Air cargo rates ease to US$3.13/kg as capacity recovers

The Analysis

Three-thirteen a kilo. Spot three sixty-two. Global average down 1% week on week, spot down 2%, capacity now about 2% above where it sat before the conflict. Read only that and you would conclude air freight got easier in the first week of July.

Then read the lane lines from the same week. Hong Kong to Europe down 12%. Taiwan to Europe up about 20%. Same week, same broad origin region, same destination continent, thirty-two points apart. The average is not wrong. It is an average, and nobody ships an average.

The physical story is simpler than the rate table. Somebody gave the belly space back. Nobody gave the main deck back, and nobody made the cargo on those two lanes interchangeable, because it never was.

Start with where the capacity came from. When passenger flying recovers, belly hold comes back with it, and belly hold is the cheapest space in the market because the aircraft was going anyway. That is what pushes a global average down a percent: a lot of small, cheap, awkwardly shaped space spread across a lot of routes. It is real capacity. It is also the wrong shape for a pallet of servers, which is too tall for a lower deck and too heavy for a narrowbody.

Hong Kong to Europe falling 12% in a week is not a capacity story. Aircraft did not multiply. That is demand leaving, because the EU stopped exempting low-value shipments from duty. Take the exemption off a parcel and its landed cost changes at the border, not at the airport, and a certain number of parcels simply stop flying. The freight rate follows the volume down. Read that as capacity being abundant and you will mis-price the entire region.

Taiwan to Europe rising 20% is the mirror image. AI equipment is dense, heavy, high-value, and it does not go into a lower-deck position without a long conversation about handling and security. It flies main deck, on specific equipment, out of specific airports, and that equipment did not come back with the passenger schedule. So a shipper pays 20% more in a week when the world average fell 1%. Both numbers are true. Neither tells you what your lane costs.

Who gets hurt depends on which lane you are on and how much you fly. A consolidator in Hong Kong moving sub-hundred-kilo e-commerce into the EU now sits in a soft freight market and a harder border. The 12% freight relief is real, and it is almost certainly smaller than the duty that just landed on the parcel. Cut the freight budget, leave landed cost alone, and month end will be a surprise. A hardware manufacturer out of Taiwan at two tonnes a week is in a tight market and will pay, because at that value density 20% is a rounding error against a missed launch. The one who gets whipsawed is the shipper in the middle, 300 to 800 kilos a week, too small to hold a blocked-space agreement and too big to ride spot quietly.

Timing has a boring answer. Belly capacity arrives with the passenger schedule, and the passenger schedule is published months ahead, so most of the recovery you are seeing now was decided last winter. The average will keep drifting on fuel and on where the passenger carriers deploy. The lane-level moves, the ones that change your invoice, come from the cargo, not from the aircraft. Watch your own booking confirmations and the duty line, not the weekly average. The next real repricing out of Taiwan will show up as equipment that cannot be found for a given week, and you will see that before any index reflects it.

What the headline never tells you is which number you are actually paying. Three thirteen is a blend of contract and spot across every lane and every product. Spot is three sixty-two. That is 0.49 a kilo, about 16% above the number in the story. If you buy ad hoc, 3.13 was never your benchmark, and you have been paying roughly 16% over the headline all along. Spot also fell 2% while the blend fell 1%, so the gap is narrowing, which is good news for anyone buying late and a warning for anyone about to sign a twelve-month contract at last quarter's level. Do the same subtraction on the lanes and the other half appears: minus 12 and plus 20 in one week is a 32-point spread. A regional average is not a benchmark. It is a rumour.

A small version, using the report's numbers and one assumption of mine. Take that 0.49 gap between spot and the blend. At eight tonnes a week, and eight tonnes is my assumption rather than a figure in the release, the gap is 3,920 a week, roughly 17,000 a month. That is not the price of the freight. It is the price of buying late. It is also the only part of the bill that moves week to week, which makes it the only part worth negotiating. The Taiwan premium sits on top of whatever base you already have, and I am not going to guess your base for you.

The trap is benchmarking a lane against a world average. The second trap is reading a freight decline as a cost decline. Hong Kong to Europe is 12% cheaper to fly and more expensive to land, and both halves belong in the same sentence or neither belongs in the budget. If the duty change is what pulled the volume out, the volume does not return because the rate fell. It returns when the landed cost works again, and that is not a freight decision. If the boxes had stayed on the aircraft, the lane would still be priced for a parcel flow that no longer exists, and the shipper would be paying for capacity built for somebody else's business. That is the counterfactual, and it is the reason a 12% drop is not automatically good news for you.

So this week, pull twelve weeks of invoices, sort them by origin airport instead of by region, and find out which of these two lanes you are actually on. Most shippers are on both and budget for one.

↑ Back to the key points

— By Ömer Kaya