Congestion holds Europe's floor, and the Q4 window narrows to fifteen days

Last month closed on a test with two outcomes: if September stopped at par, the July surcharge had simply lapsed; if it undercut June, demand was softening underneath it. Europe delivered the first answer almost to the dollar. China → Germany, the Netherlands, Belgium and the United Kingdom all printed $4,095-$5,005 for a 40GP, unchanged from August to the cent, and France closed the small gap it had opened to join them. But the same month produced 101 destinations rising against 84 falling, an average lane up 15.3% and a median up 1.5%. Those two numbers describe different markets, and both are correct.

The flat North European print is the most informative number in this snapshot, because the reason for it is not the obvious one. Volumes on the lane remain subdued and Asia-Europe spot rates have now fallen for a seventh consecutive week. On demand alone the lane should have kept sliding. It did not, because back-to-back typhoons left the Port of Shanghai with an average vessel waiting time of 4.42 days and over 2.4 million TEU of capacity disrupted across Northeast Asia — and carriers responded with port omissions, void sailings and schedule slippage. The sailings that would have carried the correction were the ones that got cancelled. Southern Europe, further from the affected rotations, kept falling: Italy −21.2% and Spain −19.1% to $4,185-$5,115, the Czech Republic −20.8%, Denmark −30.2% to $3,330-$4,070.

Read against June rather than against August, the picture is less settled than a flat month suggests. Germany now sits 11.7% below its June base, but the Netherlands, Belgium and the United Kingdom are all 9.6% above theirs. The surcharge came out of the German lane and has still not fully come out of the others. For a buyer, that is the practical asymmetry this month: the same $4,095-$5,005 quote represents a genuine retreat on one lane and a residual premium on the next one over.

Everything the European lane did not do, the southern headhauls did. China → Colombia rose 82.0% and China → Brazil, Argentina and Ecuador all printed $8,190-$10,010, up 54.2%, 54.2% and 49.2%. Chile and Mexico came in at $6,570-$8,030, up 46.0%. These are not extrapolations: Colombia, Brazil, Argentina, Ecuador, Chile and Mexico each arrived as real desk quotes this cycle. The driver our sources give is the same congestion read from the other side — with capacity repositioned to protect mainline reliability and Q3 volumes still moving, the South American lane is absorbing both the tightness and the seasonal surge at once.

That answers the question we left open in August. We wrote then that the Pacific Alliance needed one clean month before its print could be read as a level rather than a rebound, because July's read had been withheld and August's +58.7% was partly our own suppression catching up. September is that clean month, and it did not settle back — it rose again, on real quotes, by half as much again. The rebound reading was wrong. This is a repricing.

North America splits along the same line the indices show. China → Canada arrived as a real quote at $7,740-$9,460, up only 7.5%. The United States, which carries no desk quote of its own in this dataset and is derived from its cluster, prints $6,770-$8,275. That derived lower bound lands within $5 of the $6,765 per FEU independently reported for Shanghai-US west coast this week — a coincidence worth stating plainly, because it is the kind of external check we cannot run on most lanes.

The short-haul Asian lanes disagree with each other more than usual. Vietnam rose 75.0% to $630-$770 and Indonesia 30.0%, while Thailand fell 21.8% to $387-$473 and Japan and South Korea both eased 5.0%. At these absolute levels a small dollar move is a large percentage, and the ranking changes month to month without meaning much — the levels are the signal, not the deltas. South Asia moved harder: Sri Lanka rose 94.9% to $3,420-$4,180 on a real quote, Bangladesh 30.5%, and India and Pakistan both print +60.0%, which is a number we chose rather than one we received.

The Gulf produced the cleanest resolution of the month. China → Saudi Arabia rose 76.2% to $11,655-$14,245, and its air rate came in at $5.50/kg — down from the $6.90/kg that the raw desk data had carried for three consecutive months and that we finally published in August. We asked for a fourth reading to separate a structural Gulf premium from a routing artefact. The fourth reading came in lower, which settles it: the premium was real while it lasted and it is now deflating, rather than being a permanent step in the lane's cost base.

Kuwait is published this month at $11,628-$14,212, against $3,126 in our own August snapshot. That is not a 313% market move and we are not presenting it as one. Kuwait carried no desk quote in August, so our published figure was an extrapolation from its neighbours, and it was too low. When a real quote finally arrived it was rejected by our own threshold rule for being too far from the number we had invented. It lands within 0.3% of Saudi Arabia's own quote the same month, on both the 20GP and the 40GP, from the same gulf and the same transhipment ports; on air, its $6.90/kg is exactly the level the Saudi desk confirmed three months running. Neighbour corroboration beats our prior guess, so the quote is published as received — and Kuwait is excluded from its cluster's ratio maths, so that the correction of our error does not travel to the United Arab Emirates, Qatar, Oman and Bahrain as a fabricated 60% rise. Those four print +25.0%.

Two other suppressions are worth naming. India and Pakistan both arrived with a 40GP priced identically to their own 20GP, which is physically impossible, so both are extrapolated at +60.0% rather than the +120% and +109% their raw carried; the direction is right and the magnitude is ours. And eight Asia-Pacific markets — Australia, New Zealand, Singapore, Malaysia, Thailand, Vietnam, Indonesia and the Philippines — arrived with an LCL rate collapsed to the $5/cbm sentinel. That is a single systematic fault across one region's column, not eight independent errors, and their LCL is carried forward from August rather than published.

The operational fact that matters more than any rate this month is the calendar. Mid-Autumn Festival and Golden Week fall back to back, and Chinese factories and trucking capacity begin standing down after roughly 20 September. That leaves about fifteen working days to move Q4 inventory, into a market where carriers are running an aggressive blank-sailing programme, Shanghai is still working through its backlog, and eight carriers have just posted a 58.2% year-on-year rise in operating profit — which is to say they have no particular reason to discount space to hold volume. Our European desk's guidance is to have ocean cargo at the warehouse before 15 September and to switch to China-Europe rail rather than wait if a factory slips. On the North American side the constraint is rolled cargo and pickup delay at destination rather than the rate itself.

The practical consequence is that the usual sequencing reverses. For most of 2026 the question was whether to book now or wait for a better number, and waiting was often right. In a fifteen-day window against a congested origin, a booking that rolls is not a slightly worse rate — it is Q4 inventory arriving in November. Space, not price, is the binding constraint until the holiday clears.

August's lesson was that ocean and air had stopped telling the same story. September's is narrower and more useful: the average has stopped describing anybody. With the mean up 15.3% and the median up 1.5%, a shipper who reads only the headline will conclude the market rose and will be wrong on more than half of all destinations. The right question this month is not where freight is going. It is whether your box gets on a ship before the twentieth.