Express fuel surcharges hold near record highs; DHL lifts Hong Kong air rate to HK$10.30/kg

The Analysis

Half past nine on a Sunday night and somebody on the packing bench asks the least glamorous question in this business: is the number on this label still true? Two years ago that question had an answer you could give without looking anything up. The carrier published a rate, it moved roughly once a month, you typed it into your shipping table and you forgot about it for thirty days. Look at where we are now. For the week of 28 August to 4 September 2026, FedEx sat at 47.75 percent, UPS at 48.25 percent and DHL at 43.25 percent. DHL has taken its Hong Kong air-export fuel surcharge to 10.30 Hong Kong dollars a kilo for 7 to 13 September on long-haul. FedEx and UPS are both rolling weekly now. So the honest answer to the person holding that label is: it depends what time you printed it.

That is the whole story, and the percentages are not the interesting part of it. The interesting part is the clock. A surcharge that resets every seven days means your price is no longer a number you set, it is a number you rent for a week, and if your checkout page, your marketplace listing or your freight forwarder's quote sheet is even one day behind, you are selling at yesterday's cost. I have spent my career on the tail end of the journey, on the bits measured in hours, and I can tell you that nobody loses money on the headline percentage. They lose it in the gap between the moment the rate changes and the moment their own system notices.

Three numbers, then, and one date. Forty-seven point seven five, forty-eight point two five, forty-three point two five, all of them above forty percent for months now, which puts the fuel line just behind base freight as the biggest thing on your express invoice. Ten point thirty Hong Kong dollars a kilo out of Hong Kong. And the fourth number nobody writes down: seven days. That is how long your price is good for now.

Let me put the percentages on a real parcel so they stop being abstract. Assume a cross-border parcel with a chargeable weight of two point five kilos, and assume your base air freight out of Hong Kong is sixty Hong Kong dollars a kilo. Base freight is therefore a hundred and fifty dollars. At FedEx's 47.75 percent you add seventy one dollars sixty three. At UPS's 48.25 percent you add seventy two dollars thirty eight. At DHL's 43.25 percent you add sixty four dollars eighty eight. The spread between the cheapest and the most expensive of the three on that one parcel is seven dollars fifty. Seven dollars fifty sounds like nothing until you remember you are printing labels all day.

Now the trap, and this is the one I want you to take away from this piece, because it has cost people real money this month. The 43.25 percent and the 10.30 dollars a kilo are not the same kind of number. One is a percentage applied to base freight. The other is a flat amount per kilo. You cannot compare them, you cannot subtract one from the other, and you cannot say DHL is cheaper because 43.25 is lower than 48.25 or because 10.30 looks small next to a percentage sign. They are different metering bases on different products in different weeks. The 43.25 percent belongs to the 28 August to 4 September window. The 10.30 dollars a kilo belongs to the 7 to 13 September window and to Hong Kong air export on long-haul. If your finance person puts those two numbers side by side in a spreadsheet and draws a conclusion, fire the spreadsheet.

Here is the same trap expressed as money, because that is the only way it lands. On our two point five kilo parcel, DHL's per-kilo surcharge for the September window is ten dollars thirty times two point five, which is twenty five dollars seventy five. The percentage version on the same parcel would have been sixty four dollars eighty eight. Same carrier, two bases, a factor of two and a half between them. Which one applies to you depends on the product and the lane, not on which carrier you prefer. So the question you should be asking your account manager this week is not what is your surcharge. It is what base is my surcharge calculated on, and how often does that base get reset.

Let me scale it up, because a single parcel never changes anyone's behaviour. Assume you are doing eight hundred parcels a week at an average chargeable weight of two point five kilos and the same sixty dollars a kilo base. Base freight is a hundred and twenty thousand Hong Kong dollars a week. Fuel at 43.25 percent is fifty one thousand nine hundred. Fuel at 48.25 percent is fifty seven thousand nine hundred. The difference between the top and the bottom of the three carriers is six thousand Hong Kong dollars a week. Run that from the start of October to the end of the year, thirteen weeks, and you are looking at roughly seventy eight thousand Hong Kong dollars of pure spread. That is not a rounding error. That is a person's salary, and it is decided entirely by which carrier you hand the bag to.

Now the part I actually care about, which is hours. Assume those eight hundred parcels a week break down to about a hundred and fourteen a day. Assume your margin is eighteen Hong Kong dollars a parcel after everything else. Assume the surcharge moves five percentage points against you, which is exactly the gap between DHL's 43.25 and UPS's 48.25 on that August to September window. On a hundred and fifty dollar base that is seven dollars fifty a parcel, and on a hundred and fourteen parcels it is eight hundred and fifty five Hong Kong dollars. That is what one day of stale pricing costs you. If your rate refresh lags by thirty six hours, call it one thousand two hundred and eighty. One weekend of a cached rate file and you have given away a day and a half of margin on the entire book.

So the operating question for this week is not which carrier is cheapest. It is how many hours sit between the carrier publishing a change and your checkout page charging it. Measure that number today. Take the timestamp of the carrier's weekly publication, take the timestamp of the rate your system actually applied, and subtract. If the answer is more than four hours, you have found the most expensive process in your company, and it is not the one anybody is looking at.

The second clock in this story is the peak season one, and it has a hard date on it. The US Postal Service has announced a six percent temporary peak season surcharge for the 2026 holiday period, running from 4 October 2026 to 17 January 2027, covering Priority Mail, Ground Advantage and Parcel Select. It is stated as pending final regulatory approval, so do not go and promise a customer anything on the strength of it yet, treat it as very likely rather than done. The published per-piece impact runs from fifty cents up to twenty dollars eighty at retail and from forty cents up to eighteen dollars twenty at commercial rates. Read those ceilings properly. Six percent sounds like a small number and in most cases it is, but the ceiling is what hits a heavy parcel going a long way, and that is exactly the parcel profile that shows up in November.

Do the small arithmetic so you can see it. Assume a zero point nine kilo Ground Advantage parcel on a commercial base of four dollars twenty. Six percent is about twenty five cents. That is a rounding error on a low-value item and it will not change anything you do. Now assume a heavy, long-distance parcel where the surcharge lands near the eighteen dollar twenty commercial ceiling. That is not a rounding error on any item, and on a forty dollar product it eats the whole contribution margin. The lesson is not that six percent is dangerous. The lesson is that a percentage hides a per-piece ceiling, and only the ceiling tells you which listings are about to stop making money.

Put those two clocks together and September stops being a slow month. Everything you can ship before 4 October is shipped at the pre-peak rate. Everything you leave until November is shipped with fuel above forty percent, a demand surcharge on top of it, and peak capacity getting tighter every week. The window is this month. Once it closes, it closes, and there is no clever move in November that recovers it.

Capacity is where this turns from a pricing problem into a timing problem, and it is where the hours really start to bite. When peak season space tightens, the first thing that moves is not the price. It is the cut-off. The bag that used to go out at eight in the evening now has to be at the gateway by five, and the driver who used to collect at six now collects at three. Every one of those shifts removes hours from your working day, and every hour you lose at the end of the day becomes a day on the delivery promise, because nothing that misses tonight's truck moves again until tomorrow.

That is why I keep telling people to write promises in hours rather than in working days. Three to five working days is not a promise, it is a hope with a weekend hidden inside it. Say instead: orders placed before two in the afternoon leave the same day, orders after that leave the next morning, and typical door-to-door is ninety six to a hundred and twenty hours to a metro address. Then go and check whether your operation can actually hold the two o'clock number when the collection moves to three. If it cannot, change the promise before peak season, not during it. A promise you break in November costs you the account. A promise you shorten in September costs you a few per cent of conversion and nothing else.

On alternatives, there are four and they all cost something. You can change the tail carrier, and for US domestic this is the year to do the work, because USPS, FedEx, UPS and the regional carriers have different surcharge structures and different coverage maps, and the cheapest one changes by destination ZIP. The cost is integration: assume twenty four to forty eight hours of work per carrier to get labels, tracking and exceptions flowing, and do it now rather than in the first week of October. You can change the mode, moving anything that does not need to fly onto an air line-haul or a consolidated service, and the cost is three to six days of transit, which is fine for replenishment and fatal for a promised date.

You can change where the inventory sits, which is the alternative that actually scales. Assume a peak-season parcel shipped from a domestic warehouse avoids the international leg entirely; the surcharges we are talking about simply do not apply to it. The cost is the inbound freight and the holding, and the honest way to size it is to compare the per-unit inbound plus monthly holding against the per-unit surcharge stack you expect to pay in November and December. If you are shipping anything with a predictable November demand curve, run that comparison this week, because the inbound has to sail in September to be on the shelf in November.

And you can change the promise, which is the cheapest and the most underused. Raise free shipping thresholds by ten percent, drop the two-day option on your lowest-margin SKUs, or move your handling-time display from one day to two. Each of those buys you hours of slack in the operation, and hours of slack are what keep you from missing a cut-off when the collection time moves. Nobody likes doing it. It works.

The pitfalls are mostly about things that are not on the invoice until they are. Volumetric weight is one: a light bulky parcel billed on volume can move two brackets with a five centimetre change in carton size, and nobody checks carton size in September. Remote area is another, and a single remote delivery can carry a per-shipment charge in the tens of dollars on top of everything else, so run an address check before the label prints rather than after the invoice arrives. Residential surcharges, demand surcharges and peak surcharges all stack on the same parcel, and the stack is what hurts, not any individual line.

There is a specific version of that pitfall for this month. Do not compare the 43.25 percent with the 10.30 dollars a kilo, do not compare an August window number with a September window number, and do not assume that because a surcharge is described as temporary it will come off when it says it will. The three carriers have all been above forty percent for months. Plan as though they stay there through January, and treat any fall as upside rather than as budget.

Two more traps that cost hours rather than money. Weekends do not count the way you think they do. A parcel that misses a Friday cut-off may not physically move until Monday, and a three-working-day promise quoted on a Thursday afternoon is really a five-day promise. And rate caching is everywhere: the marketplace, the ERP, the plugin, the spreadsheet somebody built two years ago and never told anyone about. Every one of those is a place where this week's number goes to die quietly for seven days.

There is a budgeting version of this that catches people out because it hides inside a spreadsheet formula. Assume your quote model treats the surcharge as a fixed assumption, and assume that assumption was set at twenty percent back when that was a reasonable number. On our hundred and fifty dollar base that is thirty dollars a parcel. The real number for the 28 August to 4 September window was somewhere between sixty four dollars eighty eight and seventy two dollars thirty eight. The gap between the assumption and the reality is up to forty one dollars sixty three a parcel. On eight hundred parcels a week that is over thirty three thousand Hong Kong dollars a week of margin that was never actually there. Go and find that cell today. It is usually in a tab nobody has opened since spring.

The surcharge stack is the other place where the surprise lives. Fuel is the headline, but it does not travel alone. Residential, demand and remote-area charges all sit on the same parcel, and the remote one is the one that bites hardest per event: the international remote surcharge has been pushed to around sixty one US dollars a shipment on some inbound and outbound lanes. Assume three percent of your eight hundred weekly parcels land in a remote area. That is twenty four parcels, and at sixty one dollars that is roughly one thousand four hundred and sixty four US dollars a week on twenty four parcels. Twenty four parcels out of eight hundred is nothing on a volume report and it is a full-time job's worth of money on an invoice.

And then there is the queue, which is the cost that never appears on any invoice at all. Assume that in a normal week two percent of your parcels generate a where-is-my-order ticket, and assume each one takes eight minutes of someone's time. On eight hundred parcels that is sixteen tickets and about two hours. Now assume peak season arrives, transit slips by a day and the ticket rate doubles to four percent. That is thirty two tickets and roughly four and a half hours a week, and it keeps climbing every week the promise slips. Hours, again. Everything in this business that hurts you arrives in hours first and shows up in money about a month later.

So here is what this week looks like if you are the person who owns the parcel. Get the rate refresh interval measured and down to hours, not days. Get the September shipments pulled forward so that anything which can cross before 4 October does. Get the two metering bases separated in your costing so nobody compares a percentage with a per-kilo amount again. Get the delivery promise rewritten in hours with a cut-off you can actually hold under peak conditions. And get one tail-carrier alternative live before the end of the month, so that when space tightens you have somewhere to put the volume.

I know how this reads when you are busy. It reads like five more things on a list in a month that was supposed to be quiet. But the surcharges are not going to come down in the next six weeks, the peak window has a hard date on it, and the gap between a rate changing and your system noticing is measured in hours. Get the parcel out the door at the right price first. Everything else you can argue about later.

↑ Back to the key points

— By Nadia Pryce