Freight

TLDR: America's freight relies on inefficient trucks due to misguided policies. Physics favors ships and trains. Asia shows better systems are possible through smart processes.

Prerequisites: Congestion Pricing, Price Controls, and Land Ownership are all useful stepping-stones, but not strictly necessary.

Imagine you are the logistics manager for a company like Silfab Solar, which manufactures goods (solar panels in this case) in northern Washington and wants to get them to various destinations, such as sunny southern California. How might you do this?

Backing up a bit, this kind of transportation question is one of freight — moving goods by a choice of road, rail, sea, or air. Globally, over 80%¹ of freight is transported by sea, and for good reason. Ships are by far the most efficient way to travel between two coastal cities. This follows from common sense and basic physics:

• Airplanes are the least efficient method, at roughly $1 per tonne²-km.³

  • It takes a lot of energy to lift heavy cargo into the sky!

  • It also takes more energy the faster you go, and planes go fast! In fact, the only real reason to transport via airplane is if you want something to get to its destination fast — air travel is around an order of magnitude faster than the next-best option for anything going a significant distance.

• Trucks are significantly more efficient, at roughly $0.10 per tonne-km.⁴

  • Trucks are also extremely flexible, as there are roads everywhere there are people. This flexibility is vital for first/last-mile transport.

• Trains are even more efficient, at roughly $0.03 per tonne-km.⁵

  • Trains are more aerodynamic than trucks, and have a better container-to-cargo weight ratio.

  • In theory trains can travel faster than trucks, but this varies a lot by region.

  • Both truck tires and asphalt roads deform more than the steel-on-steel wheels and rails of a train, leading to significantly lower levels of rolling resistance.

• Ships are the most efficient, at roughly $0.01 per tonne-km.⁶

  • This insane efficiency largely stems from ultra-large container ships, which experience drag based on a wetted area that scales roughly with the square of the ship size, but can carry cargo roughly in proportion to the cube.

  • It also helps that ships don’t need infrastructure except at ports, eliminating the construction and maintenance costs of roads or rails.

  • The cheapest way to travel is very slowly, improving engine efficiency and reducing drag. Ships frequently travel at less than half the speed of a truck, and are often forced into circuitous routes around landmasses. Ship is a bad choice if you can’t wait.

Returning to our starting question, northern Washington and southern California are both coastal, as well as connected by rail and road, so all options are on the table. Solar panels don’t go bad, so we can prioritize price over speed. The start and end of the journey will almost certainly be via truck. But the middle leg of the trip, where most of the distance is covered… will also probably go via truck.

Photo by Robson Hatsukami Morgan

Why So Much Long-Haul Trucking, America?

Physics says the obvious way to move goods from Seattle to Los Angeles is via ship. And yet, only a small fraction of freight goes this way. Only about 3.6% of goods entering or leaving the LA region come from/go to domestic ports, compared to 64% via truck, 17% via foreign ports, and 14% via rail. If this balance seems wrong, that’s because it stems partly from some very poor choices.

Most blatant is the Merchant Marine Act of 1920, better known as the Jones Act. By law, goods shipped between US ports (such as Seattle and Los Angeles) can only be carried on ships built, owned, and operated by Americans. It was designed to be a form of protectionism, both to help workers in the marine trade sectors and to ensure we have strong domestic shipbuilding capacity. But after over a hundred years the clear result of the law is that we are both much poorer and have an extremely small and uncompetitive merchant fleet/industry with little to no value in war. The total cost of the Act is impossible to know, but I’d guess it’s at least a trillion dollars.⁷ I could go into depth explaining and calculating the sheer insanity, but thankfully Zvi Mowshowitz already wrote a 12k word essay on the topic. Highly recommended. (For a more compact, professional analysis, consider reading this piece by the Cato Institute.)

But the Jones Act isn’t the only form of economic distortion on freight coming from the US government. There’s also the Foreign Dredge Act of 1906, which makes it vastly harder to maintain and build new ports, especially to accommodate the largest (and thus most efficient) cargo ships. Similar laws exist restricting aircraft, but these have less of an impact, given that air freight is naturally a niche service.

Okay, so domestic shipping is off the table because of self-defeating protectionism, but surely this doesn’t explain why more goods don’t travel by rail. There’s no law preventing Japanese trains from being imported and run on American tracks, for example. Well… except for all the standards being different, resulting in a need to overhaul any imported design… But still! There must be a better story for why American trains are largely dominated by trucks when it comes to long-haul freight.

The United States was built on rail, after all. In the 1800s, rail was the primary mode of travel, and major fortunes were made building out lines crisscrossing America, carrying both freight and passengers. This was made possible by strong governmental support, allowing railroads to take valuable land at low cost (sometimes in cruel ways), and also by the de facto monopoly on transport that the railroads provided.

But during the transition into the 1900s, both of those things changed dramatically. The Interstate Commerce Commission (ICC) was established in 1887 to reign in railroad monopolies, implementing strict price controls and routing regulations. And while this addressed some genuine abuses, it also made it nearly impossible for railroads to adapt to changing market conditions. The Transportation Act of 1920 then gave the ICC even more power, requiring railroads to maintain unprofitable routes and services, even during the great depression.

The pseudo-monopoly on freight also came to an end as automobiles (and airplanes) entered the picture, especially after the construction of the federal highway system in the second half of the century. With a changing world and heavily regulatory barriers plus pressure from unions, railroads became less and less profitable over time. Eventually, in 1980, Jimmy Carter⁸ deregulated the railroads (along with the airlines and trucking), finally allowing some life back into the industry. But alas, passenger rail in the USA never quite recovered — Americans just loved the convenience of cars and the speed of airplanes too much. Rail-based freight, however, made a serious comeback, and by the end of the century it was a strong option once again.

But despite a substantial recovery, there are still three major obstacles that are keeping trucks in the lead, despite rail’s natural advantages:

• The Federal Highway Trust Fund has historically undercharged heavy trucks for their road damage, effectively subsidizing truck freight. Studies suggest heavy trucks only pay for about 35% of their road damage through fuel taxes and fees, with the rest covered by general taxpayers and lighter vehicles. This hidden subsidy boosts trucking’s competitiveness with rail.

• Unlike highways, which receive consistent public funding, rail companies must privately finance their infrastructure. Modern rail projects face astronomical costs due to environmental reviews, land acquisition challenges, and complex permitting requirements. For example, grade separation projects often cost $50-100 million per crossing, making system-wide improvements prohibitively expensive in an America that has generally lost the will to build.

• Freight moving through American rail terminals has an average dwell time of 24 hours, but can sometimes be held up unexpectedly or even be lost, due to legacy systems and human error. Worse, these delays often mix poorly with hours-of-service regulations, creating compounding problems in the supply-chain. More hands carrying the goods often means more costs and complexity.

Trucks are thus, in practice, usually faster, simpler, safer, and sometimes cheaper.

To really make railroads and ships the best option, they need to be fast and efficient, particularly when transferring cargo in a terminal. So let’s turn our attention away from America’s political dysfunctions and look to see how the best freight terminals in the world reach their high levels of speed, efficiency, and throughput.

The Strength of Asia

There are some excellent freight terminals all across the world, including Europe, the Middle-East, South America, and even in the USA. But I believe the clear front-runner in freight is East Asia. Riding high off of late industrialization and cheap labor, the Japanese, Chinese, Koreans, Singaporeans, and others made it a priority to build up factories capable of global-scale production and invest in ports and freight networks capable of getting those cheap consumer-goods into the hands of rich buyers across the world.

Aiding the rise of Asia was the basic technology of standardized shipping containers, which arrived in their modern form in the 1950s, and slowly became ubiquitous.⁹ Standard containers allowed for goods of all sorts to be treated identically, and thus moved by purpose-built machines instead of wrangled by human laborers. Indeed, Asian ports (along with some others, such as the port of Rotterdam in the Netherlands) are famously high-tech and automated, and this automation would simply not be possible without standardization.

But, in my reading of freight efficiency, it seems to me that people often get distracted by automation, and don’t pay enough attention to the container itself. Robots are cool, but if it takes just as much money to build and maintain robotic trucks as it would to hire a human to drive them, what’s the point? Yes, the port of Singapore is one of the most advanced in the world, and has all sorts of fancy gadgets, but the ports of Los Angeles and Long Beach (“LA/LB”) are also highly automated, and are extremely inefficient by comparison. The port of Yokohama is arguably less automated than LA/LB, but still managed to be significantly more efficient. The general story is that automation is a mixed bag, and certainly isn’t a panacea for low-efficiency.

The real key is process. If you replace a human gate operator with a machine, the best case scenario is eliminating the cost of those workers’ wages, and more realistically it means trading those already low costs for slightly lower costs of the machinery. But if you put in an automated RFID system that not only means the truck drivers entering the port don’t have to stop, but can also be automatically directed to the right loading zone, then more containers can be moved per hour, the efficiency of the entire system increases, and the terminal will become more appealing to customers.

Notably, process doesn’t have to depend on technology. Part of what makes Yokohama such an exemplar of efficiency is their digital logistics software and systems like CONPAS, but part of it is also simply that the Japanese are obsessive about being on-time and making sure things are flowing smoothly. Similar stories can be told from Yangshan to Busan. Good organizational systems, foundational technologies like intermodal containers and deep-water berths for ultra-large ships, and cultures of perfectionism are the reason that Asia’s terminals are the envy of the world.

Utopian Freight

In a more Utopian world, governments are far less involved in protecting small groups of people, such as unions and shipbuilders, and far more involved in forcing market prices to reflect externalities. Carbon taxes put pressure on freight to be more efficient and clean, and instead of subsidizing highways, the governments of Utopia tax vehicles according to the fourth power of their weight, making heavy trucks significantly more expensive to operate.

Land-value taxes and forced land auctions every 16 years make building infrastructure much easier in Utopia, but make holding onto it potentially more expensive. Railroads essentially have access to a form of eminent domain, but are forced by the laws of economics to be more productive than the marginal alternative use for that land. To avoid getting into expensive bidding-wars with other developers, rail in urbanizing areas often gets capped and turned into tunnels.¹⁰ Thanks to land-value taxes being on the unimproved land, the taxes on such tunnels are near zero.¹¹ Grade-separations and other developments to allow trains to maintain higher speeds are likewise not punished in the way that property taxes do to such improvements in our world.

Ships are even more attractive in Utopia, however, since they are vastly more environmentally friendly than the alternatives and have little need for large-scale infrastructure. Large efforts are made to dredge ports so that most coastal cities can take advantage of colossal ships buried under mountains of shipping containers. Such ports are almost always equipped with both a set of railroads and normal roads to facilitate rapid drayage. Thanks to a near-complete absence of tariffs and other forms of trade regulations, norms against searching, and incentives for efficiency, it is normal for containers to be unloaded from a ship directly onto a train, and Utopians pride themselves on maintaining a median freight terminal container dwell time of less than 24 hours.

To achieve such low turnaround times, Utopian terminals run around the clock, not even shutting down for holidays. Truckers enjoy picking up cargo in the middle of the night to take advantage of low traffic on the roads, and thanks to pervasive investment in digital tracking technologies, containers are usually ready to load onto the truck as soon as it arrives, with no need to wait in lines or fill out paperwork. Weigh stations connected to the Internet are embedded into the loading zones, allowing trucks to avoid having to needlessly stop at arbitrary points along their journey.

In general, Utopia pushes for market-based solutions to freight. If a port is doing things the wrong way, an efficiency prize is created that might lead to another, competing port getting dredged. The government fights against monopolization not by detailed regulation or breaking up large companies, but simply by banning vertical integration of infrastructure — railroads must be owned and maintained by a different company than owns and operates the trains, and railroads cannot play favorites in the rates they charge their users.

As a result of all these things, Utopia has a richer, more interconnected world-market than we do. Goods travel faster, more cheaply, and in a way that causes less environmental harm. Utopians spend less time fantasizing about self-driving trucks, and more time focused on whether there are ways to make faster trains and bigger ships.

1. This is a weirdly hard statistic to source! Statista claims it’s “some 80 percent of all goods” but doesn’t clarify whether that’s by mass or by mass × distance, which seems like the more sensible measure. DHL claims it’s “around 90%” and again doesn’t clarify the units, though does provide an infographic citing Statista that indicates it was ~85% by tonne-kilometer in 2010. DHL (and a couple other sources I found) link to an OECD page which is now dead, but the Internet Archive says claimed “around 90%” without any further citation.

2. The global standard unit for measuring freight seems to use the metric ton and kilometer. To note that I’m not talking about American short-tons, I’ll be using the spelling “tonne” for metric tons. Very obnoxiously, I’m still going to use the American spelling of “kilometer.” Deal with it.

3. Global air freight revenue ≈ 1.6e11 dollars
Global air freight mass ≈ 7e7 tonnes
Revenue per mass ≈ 2.3e3 dollars per tonne
Mean transport distance ≈ 2.3e3 kilometers (this is a pure guess)
Revenue per mass × distance ≈ 1 dollar per tonne-kilometer

This report by the Australian government says air freight costs $1.84 AUD per tonne-kilometer, which is about $1.15 USD as of time of writing.

4. I have two sources for estimating truck efficiency. This one from the World Bank (2022) says it’s about 5 cents per tonne-kilometer in India and 13 in the USA. This one from the Congressional Budget Office (2015) asserts that it’s 15.6 cents per ton-mile, which I calculate as about 11 cents per tonne-kilometer. I’m not sure how much the global average costs are more like India or more like other countries, and arguably the USA cost should be a few cents higher to account for the implicit subsidy of the highway system, but I’m confident that the true price is somewhere in the ballpark of $0.10/tonne-km.

5. The Congressional Budget Office report from the previous footnote claims rail freight costs of 5.1 cents per ton-mile, which is 3.5 cents per tonne-kilometer. This random forum poster claims a price (that after converting currencies and adjusting for inflation is) closer to 1.5 cents, but also agrees that it’s about 1/3 the price of trucking. The Australian Government report from footnote #3 claims ~2.5 cents (after converting currencies), but also claims the prices is slightly less than half the price of trucking.

6. This Dutch report is a little confusing for me, but seems to indicate sea freight costs per tonne-km as low as 0.0014 euros for a bulk carrier ships, but gives costs closer to rail for most ships. This source agrees with the low number, claiming 0.16 cents per tonne-kilometer. But shipping containers seem to be less efficient (per tonne) than bulk goods like grain.

Suppose it costs $4,250 to ship a 40 foot container from Shanghai to London, a travel distance of about 22,000 km. And suppose such a container holds around 20 tonnes of cargo. This works out to ~1 cent per tonne-kilometer, and seems reasonable.

7. Again, it’s impossible to really tally the costs of the Jones Act, but just taking a loose estimate by looking at tangible costs (and ignoring more speculative things like knock-on effects of America having lower shipping costs, or better responses to disasters), Zvi and Cato agree that the damage is likely in the tens of billions of dollars, annually. Let’s go with an estimate of $27 billion, which is conveniently ~0.1% of GDP. We can then construct a very basic model by asking how much wealthier we’d be if the US GDP had been 0.1% higher since 1920. Note that this is a very pessimistic model, that does not assume any increase in GDP growth, just a fixed increase in GDP.

This Wikipedia table (you may need to click to show), indicates that the GDP in 1920 was around $609B, in 2006 dollars, and that the GDP in 2006 was $11319B. Let f(t) be the GDP (in inflation-adjusted billions of dollars) at year t-1920, and assume an exponential structure:
f(t) = 609e^tk
f(101) = 11319 = 609e^101k
k = ln(11319/609)/101 ≈ 2.9%

The total production from 1920 to 2025 can be estimated by taking the integral from t=0 to t=105:
609((1/k)e^105k - (1/k)e^0) = (609/k)(e^105k - 1) ≈ 420210

420210 is the number of billions of dollars in 2006, which translates to roughly 672 trillion 2025 dollars in total economic production. 0.1% of that — a conservative estimate of total cost — is thus 672 billion. Given that this ignores things like increased wealth allowing for more growth, I think the real number is clearly in the trillions. (A 0.1% increase in GDP growth, for instance, results in over a hundred trillion dollars in lost production over the last century (if I did my math right), for instance.)

8. While all these acts of deregulation were bipartisan efforts spanning multiple administrations, I wanted to explicitly call out Jimmy Carter by name because I think people have an incorrect notion that Ronald Reagan and the republicans were centrally responsible for deregulation during this time period. In fact, these pieces of pro-market legislation were passed by a Democrat trifecta.

9. On the earlier topic of terrible policies, the International Longshoremen’s Association negotiated for a rule in the late 60s that forced containers to be loaded and unloaded by union men, dramatically cutting their value. Thankfully, this was struck down by the supreme court in 1977.

10. While whole-buildings can’t be subdivided into different parcels by the land tax mechanism, railroads (and indeed, regular roads) can. That said, there is a minimum track length for any parcel containing a railroad, to reduce bidding harassment. For similar reasons, tunnels and bridges count as permanent structures, and must be auctioned as a whole.

11. One might expect that there would be Pigouvian taxation on the noise and vibration from such tunnels, but as long as the rail was built first, such externalities are grandfathered in, and it’s not the railroad’s burden to be quiet. Instead, we might imagine a deal where the railroad is rewarded with a prize (perhaps in the form of an annual subsidy) for being quieter than the counterfactual.