North European port strikes: German 48-hour stoppage and Dutch action strand ~160,000 TEU

The Analysis

Stand at the mouth of the Elbe with the North Sea range in front of you and the geography looks generous. Hamburg, Bremerhaven, Bremen, Wilhelmshaven, Emden, Brake, six gateways inside a stretch of coast a truck can cross in an afternoon, with Rotterdam, Amsterdam and Zeeland a little further west behind them. Nine doors into the richest consumer market on earth. That is what redundancy is supposed to look like on a map.

Then a wage negotiation stopped three point one percentage points short of agreement, and for forty-eight hours from the second to the fourth of September six German ports halted together, the Dutch ports followed on the fourth, and Linerlytica counted more than a hundred and sixty thousand TEU of boxship capacity waiting at anchor off Rotterdam, Hamburg and Bremerhaven.

Nine doors, and every one of them shut in the same week. That is the thing to sit with, because the arithmetic of redundancy quietly assumed those doors fail independently. They do not, and the calendar this month shows exactly why.

The small fact explains the large one. ver.di asked for 8.2 percent, the employers offered 5.1, and the distance between those two figures is 3.1 percentage points. In labour terms that is a negotiating room, not a chasm. A gap that size gets settled. What it does not do is get settled without a stoppage, so the stoppage is the price of the settlement rather than evidence of a breakdown.

Which is why I would not read this as German ports in crisis. I would read it as a 3.1 point gap being worth two days of national port capacity, and as the same gap being worth the same again in the next round, because the structure that produces it has not moved. That distinction matters more than it sounds: a crisis is something you insure against, a recurring cost is something you plan around.

A port strike stops more than cranes. It stops the pilot boat, the tugs, the lashers, the gate staff, and the customs presence that has to exist before a box can travel from quay to yard to road. The vessel rides at anchor with the cargo aboard, and the cargo is not simply late. It is in the wrong place, burning free time that was never designed for a queue.

When the stoppage ends, the interesting part begins, because a port restarting is not a port catching up. Berth windows are allocated, not improvised. The ships that waited arrive in some order, and that order decides who waits another week. Two days of lost ship-working produce a queue whose length is set not by the two days but by how many vessels were due inside them, and in a range handling this volume that is not a small number.

Who hurts first is mostly a question of how many berth windows a shipment needs. Transshipment cargo needs two, one at the relay port and one at the destination, which is why the guidance coming out of this event was delayed transshipment and rolled cargo rather than merely late arrivals. Deep-sea import cargo needs one window and can be recovered by the next call. Export cargo is last to hurt and hurts longest, because the empty has to come back before the booking can be honoured, and the empties were on the ships that were waiting.

Size matters less than contract structure. An importer with a direct call into a North European gateway and a fixed weekly allocation has a number to ring. An importer buying space on the spot, or riding a feeder into a relay hub, is the one who gets rolled, and gets rolled twice, because the feeder rotation is the first thing a hub sacrifices when it falls behind.

The clock runs like this. Ship-working stops on the second of September and resumes around the fourth. The queue at anchor builds through the fifth and sixth. The Dutch action on the fourth removes the obvious alternative on the same day, so there is no relief valve in the first window. The yard starts filling from the fifth, because boxes discharge faster than they can leave. And on the ninth a nationwide rail strike is called.

The ninth is the date to circle. Everything discharged between the ports reopening and the rails stopping has to leave by road or not leave at all, and road capacity across the range is not sized to absorb a fortnight of rail volume in four days. The inland leg becomes the binding constraint at exactly the moment the water side recovers. That is the pattern worth remembering about this month: the constraint did not disappear, it migrated, from the berth to the yard to the railhead, and each migration landed on an operator with less ability to absorb it than the last.

Which brings me to the figure everybody quoted and almost nobody converted. A hundred and sixty thousand TEU of capacity at anchor is a stock, not a flow. It is the nominal capacity of the ships waiting, and only part of that is cargo destined for these ports. Some is transit for inland Europe, some is bound for other ranges, some is empty equipment going the wrong way. The headline therefore overstates the work to be done locally and understates the network effect, because the boxes that do not get worked here do not get worked anywhere downstream either.

Convert it anyway, with an assumption I will flag as mine and not the source's. If the three ports together can work off a backlog on the order of ten thousand TEU a day once fully running, a hundred and sixty thousand TEU is roughly a fortnight of drain with nothing else going wrong. Put a rail strike in the middle of that fortnight and the drain slows, because discharge is capped by yard capacity and yard capacity is capped by whether the box can leave at all. A forty-eight hour stoppage becomes a month of disruption, and the month is built out of the coupling, not out of the strike.

That is the part I would underline for anyone planning the fourth quarter. Two labour disputes in two countries in the same week, with a third one on the rails five days later, is not three news items. It is one failure domain with a shared calendar. European port labour stopped being a tail risk some years ago and became a seasonal one, and seasonal risks get planned for rather than insured against.

A piece of arithmetic, assumptions mine again. Take a European importer carrying a fortnight of inbound cover on a product line, with one forty-foot container a week arriving through a North European gateway. Take the box that was due in the first week of September and assume it arrives in the third. That is fourteen days of delay against fourteen days of cover, so the line runs to zero exactly as the vessel berths, and anything that slips beyond that point is a stockout rather than a delay.

Now take a larger importer, four containers a week, twenty-one days of cover. The same event eats two thirds of the buffer and leaves seven days. Assume the rail strike pushes the inland leg by another five days, which is not a dramatic assumption when a national rail stoppage takes days to unwind. Seven minus five is two. That is the whole margin, on a three week buffer, produced by one wage round in one country. Nobody in that planning meeting thought they were running a two day margin.

Without the rail action this would have been a bad week with a good story attached. Ports reopen, queue drains, boxes move, and by the middle of September the range is running to proforma again. The rail strike is what turns a week into a month, and it was called by a different union in a different dispute over a different contract. Nothing coordinated the two except the calendar, and the calendar was sufficient.

That is the structural point I keep returning to with European gateways. The range looks redundant because the ports sit close together. It behaves as a single node because the labour cycles, the inland connections and the feeder rotations are shared. Redundancy that shares a failure mode is not redundancy. It is the appearance of it, and it has been priced as though it were the real thing for a very long time, in every network design that treated Antwerp, Rotterdam and Hamburg as three separate answers to the same question.

What to do for the fourth quarter. Before the end of the second week of September, take every booking with a European arrival between the tenth and the thirtieth and re-confirm the inland leg separately from the ocean leg, because the ocean leg is the one being reported and the inland leg is the one that will decide your delivery date. Assume nothing about rail for the fortnight after the ninth.

Split any allocation that runs more than half its European volume through a single range across at least two, and check that the second range does not feed the same relay hub on the same feeder rotation, or you have moved a booking without moving a risk. For the top fifth of SKUs by revenue, extend cover to twenty-eight days before the end of September and leave the rest alone.

Ask your carrier, in writing, when free time starts on boxes that were at anchor during the closure, and get the answer before the invoices arrive. Detention clocks that began while the ship waited are the quietest cost in this whole event, and they land on the importer rather than the carrier.

The trap is the reflex to book the next available sailing without reading its rotation. A vessel that calls Hamburg and then Rotterdam gave itself two chances to be late in the first half of September, and it took both. Two calls in one loop is not a diversified booking, it is the same bet placed twice.

↑ Back to the key points

— By Tomás Ferreira