European dock strikes hit 2-4 Sept as Shanghai typhoon backlog hits 9-day waits

The Analysis

Put a finger on the Asia-Europe loop and you are not touching a line. You are touching a pendulum. Eight or nine hulls swing between two continents on a fixed rhythm, and the rhythm is the product. The ships are the easy part. The hard part is the empty box going back, the berth window printed eighteen months ago, the truck slot booked eleven days out. None of that travels with the ship.

In the first days of September both ends of that pendulum were hit inside seventy-two hours. A forty-eight-hour German dockworker warning strike ran from 2 to 4 September across Hamburg, Bremerhaven, Wilhelmshaven and others; a Dutch nationwide walkout landed on 4 September against truck operations at Rotterdam, Amsterdam and Zeeland; Shanghai was still working off a typhoon backlog that left vessels waiting about nine days, and was called the most critical gateway in the world in the update that carried the figure.

Most coverage will file those as two stories that shared a week. I file them as one story about slack, and about what happens when two claims land on the same pool of it. The rest is written from the importer's side of the fence.

A warning strike is an odd instrument. It is not meant to stop a port for forty-eight hours; it is meant to prove that it could. What is being demonstrated is coverage. Hamburg, Bremerhaven and Wilhelmshaven are not three independent entries on a menu. They draw on the same labour market, the same bargaining round, the same rail corridors, the same drivers. When one round of talks collapses and six ports stop together, the substitution set collapses with them. Counting ports is the wrong measure. The measure is how much of your alternate capacity sat inside the blast radius, and that week it was nearly all of it.

Shanghai's nine days needs its own reading, because it is a ship-side figure quoted as a cargo-side one. Waiting time counts vessels at anchor. It does not count containers in a yard that was shut, nor the booking rolled because the vessel that should have lifted it arrived nine days late and had to cut a port to recover its proforma. A rolled box never anchors, so it never enters the statistic. The headline understates the cargo damage by exactly what the booking desk has already absorbed.

A pendulum service has a fixed round trip and a fixed number of hulls, and the slack is spread along it: a day at the load port, a day at the discharge port, a day in transit, a day in the yard. In a normal week a bad day at one end is paid out of the buffer at the other, and nobody outside the operations desk notices. That works until both ends are bad at once. Then the same buffer is spent twice, the round trip stretches by more than the sum of the two delays, and there is no fifth week to borrow. What looks like coincidence is arithmetic: one loop, one pool, two claims.

The pain arrives in a predictable order. Exporters whose cargo was already on the quay when Shanghai shut come first, because they are not waiting for a ship, they are waiting for their own boxes to be found and re-stowed. Importers on fixed retail dates come next, since their exposure is a calendar, not a freight bill. Furthest out are the people who move empties, who feel it in inland depots three to six weeks later. Nearly untouched: goods that can fly, intra-Asia traffic, and anyone whose contract carries a generous transit allowance. Contract shape matters more than size here. One string with no routing clause is worse than three, and the LCL shipper is worst off, because the consolidator protects the base and rolls the tail.

Dates make it concrete. A strike is a step: it stops, it restarts, and the gate queue is released as a slug the moment the gates open. You feel that from day one to day three, in truck appointments, in a box that was loading on the fourth and now loads on the seventh. A typhoon is a ramp: the terminal reopens, but the backlog drains over weeks, and it drains slower than it built, because a terminal at capacity cannot also clear a queue. You feel the ramp at the booking desk ten to fourteen days later, when the allocation comes back smaller than you asked. The equipment bill lands four to six weeks out, in an inland depot: there are no boxes.

What I have not seen said, and the reason I will not call this a bad week, is that the two failure modes have different shapes and the shapes collided. A labour action releases its queue instantly and then behaves normally. A weather backlog decays slowly and keeps misbehaving for weeks. Run them on one loop and the instant release at one end lands while the slow decay at the other is still going, so the restart slug meets a standing queue instead of a draining one. That is the difference between a port that is busy and a port that is stuck. Three weeks apart, the loop would have absorbed each separately, and in my estimate the total would have come out near half.

Run a small version, with numbers that are mine and not the source's. Assume forty FEU a month from Shanghai into North Europe, three weeks of stock at the destination, and a backlog that rolls fifteen per cent of one sailing, six boxes, at ten days a roll. Assume the European stoppage adds three days at the gate and bunching adds four more at sea, and that half of the six land in that window. Those boxes lose seventeen days against a twenty-one-day buffer and arrive with four days of cover left. That is not a stockout, which is why nobody escalates: the system reports success, the customer gets the goods, and you enter the autumn build with nothing in reserve. The carrying cost is a few hundred dollars a box. The expensive bill is paid in options, and you spent all of them in one week.

Had the German talks held, or Shanghai reopened a week earlier, one end would have kept its slack. Either alone is a normal bad month. Neither alone forces a choice between base volume and spot volume, and that choice, not the delay, is the expensive part. Once you are choosing, you are paying marginal prices for volume you already contracted.

What to do is less dramatic. Confirm berth windows in writing for anything sailing in the next ten days, and treat an unconfirmed window as a cancellation. Decide now which tenth of your volume is marginal, so the call is not made at four on a Friday with a customer holding. Write down, per SKU, the one string it depends on.

The trap in a week like this is to measure the event and miss the state. Forty-eight hours is easy to put in a report; the absence of lateral slack is harder to write down and will still be true in October. The other trap is treating the secondary gateway as a substitute when it is a different product: a shorter wait, paid for in touches, one more lift, one more transshipment, one more hand on the paperwork. Ports are not interchangeable sockets. Each is a node with its own hinterland, its own labour, its own rail.

The map did not change this month. Two knots were pulled at once and the string went taut, and a taut string has no give left for the next pull. That pull comes before the peak, not after. Watch the buffer, not the headline.

↑ Back to the key points

— By Tomás Ferreira