Research Note
The Empty Seat
Truckload prices are back near their 2022 peak and nothing extra is moving.
On 31 August the US Department of Transportation announced that it had shut down 270 truck driving schools, 110 of them for violations of the English language proficiency requirement and 160 for failing federal training rules. The same announcement put the number of drivers placed out of service under the language requirement at close to 30,000, and the number of non-domiciled commercial driver's licences cancelled since the April 2025 executive order at more than 28,000.
Set against a truck transportation payroll of 1.47 million, those are large numbers. They are also the latest instalment of something that has been running for more than a year, and they land in a freight market that has spent 2026 doing what a freight market is not supposed to do: pricing as though it were short of capacity while carrying no more freight than it did a year ago. That combination is the subject of this note. If the price is not being set by the quantity of freight, what is setting it, and what would end it?
What the Freight Market Is Reading
The number the truckload market is judged on is the rate, and in 2026 the rate went up. The Bureau of Labor Statistics producer price index for long-distance truckload freight averaged 193.2 in the first half of 2026 against 174.8 in the first half of 2025, a rise of 10.5 percent. That is the largest increase since 2022 and it takes the index to 93 percent of its 2022 peak.
Load-board reporting through the summer has been considerably more dramatic than that, putting dry van spot rates up something close to 50 percent on a year earlier. Those figures are proprietary, they are not federal statistics, and they describe the spot market, which is the minority of the freight and the part that moves first. This note uses the producer price index throughout, which is contract-weighted and slower, and it is worth saying plainly that the two are measuring different things rather than disagreeing.
Either way the reading is the same, and it is the reading the trade press has settled on: after three years of a freight recession severe enough to bankrupt a great many carriers, the cycle has turned. Demand is back, capacity is tight, and the rate is doing what a rate does at the start of an upturn.
The difficulty with that reading is the freight.
The Freight Did Not Come Back
The Bureau of Transportation Statistics publishes a truck tonnage index, which measures the quantity trucks actually carry rather than what they charge for it.
The Price Recovered. The Freight Never Moved.
US long-distance truckload producer prices against the volume of freight trucks carried, rebased to first half 2018 = 100. January to June average of each year, 2018 to 2026
Monthly index values averaged over January to June of each year by Seven Measures, then rebased to the 2018 half-year. The underlying index levels are as published.
Behind the rebasing, the published half-year averages run from 132.7 to 206.9 for the price index and from 111.9 to 116.6 for the tonnage index.
Source: US Bureau of Labor Statistics; US Bureau of Transportation Statistics; Seven Measures calculations.
Over nine years the price index travels through a range of 56 points and the tonnage index through 4. In the first half of 2026 tonnage was 1.4 percent above the first half of 2025 and 1.3 percent below the first half of 2019. The freight recovery, measured as freight, has not happened. American trucks are carrying about as much as they were carrying before the pandemic, and slightly less than they were carrying at the 2019 high.
This is the observation the rest of the note is built on, so it is worth being precise about what it does and does not say. It does not say that demand is weak in some absolute sense; tonnage is up a little, not down. It says that the change in the quantity of freight between 2025 and 2026 is far too small to account for a 10.5 percent change in its price, and that anyone attributing the price to the freight has to explain how 1.4 percent of extra volume produced it.
Rail Was Offered the Same Freight and Did Not Reprice
There is a way to test that which does not depend on any judgement about how elastic freight pricing ought to be. Trucks are not the only way to move freight across the country. If what changed in 2026 was the demand for moving goods, the other modes were standing in the same market and should show it.
What Repriced Was Trucking, Not Freight
Change in US producer prices for surface freight by mode, first half 2025 to first half 2026, percent
Rail carried more freight over the period, not less: rail freight carloads rose 3.1 percent between the two half-years, against 1.4 percent for truck tonnage.
Source: US Bureau of Labor Statistics; Seven Measures calculations.
Line-haul rail producer prices rose 0.6 percent between the two half-years. Over the same period rail freight carloads rose 3.1 percent, which is to say the railroads carried more freight than the year before and charged essentially the same for it. Both trucking modes repriced: truckload by 10.5 percent and less-than-truckload by 12.8 percent.
That is a clean split and it is hard to get to from a demand story. A shipper with more goods to move bids for capacity wherever capacity is; the railroads had it and did not charge for it. What went up was the price of moving freight by truck, specifically, and it went up in both of the trucking modes at once, which is what you would expect if the scarce thing were common to both. The input truckload and less-than-truckload share, and do not share with rail, is a qualified driver per load moved.
The Price Rose and the Payrolls Fell
Which brings the question to the drivers themselves, where the series behaves in a way it has not behaved before in this record.
Prices and Payrolls Moved Together Until This Year
Year-on-year change in the US long-distance truckload producer price index and in truck transportation employment, first half against first half, percent, 2019 to 2026
Of the four years here in which the truckload price rose, 2026 is the only one in which truck payrolls fell.
Employment is 110,800 jobs below its first half 2023 peak and below its first half 2018 level. The establishment survey counts payroll jobs at trucking firms; owner-operators are self-employed and are not in it.
Source: US Bureau of Labor Statistics; Seven Measures calculations.
In every one of the previous seven years, truck transportation payrolls moved with the truckload price or against it in the direction a lag would predict. Prices and employment rose together in 2019, 2021 and 2022. They fell together in 2020, 2024 and 2025. In 2023 the price fell while employment was still drifting up by half a percent, which is the ordinary behaviour of a labour force at a cycle turn.
2026 is the only year in the series in which the truckload price rose and payrolls fell. The price is up 10.5 percent and employment is down 1.5 percent, leaving truck transportation with 110,800 fewer jobs than at its first-half 2023 peak and fewer than it had in the first half of 2018. Whatever is happening, the price is not pulling labour back into the industry, which is what a price is for.
Nobody Is Ordering Trucks
Nor is it pulling in equipment.
Nobody Is Buying Trucks Into the Price
US retail sales of heavy weight trucks, millions of units at a seasonally adjusted annual rate. January to June average of each year, 2018 to 2026
The weakest first half since 2020 and 11.4 percent below the first half of 2025. Sales are all heavy trucks sold into the US market, not tractors bought by for-hire carriers, which is a subset of them.
Source: US Bureau of Economic Analysis.
Retail sales of heavy trucks averaged 404,000 units at an annual rate in the first half of 2026, down 11.4 percent on the first half of 2025 and the weakest first half since 2020. Carriers looking at a 10.5 percent increase in the price of hauling a load are buying fewer trucks to haul it with.
One reading of that is caution: carriers who lost money for three years do not believe the rate will hold and will not underwrite a tractor against it. The other reading is that the tractor is not what is missing. Both readings are consistent with the exhibit and this note does not claim to separate them. What the exhibit does establish is that the ordinary self-correcting mechanism of a freight upturn, in which a high rate calls capacity back and competes the rate away, is not currently running in either the labour or the equipment channel.
What Is Being Taken Out
The thing that has been changing over the same period is who is permitted to drive.
An executive order in April 2025 directed stricter enforcement of the long-standing English language proficiency requirement for commercial drivers, and the Commercial Vehicle Safety Alliance added failure of it to the criteria that put a driver out of service at the roadside. A final rule effective 16 March 2026 narrowed eligibility for non-domiciled commercial driver's licences to a short list of visa categories. The announcement of 31 August adds the training schools to the same programme.
The figures the Department has announced are, in its own categories, more than 28,000 non-domiciled licences cancelled, more than 24,000 drivers removed under the language requirement since April 2025, and close to 30,000 placed out of service under it. These are announcements rather than a measured statistical series, the categories plainly overlap, and they should not be added together. The point of quoting them is only that the order of magnitude is tens of thousands of drivers rather than hundreds, in an industry whose payroll fell by 22,700 over the year in question.
Investment Implications
The firm reads three things from this. First, that the 2026 truckload rate is a capacity event and should be underwritten as one. A rate driven by freight volume is a bet on the economy and mean-reverts when volume does. A rate driven by the withdrawal of a licensed input does not mean-revert on the same clock, because the input is restored by a licensing decision rather than by a purchase order. Two different things are being priced and they are being quoted under one name.
Second, that the duration of this is a policy question and should be watched as one. The series to follow are commercial licence issuance, the count of carriers holding operating authority, and the out-of-service rate at the roadside, none of which is a freight statistic. Class 8 order books, which is where the industry conventionally looks for the top of a rate cycle, will not answer it: exhibit 4 shows equipment buying falling into the price already, and it will keep falling while the constraint is the seat rather than the truck.
Third, that the pass-through is incomplete and asymmetrically held. The contract index has moved 10.5 percent against spot reporting several times that, which means the repricing is still working through contract books and that the carriers with the most contract exposure have the most of it still to come. Brokers earn the spread between the two prices and are therefore on the other side of the same movement, which is a statement about how the models are built rather than a claim about anybody's reported results. That distinction between business models is the more useful output here than any view on the sector.
The work this note has not done, and which would turn a reading into a position, is the count of active operating authorities month by month against the enforcement calendar. That is answerable from the carrier census and it is the thing that would establish causation rather than infer it.