What moves a freight rate

The cost of moving goods passes into the price of nearly everything else, yet how much it responds to fuel, to volume, or to an empty return leg is poorly measured. We set out what the published evidence can and cannot say.

By February 2021 the long-distance truckload rate index was already 18 percent above its pandemic low and climbing toward a peak 65 percent above it, while diesel that month sold for $2.88 a gallon, the cheapest it had been in two years. Trucking got expensive when diesel was cheap, because capacity could not keep up with demand in the post-pandemic boom. In 2026 the order reversed: when the Strait of Hormuz closed in March, retail diesel went from $3.90 to $5.64 a gallon in five weeks, and the rate index followed it up 13 percent in five months. Two spikes of similar size, with substantially different economics behind them. Telling them apart requires decomposing a freight rate into three components and estimating how sensitive each one is to market movements. In this first article of our cost-to-serve series we describe those components, how they depend on the full map of lanes, and how they respond to factors such as volumes and fuel prices.

Consider the relative magnitudes. Diesel is about 30 percent of a truck's variable cost per mile, so a 10 percent move in diesel lifts that cost about 3 percent, and less than that reaches the rate once carriers reroute around expensive fuel. A lane that thins out by 10 percent raises what it costs a carrier to cover a load on it by roughly the same amount. Fuel is not a bigger lever than the rest of the market. It is the only one with a weekly price.

The physical cost, referred to as "marginal cost," is the most measurable part of the rate, whereas the other components (an opportunity cost and carrier rents) bring more complexity and are highly context- and timing-sensitive. Recent empirical work has measured how sensitive rates are to each of the three, and we put those estimates side by side at the end of this piece.

Marginal cost: what the miles cost a truck

The first part is marginal cost: what it costs the carrier to run this load rather than another. Fuel, the driver's time, maintenance and tire wear, tolls, and the empty miles the truck runs to reach the pickup.

This is the part most people picture when they think about the cost of a haul. It scales with distance and weight, and makes up the bulk of the floor below which no carrier will accept a load.

Opportunity cost: how attractive other loads are to a carrier

In economics, an opportunity cost is the next-best alternative given up with every choice made: the opportunity cost of taking public transit to the airport instead of a taxi includes the additional time you spend on the bus or train. This idea shows up in several ways in freight markets. For a single load, it is the value of the best other load the carrier could have run with the same truck and the same hours. That value in turn is also a function of where the load's destination is and how valuable the truck's opportunities at that destination are.

Scroll to see why where a load goes affects your quoted rate. ↓

Tightness is only half of it. Opportunity cost is also about where a load leaves the truck. A truck traveling somewhere with little outgoing freight faces empty miles, known as deadhead, or a long wait before it can earn again. That cost rides on your load, even though none of it is about the miles you paid for.

Carriers make forward-looking decisions

The simplest view of trucking says a load is worth taking whenever its rate covers the running cost. That view cannot explain why the same lane costs a different amount in each direction. Carriers make forward-looking decisions that account for haul opportunities at the truck's next destination.

A carrier takes a load only when hauling it beats its other options (waiting for tomorrow's board, or running empty to a better market), so the destination's future value is baked into every quote. This comparison is the opportunity cost of the previous section. If the destination is worth more than where the truck stands (Chicago in a tight week), a carrier can rationally haul the load below its running cost, because the load doubles as paid repositioning. If the destination is worth less, as is the case with inbound Florida hauls most of the year, the rate charges you for the exit before the truck ever arrives. Rates are directional because a lane's two ends leave the truck in different places.

Brancaccio, Kalouptsidi and Papageorgiou (2020) wrote down and estimated this model for dry-bulk ships, which work like taxis of the ocean: 42 percent of ship-miles are sailed empty, and the same lane costs about $10,000 a day toward the port with poor reload prospects but only about $7,500 back. Their follow-on paper (2023) prices the friction itself: better pricing alone would raise welfare about 8 percent, and perfectly frictionless matching of ships with loads about 19 percent. The second figure is an upper bound; a platform that actually centralizes matching delivers about 9 percent, because it trades search friction for market power.

Carrier rent: premiums earned above costs

The third part is carrier rent: whatever the carrier captures above marginal and opportunity costs. A carrier can hold out for it when it has the upper hand, for example when few trucks are available for a load that has to move.

Rent is the part most exposed to competition. On dense lanes where many carriers bid, it is thin to begin with; on lanes served by fewer carriers, markups run higher.

Each part of the breakdown points to a different data source and a different driver when economic events hit.

How sensitive rates are to each factor

An elasticity is a ratio of two percentage changes: when a factor (such as fuel costs) rises by one percent, by what percent does the outcome move? An elasticity of 0.3 implies that a 10 percent rise in the factor moves the outcome by roughly 3 percent. The estimates below come from structural models of freight and bulk shipping markets, from instrumental-variable regressions, and in one case from a cost identity. The chart applies a 10 percent move to each factor, stated in the direction that makes conditions worse, so the values read directly as percent changes and a point to the right indicates a higher price. Together they give the relative sensitivity of the three components to the movements an operator encounters, such as a fuel shock or a decline in volume on a lane.

Marginal cost follows from an accounting identity rather than an estimate. Fuel is about 30 percent of a truck's variable cost per mile, so holding fuel efficiency and all other cost lines fixed, a 10 percent rise in diesel raises variable cost by about 3 percent. How much of that reaches the rate is a separate question, and the only structural answer available comes from ocean shipping. In Brancaccio, Kalouptsidi and Papageorgiou's model, a 10 percent change in fuel cost moves prices by 2.9 percent when carriers' routing is held fixed, and by 1.7 percent when they are free to re-optimize. Just over 40 percent of the direct cost effect is therefore offset by carriers' own routing adjustments rather than passed into the price.

Opportunity cost is the component with no observable market price, and the available estimates come from three settings. In dry bulk, a destination whose expected empty leg is 10 percent longer carries a price about 1.7 percent higher per day. That figure is an association rather than a causal estimate, because the regressor is a property of the destination and the specification cannot absorb destination fixed effects. In US spot trucking, operational costs are about 1.3 percent higher where backhauls are 10 percent scarcer, and the cost of covering a spot load is about 2.8 percent higher on a lane 10 percent thinner. The first enters its paper as a control rather than an object of study and is not separately instrumented; the second is estimated by two-stage least squares using predicted interstate trade flows.

One estimate in the chart points the other way. Everything so far implies that the busy direction of an imbalanced lane should be the expensive one, since that is the direction that leaves the truck stranded. In Japanese trucking, Tanaka and Tsubota find the reverse: the more imbalanced the lane, the cheaper its busy direction becomes relative to the quiet one. Running more freight over the same road lowers the cost of each load, and on those lanes that saving outweighs the cost of returning empty. Because the method compares a lane's two directions against each other, it recovers the net of the two forces rather than either one alone, so it shows which dominates in Japan without showing how large each is. The equivalent test has not been run on US lanes.

Carrier rent tracks competition, and its measured sensitivity is the smallest in the chart. In Colombian trucking, a doubling of a carrier's market share of a route raises the rate it charges by 4 to 7 percent. The effect concentrates on small and medium routes; on the busiest routes, served by hundreds of firms, the authors find little evidence of market power.

Evidence on the elasticity of opportunity costs to volumes, demand, or fuel price spikes in US freight markets is thin. No published study estimates how much a destination's onward prospects move a US truckload rate. The nearest evidence is an association from ocean shipping and a single published estimate from Japan. The gap is consequential, because lane selection, shipment timing, and any forecast of what a load will cost to serve all require a price for precisely this quantity. We are estimating it directly, using the frameworks in these papers.

What's next in the series

Part two documents the pricing model we are building on a live brokerage book. The construction adapts recent papers in freight markets cited in this article. We're working hard to distill insights from this research using a range of data inputs and deploy their predictive power in real-life settings to enhance performance for our customers. We will set out how the model is built, which parts of the theory survive contact with that data, and how the industry can benefit from these tools.