation rule that is not written down gets decided by whoever argues hardest; and a 3PL's own cut-off moves during a backlog, so the promise needs its cut-off confirmed weekly rather than assumed.

For Procurement Teams

• By 15 September: get terminal-level wait times and free-time periods from both carriers and both forwarders in writing.
• By 19 September: target a lift in contract share from 70% to 85% with the top two carriers, with a named-terminal clause.
• By 22 September: agree a typhoon and force majeure clause covering notice period, re-routing right and demurrage ownership.
• By 26 September: qualify a second port and a second carrier for the top five imported materials with transit times recorded.
• By 30 September: cap any single terminal at 50% of volume on a critical lane.
• By 31 October: add a re-routing clause with a cost split; at a $700 spot premium, 180 FEU is $126,000 a year.

Procurement is where the port queue turns into a contract term. Shanghai at 4.72 days, Ningbo at 3.58 days and Yantian at 2.45 days are three different risk levels on the same lane, and the 78-hour Saudel closure at Ningbo is the evidence that a terminal can stop for three days with no notice. Rotterdam's APMT Maasvlakte II yard at 85-90% and the 4 September Antwerp strike say the same thing about the other end of the route.

Assume 1,200 FEU a year on the affected lane, split 70% on annual contract, which is 840 FEU, and 30% on spot, which is 360 FEU. Assume spot runs $700 a FEU above contract during a backlog. The spot tail costs 360 x $700, or $252,000 a year. Lifting the contract share from 70% to 85%, which means 1,020 FEU on contract and 180 FEU on spot, saves 180 x $700, or $126,000 a year. That is the number to bring to the carrier, and it is why the conversation belongs in September rather than in the week the terminal closes. One more line: assume a force majeure notice costs $40,000 a shipment when a buyer rejects the excuse and buys replacement goods; one clause tested in writing on 15 September is worth more than a rate card renegotiated in October.

By 15 September, get terminal-level wait times and free-time periods from both carriers and both forwarders in writing, because a named terminal with a nine-day wait is a different supplier from the port average. By 19 September, open the contract-share conversation with the two carriers holding the most volume, targeting a lift from 70% to 85% with the extra volume in the July to October window, and ask for a named-terminal clause rather than a port-level one. By 22 September, agree a typhoon and force majeure clause covering the notice period, the re-routing right and who owns the demurrage while the terminal is shut.

By 26 September, qualify a second port and a second carrier for the top five imported materials and record the transit time for each. By 30 September, put the port mix in writing: no single terminal above 50% of the volume on a critical lane. By 31 October, add a re-routing clause with a cost-sharing split, because in a backlog the carrier's first offer will be to re-route at your cost.

Alternatives and their price. Buying more contract share protects against the spot tail but locks volume into a carrier that may itself be late, so cap the share. Keeping a spot tranche preserves flexibility and pays the premium. Splitting across three ports lowers single-terminal risk and adds inland cost on every box. The traps: a port-level service clause is satisfied by a carrier that calls at the one terminal in that port which is still working; force majeure gets invoked by the carrier as readily as by you, so the clause has to work both ways; and demurrage responsibility during a terminal closure is usually left silent in a standard contract, which means it is settled after the fact against whoever has the weaker file.

↑ Back to the key points