Global air cargo rates fall for a third week as Asia-Europe lanes cool

The Analysis

Three weeks of falling numbers, and half the industry reads them as one thing: air freight is getting cheaper. I have run this lane long enough to distrust that sentence on sight. Week 28 of 2026 closed on 12 July with the worldwide average rate at $3.03 per kilo, down 3% on the week and down for the third week running, and the same report says that number is still 24% above where it sat a year earlier. Those two halves of one sentence do not agree, and whichever half you build a budget on decides what you pay in the fourth quarter.

The lane I care about is much narrower than the world. Asia Pacific to Europe spot came in at $4.82 per kilo, down 5%, with China origins at $4.45, down 9%, which puts the lane back where it printed in late March. Two things happened to that lane inside a month: the European Union removed its €150 de minimis exemption on 1 July, and typhoon Bavi took lift out of Taiwan and China. Week 28 ran from 6 to 12 July, so every kilo inside that $4.45 was flown after the exemption had gone.

What I want to do below is narrow: turn those four prices into a bill for one shipment you would recognise, show where the falling-3% reading and the still-24%-higher reading send people wrong, and then talk about the only part of this I actually know, which is lanes. Which roads are open, how long recovery takes when a storm shuts one, and what a 23% year-on-year hole in Hong Kong to Europe tonnage means for the way you route.

Start with the four prices, because everything hangs off them. In WorldACD's week 28 of 2026, the week running 6 to 12 July, the worldwide average rate printed at $3.03 per kilo, down 3% on the week and down for a third consecutive week. Worldwide spot averaged $3.47 per kilo, also down 3%. Asia Pacific to Europe spot printed $4.82 per kilo, down 5%. China origins into Europe printed $4.45 per kilo, down 9%, and the report calls that a return to late-March levels. Tonnages moved the same way: global volumes down 4% week on week, Asia Pacific origins down 7%, and Hong Kong to Europe down 23% against the same week a year earlier. Typhoon Bavi is named as the reason capacity out of Taiwan and China was curtailed, and the European Union's €150 de minimis exemption ended on 1 July.

Now the part that costs people money, which is what those four numbers actually are. The $3.03 is an average across every lane on earth, contract and spot mixed together, short-haul and long-haul mixed together. The $3.47 is spot only, still worldwide. The $4.82 is one corridor. The $4.45 is one origin inside that corridor. If you are a European buyer flying out of China, the only one of the four you can honestly hold your own quote against is the last one, and even then only if your quote is spot, uncontracted and quoted the same way. I have watched a buyer take a $4.60 China-Europe quote, hold it against a $3.03 headline and spend a week arguing with a forwarder about a discount that was never on the table. That gap is not somebody's margin. It is a different road.

Two more numbers fall out of the report if you are willing to do the arithmetic in public. The worldwide average of $3.03 is 24% above a year earlier, so a year ago that index was sitting near $2.44 a kilo. The report gives this week's 3% and says the decline is the third in a row; it does not publish the size of the two before it. Assume they were also about 3% and three weeks ago the index sat near $3.32. That assumption is mine and not WorldACD's, and if the real figures were 1% or 5% the shape changes a little while the argument does not change at all.

On those numbers the run-up from last year was roughly $0.88 a kilo and the three-week slide has handed back about $0.29, which is close to a third of it. Keep that picture: a lane that has given back a third of its increase is not a cheap lane. It is a lane that stopped getting more expensive three weeks ago.

One more thing about reading a weekly index, and it is the thing that costs the most. Week 28 closed on 12 July. If you are reading this on the day it was published, 27 July, the cargo in that number stopped moving fifteen days ago. The air spot quotes that cross my desk carry a validity measured in hours, sometimes a day, rarely longer. A number that is fifteen days old is a photograph of a road, not a report on today's traffic. Use the index to decide what to argue about. Use a quote dated this morning to decide what to pay.

Which brings me to the tonnage figures, and here I want to be careful, because a week-on-week number taken during a typhoon is not a demand number. Typhoon Bavi curtailed capacity out of Taiwan and China. When lift disappears for a few days, kilos that were booked do not evaporate; they sit in a warehouse waiting for the next departure, and the week that loses the flights also loses the tonnage. Global volumes down 4% and Asia Pacific origins down 7% in a week like that is exactly what a weather event looks like.

Whether it is only weather is a question the next two prints answer for free. Capacity that was cancelled comes back within days to about a week once the airport reopens, and the deferred kilos then show up as an unusually strong bounce in week 29 or week 30. Demand that has genuinely gone soft does not bounce. I would not re-plan a programme off one week until I had seen which of the two I was looking at.

So here is the bill, and the assumptions go first so you can argue with them. One shipment, China origin to a European gateway, chargeable weight 1,000 kilos. I am assuming chargeable weight equals gross weight, which is another way of saying the cargo is dense. Freight is quoted as a single rate per kilo with no fuel surcharge, no security surcharge, no peak-season surcharge, no screening or war-risk add-on stacked on top, no contract discount applied, and nothing included for pickup, origin handling, customs clearance or delivery at the far end. Real invoices never look like this. The point of stripping it down is that the only thing left moving is the rate itself, which is the only thing this report is about.

At $4.45 a kilo that shipment costs $4,450 in freight. The report says China origins fell 9% to reach $4.45, so the week before the same lane was about $4.89 a kilo and the same shipment cost $4,890: the slide is worth $440 on one 1,000-kilo shipment. Scale it to a programme I recognise, say 20,000 kilos a month out of China into Europe, and the same 9% is worth $8,800 a month and $105,600 over a year, on the freight line alone. Then take the second comparison, the one that hides in plain sight. The Asia Pacific to Europe average is $4.82 and China origins are $4.45. That $0.37 is not a rounding difference. It is $370 on one shipment and $7,400 a month on 20,000 kilos, which is $88,800 a year. If your forwarder is pricing you at the corridor average rather than at the origin number, that is what you are paying, and nothing on the invoice will tell you so.

And here is the trap the headline sets, the one I said I would come back to. Rates falling 3% for a third week sounds like a market going soft. The same report says $3.03 is 24% above a year earlier. Both are true, and they are true about different things. The 24% belongs to the worldwide average, not to your lane. The temptation is to import it: divide $4.45 by 1.24 and you get about $3.59, and you have just invented a China-to-Europe rate for July 2025 that nobody published. Do not do that. What the report does give you for your lane is a different anchor, and it is a better one: $4.45 is late-March pricing.

Late March is four months back, not twelve. A lane that has retraced four months has given up its summer premium. It has not gone back to last year. If your 2026 budget was built on 2025 air rates, this print does not rescue it, and if your 2027 budget is going to be built on this print, you are anchoring on a week in which a typhoon was removing lift from the busiest export region on earth.

One assumption in that example deserves pulling out, because it is where the money usually hides. I assumed chargeable weight equals gross weight. Air freight is billed on whichever is greater, gross or volumetric, and volumetric weight at the usual 1:6000 divisor means the carton's dimensions rather than its contents set the bill once the cargo is light. Take my 1,000-kilo shipment and assume instead that it is bulky enough for volumetric weight to rule at, say, 1,200 kilos: the same rate produces $5,340 instead of $4,450, and the $890 difference has nothing at all to do with the market. It also opens a lever that does not depend on the market. Cut the packed volume by 5% and chargeable weight drops by 50 kilos, which at $4.45 is $222.50 on one shipment and $4,450 a month on 20,000 kilos. That is real money obtained without a single conversation about rates, and it survives every future move in this index.

Then there is the number I keep coming back to, because it is the only one in the report measured against a year rather than against a week. Hong Kong to Europe tonnage is down 23% year on year. Twenty-three per cent of a lane's kilos gone in twelve months is not weather and it is not a week's noise. It is a route being used differently. I do not know how many weekly departures are left on it and I am not going to pretend to.

What I know is what a lane looks like after it loses that much. Frequencies thin out, the aircraft that remain get fuller, and the penalty for missing one goes up. Think of a ferry rather than a road. On a busy crossing, missing a sailing costs you an hour. On a crossing that has lost a quarter of its traffic, missing one can cost you a day, and a day on a Europe-bound air shipment is a day of inventory somebody was counting on.

Put the map down for a moment, because the lanes in this report are not abstractions. Hong Kong to Europe is a crossing fed by truck capacity out of the Pearl River Delta, across a border, into an airport that has been among the busiest air cargo gateways in the world for most of my working life. This lane I have run, and I can tell you that when it loses twenty-three per cent of its tonnage in a year, the trucks do not stop running.

The freight on them goes somewhere else, and the somewhere else is usually a mainland gateway with its own border, its own cut-off times and its own exposure to storms. That is a route decision, not a rate decision, and it is the kind of decision that takes a month to unwind once you have made it. Which is why I would rather you asked the frequency question now, in a week when nothing is on fire, than in October when the answer is no.

So ask it, and do not underestimate that gateway while you are at it. If your cargo reaches Hong Kong by truck, two numbers matter more than the rate, and you should have both in writing before this week is out: how many Europe-bound departures a week your forwarder can actually book on, and what the recovery time is if you miss one. If the answer is a departure a day and next-day recovery, fine. If the answer is three a week and recovery measured in days, then your air programme has a single point of failure built into it and the fix is not a better rate. The fix is a second door: a mainland gateway on the same programme, or fewer and larger consignments so that one missed sailing costs you less of your month.

On the €150 exemption I will say what I know and hand the rest over. It ended on 1 July. Week 28 ran from 6 to 12 July, so every kilo in the $4.45 was flown after it ended, and the lane printed at late-March levels inside that week. What changed in the declaration scope, what that means for how a low-value consignment has to be entered, and who carries the liability if it is entered wrongly is not my subject and I am not going to improvise on it. Take it to your broker and get the answer in writing.

What I will watch from the lane side is whether the thinning I described has anything to do with it, because removing a threshold like that is the kind of change that takes a whole class of low-value consolidated freight out of the air market, and if that is what happened, the kilos do not come back in week 29 or week 30. That is a testable difference, and it is worth two more weekly prints before anybody re-plans a network around it.

What I would actually do with this week, then. Re-quote the whole China-to-Europe programme today, in writing, on a rate-per-kilo basis with surcharges listed in a separate column, and compare every line against $4.45 rather than against $3.03. Ask for quote validity in hours and put the expiry time on the booking. Take the two heaviest months of your fourth quarter and price them twice, once at $4.45 and once at the $4.89 the lane printed a week earlier, so you can see what a 9% swing does to the landed cost of the programme. Set the trigger before you need it: if a quote comes back above the $5.07 a kilo that the $4.82 print implies for the week before, escalate it the same day rather than discovering it in the monthly accrual. And decide now what you will do if typhoon season takes lift out again, because it is July in this part of the world and another one before September is not an unlikely event.

One more piece of arithmetic, because timing a re-quote is the question I get asked most often. If the worldwide average has fallen about 3% in each of three weeks, a fourth week of the same size would take $3.03 to roughly $2.94, and a week that snapped back 3% would take it to $3.12. On my 20,000-kilo programme that is a swing of about $1,800 a month at the worldwide average between waiting one more week and not waiting. At $4.45 a kilo the same 3% is about $0.13 a kilo and roughly $2,670 a month.

So the decision is not whether rates will fall. It is whether you can tolerate the other side of the bet. If a programme has to move in August, the cost of waiting for a fourth decline that may never come is one more week of capacity risk in typhoon season. If a programme can slip three weeks, the arithmetic says wait and re-quote every week. I have never seen a shipper regret re-quoting weekly. I have seen several regret signing a three-month air rate in the third week of a decline.

Ownership matters more than the numbers here, so put names against them. The person who owns the rate comparison is whoever signs the forwarder's invoice, and that person should be working from a quote dated this week rather than from this report. The person who owns the departure-frequency question is whoever books the space, and they should have the answer in writing before Friday. The person who owns the volumetric question is whoever specifies the carton, which in most organisations is packaging, a department that has never once been told what a kilo of air freight costs. Tell them. A 5% cut in packed volume is $4,450 a month on my numbers, and it is the cheapest item on this entire list because nobody has to be negotiated with to get it.

And for the slice of the programme that does not need to fly at all, the decision is not about price, it is about whether extra weeks of transit break your inventory plan. Check it against your own stock cover and your own promise dates. I am not going to put ocean comparisons into a piece about air lanes; I will only say that the road you choose is the road you have to live with when it rains, and the clock on the slow road is measured in weeks rather than hours.

A short list of the ways this goes wrong, every one of which I have paid for at some point. An all-in quote that bundles the surcharges hides the rate you are arguing about, so refuse it and get the columns separated. A quote in a currency that is not your budget currency adds a variable you did not ask for. A booking accepted on a verbal rate with no expiry time is not a booking, it is a wish. A volumetric surprise at the origin warehouse turns a $4,450 shipment into a $5,340 one after you have already promised a landed cost to somebody. A routing that depends on a single departure a week has no recovery path. And a typhoon clause nobody read means the extra cost of rerouting lands on you by default rather than by negotiation.

Strip all of it down and the week is smaller than the headline and more useful than the headline. Rates came off a summer peak by about a third of the run-up, on numbers that are a fortnight old, in a week when a storm removed aircraft from the busiest export region on earth, on a lane that is back where it was in March and nowhere near where it was last year. That makes it a good week to re-quote, a poor week to re-baseline a budget, and a very good week to find out how many ways out of your origin you actually have. Roads close and ferries thin out, and the shipper who knows three ways to the airport is the shipper who pays least when one of them shuts. You do not change all four tyres because the gauge moved. You check the pressure, you note which roads are open, and you keep one spare in the boot.

↑ Back to the key points

— By Ömer Kaya