In October of 2000, a passenger train derailed just north of London, leaving four dead and more than 70 injured. The cause: a cracked rail that should have been replaced. A judge called the accident “one of the worst examples of sustained industrial negligence in a high-risk industry.” 

To many, the accident was a cautionary tale, explains David Besanko, a professor of strategy at the Kellogg School. “A lot of people said that it really shows that the way Britain organized its passenger train system was flawed,” he says. 

At issue is a system economists call “vertical separation,” where one company builds and maintains the rail tracks, while other companies run the day-to-day train operations. The UK’s railroads have been structured this way since they were privatized in the 1990s. In contrast, “horizontal separation” has the same company managing both infrastructure and day-to-day operations, and competing with other companies doing the same.

Vertical separation can be good for consumers because competition among train operators provides consumers more options. But critics warn that it also runs a risk of creating mismatched incentives between the infrastructure company and the train operators. For example, train operators want high-quality rails to attract customers, but the infrastructure provider might focus on keeping costs low.

That disconnect can result in a rail system that is undesirable to riders—or even dangerous.

“It’s led people to wonder whether vertical separation potentially creates compromises to safety,” Besanko says.

This finding could have big implications for how regulators manage all sorts of public infrastructure and utilities.

New research from Besanko and Shana Cui, a visiting scholar from Beijing Jiaotong University, suggests that those critics may be right. While both vertical and horizontal separation can be found in national freight railway systems across the globe, the new study shows that horizontal separation typically results in a safer, better functioning rail system.

This finding could have big implications for how regulators manage all sorts of public infrastructure and utilities—such as bridges, airports, or even the energy sector, Besanko says. The takeaway could also prove instructive for countries that are currently weighing how to privatize their national railway systems.

Separation Anxiety

Economists have long acknowledged that both vertical and horizontal separation have their advantages, regardless of the sector.

For example, in the realm of rail transport, vertical separation often gives customers a wider choice of services, since multiple companies are running the trains, while horizontal separation can help companies operate more efficiently. (The United States employs a horizontal rail system.)

Yet neither industry experts nor scholars agree on which kind of rail system generates larger benefits for society as a whole.

So the researchers set out to model how a rail system would work under both scenarios, and which would produce a higher-quality infrastructure—that is, one that delivers increased levels of safety and satisfaction for customers.

“Think of quality as things that improve the reliability of the transport services,” Besanko says. That could include regular maintenance for safety, as well as sophisticated switching software to help trains run more efficiently.

The researchers imagined a vertical-separation scenario in which a country’s railroad tracks are managed by a central “network firm” that chooses and manages the quality of the rail network. Two different transport operators then pay the network firm an access fee to use the tracks and compete with one another for customers on those tracks. The access fee is determined by an independent regulator, who has the country’s best interests in mind.

Then they imagined a horizontal-separation scenario, in which two vertically integrated companies—each of which oversees the entirety of its freight service, from maintaining train tracks to making sure the trains are on time—operate separately. In areas where both companies have train tracks, they must compete for customers.

Which of these hypothetical scenarios would result in a higher-quality rail system? The researchers put both into mathematical language and compared the outcomes.

The Horizontal Advantage

Under a majority of circumstances, they found that horizontal separation produces the better rail system.

For starters, under the vertical-separation scenario, the network operator does not have any control over the access charge, since it is set by an independent regulator.

“So the network operator cares almost entirely about volume,” Besanko says, since only by increasing volume carried—not quality—can they increase their profits. As a result, the quality of the infrastructure is likely to fall by the wayside.

Not only is the price out of the network operator’s hands, but the regulator is also likely to set the price too low, further diminishing the network firm’s incentive to improve quality.

This happens because a regulator concerned with social welfare will frequently set the access fee for train operators at a level that equals the extra maintenance cost incurred by allowing one more train to access the network. This system provides the greatest social good, since it neither encourages traffic whose benefit would be less than the cost of maintenance nor discourages traffic whose benefit would be greater than the cost. However, this means that the network firm is forever doomed to low profit margins and thus has no incentive to lure more train operators to its tracks by improving quality. 

“So there’s no incentive for the network operator to boost network quality because there’s basically no gain for doing so,” Besanko says.

He did not look at a scenario of vertical separation without an independent regulator, because in nearly all parts of the world, network-access prices are regulated.

While horizontal separation typically outperforms vertical separation, Besanko emphasizes that there are some exceptions. In situations where horizontally separated companies do not have to compete—if, for instance, the different companies’ train lines serve completely different areas and do not overlap—the advantages of horizontal separation evaporate, as the companies have no reason to improve quality.

“The best-case scenario for vertical separation is one where you have very intense competition among the train operators.”

Brazil, China, and Beyond

Besanko thinks his findings will be of interest to any nationalized industry with what economists call a “bottleneck facility”—an expansive physical infrastructure upon which value is created.

He points to Brazil’s energy sector as one example.

Much of the nation’s natural-gas industry—from production to refining to distribution—is controlled by one state-backed company called Petrobras. But as the country grapples with how to restructure the industry to foster competition and cut corruption, Besanko thinks his research offers some options.

“Just break Petrobras into a bunch of mini-Petrobrases, and those guys could compete with one another, and that would be like horizontal separation,” Besanko says. Alternatively, the country could maintain a monopoly on the pipeline itself, but open up production to outside companies. “And that would be like vertical separation.”

Of course, the research also has clear consequences for countries looking to open up their nationalized rails. China, in particular, is considering how to privatize its state-owned freight system.

What should China do? While vertical separation could work well in some contexts, it requires regulatory institutions to ensure that transport providers are competing on a truly level playing field.

“It’s not clear to me that China’s institutions are quite ready for that,” Besanko says.

On the other hand, he says, horizontal separation works well when there are a handful of enterprises competing on roughly the same territory but spread out enough such that competition is not cut-throat. (Under cut-throat competition, the model implies that firms might have to spend so much on quality that they will run themselves, and each other, into the ground.) China’s Eastern corridor, running from Beijing to Shanghai, seems to fit the bill.

“You could probably have 2 or 3 competing horizontally separated systems,” he says.

Besanko hopes Chinese policymakers make the decision carefully—as his findings show the choice will have implications for far more than just the railway companies. Changing the quality of the infrastructure could impact public safety and determine how efficiently firms around the country get their cargo to customers.

“The rail infrastructure is going to be a very important part of how a country develops,” he says.