Should we subsidise railways?
The Swiss, once again, have the answer
Every year, the British Government spends about £12.5 bn on subsidising the railways, about 1% of total expenditure, or 0.4% of GDP. Should it?
Scepticism towards subsidy
All railways face a huge problem: they are expensive. The advantage of railways is that they can move very large numbers of people reliably using minimal space. The disadvantage is that they cost a lot to build and to operate, and usually cost much more than they can recover in fares.
As a result, nearly all railway systems are reliant on some kind of subsidy. There are two types of subsidy. First are subsidies for capital expenditure, i.e. the state pays for capital projects like line upgrades and new stations. The second are subsidies for routine operations, including regular capital maintenance. This post is about the second category.
It is untrue that passenger railways cannot operate profitably. I think this myth has spread because of Americabrain; no railway in North America is profitable. But outside of America, many metros in dense cities operate subsidy-free: the London Underground does, and so do the Munich U-Bahn, the Hong Kong MTR, and the privatised Japanese railway companies. Many countries’ long-distance networks require no subsidy either: the Swiss intercity network, Eurostar and the Shinkansen in Japan are all profitable, and so was British Rail’s Intercity before privatisation.
If a service can run without operational subsidy, it should: the state should not throw money at the railways to reduce ticket prices artificially.
There are a few reasons for this. Crowding is the obvious one: if you reduce prices, more people will use the railway, and nearly every successful railway system already struggles with the problem of overcrowding. Even though my wallet often wishes Tube fares were cheaper, no line has the spare capacity to make this feasible.
The second is opportunity cost: the money used to subsidise ticket prices can almost certainly be used more effectively elsewhere. We live in a world of finite resources, and need to allocate them in the way that most effectively achieves our goals.
Third, subsidised fares can have perverse political-economic effects. One of them is the moral hazard problem: the transit agency has no incentive to reduce its operational costs if it knows the taxpayer will pick up the bill. Similarly, subsidy can lead to the proliferation of pointless services which are underused, or it can be spent on things that don’t actually improve transit provision.
Probably more importantly, the transit provider stops optimising for delivering a good product, and instead optimises for getting subsidy. It is in the operator’s interest to try to squeeze the state for as much money as possible, inevitably forecasting dire consequences (‘we’ll have to halve service if we don’t get this money’) to trap politicians into giving it what it wants. This often has another pernicious consequence: money that could be spent on infrastructure upgrades instead gets channelled into operational subsidy to plug the gap in the budget.
This is not to say that these problems are inevitable: as we shall see, countries with high state capacity have well-designed transit subsidies that largely avoid these problems. But we should be extremely alert to them.
Making transit profitable
But there are many railway systems that struggle to make money. Short of shutting them down, what can we do?
Some systems definitely could be profitable, but are not, due to incompetence. The New York City Subway falls into this category: it uses operating practices that are stuck in the 1950s (or earlier), which leads to poor service that depresses passenger numbers. The same could be said for all other operators in America, and some in Britain and continental Europe too.
What is the path to profitability? It means attracting passengers by learning from what successful places do well.
All lines should be electrified and operated with electric multiple units, not locomotives pulling carriages.
Infrastructure should be maintained and be kept up to date. Signalling should be digital.
The trains should go directly into the city centre.
The trains and platforms should be designed for getting people on and off the train quickly: wide doors, and no big horizontal or vertical gaps between the train and the platform.
The same basic service pattern should repeat every hour. Trains should operate frequently enough, and be long enough, to soak up most of the passenger demand.
Stations and trains should be clean, safe and comfortable.
Timetables should be designed so that they are robust and reliable.
There shouldn’t be one agency owning the tracks, and another operating the trains; infrastructure and operations need to be planned together.
The minimum number of staff required should be used to operate the service.
We should also look at the other side of the equation: increasing demand, so that more people want to use the trains. The term ‘transit-oriented development’ is ugly and American, but the concept is common sense: building homes and amenities near stations in the suburbs, and offices and amenities near stations in the city centre.
The case for subsidy
Getting these things right will increase passenger numbers and fares. But there are some cities where, even with the best operations in the world and plenty of transit-oriented development, it will still be hard for the transit network to be profitable.
Mass transit needs a critical mass of people using it: even though the Munich U-Bahn is profitable, transit in the surrounding region only covers 80% of its costs, due to the lower overall population density. I could imagine something similar happening in most of Britain’s secondary cities, unless they massively expanded in population.
In most cities with a mature transit system, the social benefits of railways are substantial, even if not all of them are captured through fare revenue. Roads are a much lower capacity mode of transport than a railway, which means they impose more social costs. Congestion is the most obvious one, but there is also car-centricity: cities need to be redesigned around the car so that people can still get where they want to go.
Trains help mitigate these problems. They get cars off the roads, and facilitate good urbanism. Subsidy is the price we pay for avoiding living like Americans.
But there are some types of railway line where the case for subsidy is much murkier than it is in the Munich city region or in Liverpool. Take rural railway lines: they are never going to have enough demand to cover their high fixed operational costs, and their role in reducing the social costs of cars is less clear.
Whether subsidising these lines is a poor use of resources is debatable, but in practice they still remain open. Sometimes rural lines really do have a lot of potential, either because they help tourists get around, or because they are used by people who want to work in the city and live in the country. Voters really like having their little local railway line trundling along, because having the option to use it is still valuable, even if the option is only exercised infrequently. This is probably why, in Britain, the memory of the Beeching cuts, where about a third of the British railway network was cut back in the 1960s, still runs deep. Whatever the arguments for or against subsidising unprofitable lines, the logic of political economy suggests that they will get money from the public purse.
In summary, there are three categories of railway network.
Networks serving big, dense cities, which can and should be operated profitably. Intercity networks also fall into this category.
Networks serving smaller cities and their city regions, which cover most of their costs, but often need some subsidy to get over the line. This is justified because they mitigate the social costs of cars.
Rural branch lines, where voters are keen on subsidy, even if it might not strictly be justified by pure cost:benefit analysis.
The last two categories will need money from the public purse. So – how do we make sure that the railways don’t fall into the traps described above?
Doing subsidy right
Switzerland has the answer. The country uses the so-called Bestellerprinzip, or customer principle: in effect, the government ‘buys’ unprofitable local services from the railways and bus operators. The Bestellerprinzip doesn’t apply to long-distance services, which are expected to operate profitably.
The Swiss cantons are responsible for deciding what level of service they want to operate: hourly, half hourly, and so forth. The cantons have similar powers to US states, but are much smaller: the largest, Zurich, only has a population of 1.5 mn, and the median canton has 240,000 people. Together with the Swiss federal government, the cantons pay to make up the revenue shortfall, usually in a 50/50 split.
This is the right way to do subsidies. The wrong way is just to give a lump sum to the transit agency to carry on operating. This leads to all of the perverse political-economic dynamics I described above. American transit agencies optimise for acquiring subsidy, not for providing good service. Equally, if the amount of money is insufficient, then the transit agency has to cut service, but the politicians can just blame the transit agency for operating inefficiently. Neither side takes responsibility for the service, nor reckons with both the costs and the benefits of its decisions, nor has much of an incentive to invest lots of money today to improve the operations, so that subsidy reduces tomorrow. Investment can perversely act as a tax on subsidy.
Britain does things differently, but not better. Since the pandemic, all fare revenue goes to the Department for Transport, which then pays the operator to run the service, and makes up the shortfall. DfT holds the purse strings, so ultimately controls the operator – but there is no reason to think it will be optimising for delivering a good product to passengers, or even for the long-term profitability of the railways. It is much more likely that it will optimise for short-term cost control, which means service cuts.
The Bestellerprinzip does not entirely avoid these problems, but it is definitely superior. Political actors have to reckon with both the benefits and the costs of their decisions. There is a clear feedback loop: expenditure levels directly determine service provision. The canton can choose for itself whether (say) it wants to build a new school or increase the frequency on a rural branch line.
As well as the opportunity cost, there is another built-in safeguard against pointless services: the Swiss method of doing timetabling means there is little reason to spend money on a poorly-used direct service from A to C, because the few passengers who want to get there can easily change at B.
The right politicians also make the decisions. Decision-making at the level of the Swiss federal government may well lead to a principal-agent problem: the Swiss government purports to act in the interests of all Swiss, but in reality it would be extremely tempting for the government in Bern to cut back service to distant Alpine communities unless they protested loudly. Conversely, if (as is true in practice) the Alpine villages were politically powerful, they might get a better service than they merit. And if decision-making were done more locally than the canton, at the level of the town or village, there might be a co-ordination problem: every town along the line would need to agree on subsidising it, which might prove difficult.
The canton is the right scale of government to decide on local transport provision. If the option to use the rural railway service provides enough value to the canton’s residents, then they are able to use their taxes to pay for it. The trade-offs are being made at the right level.
Empirically, this approach works: the Swiss have the best operations in the world, and provide the best possible product to passengers. But the subsidies are at the end of the chain. The Swiss, first and foremost, have good operations: the list of best practices I gave above essentially describes Switzerland. As a result, they run their intercity network profitably. Subsidies are not alms doled out to the railway, but the result of local politicians making the right trade-offs for their area. Britain, needless to say, could learn much from this approach.
"...no railway in North America is profitable."
I realize this is about passenger rail and not freight, but I find the snide dismissal of the US railroad system annoying, inasmuch as we have an excellent and highly-profitable railroad system in this country: it just doesn't carry passengers, is all. How many publicly-traded, privately-owned railroad companies does Europe have, I wonder? Because we certainly have plenty.
As an aside, it would probably be impossible to have a European-style passenger rail system outside the Northeast--indeed as I understand it Amtrak is fairly close to breaking even in the Northeast Corridor--owing to the punitive passenger-to-distance ratios and the unacceptable amount of time railroad journeys outside the Corridor would take compared to flying. At the height of the railroad era, it took fifteen hours to take either the Pennsylvania Railroad or the New York Central System from New York to Chicago: the flights take about an hour and are on clocker schedules.
If you're traveling inter-city in the US, the received wisdom is that people will generally drive for trips that take less than three hours: more than that, they'll generally fly. Indeed, on this logic Amtrak spent a great deal of money over the years making the main lines (Boston-New York-Washington) able to sustain high-speed trains and electrifying the line between New Haven and Boston. And it has paid off for them: people do, in fact, travel using Amtrak between New York and Boston or Washington, now that improvements have driven both those routes below the magic three-hour threshold.
As to urban and commuter rail systems, it's become apparent that buses using express lanes are faster, cheaper, and (crucially) far more flexible than rail. So that's what we mostly do here, despite the best efforts of railfans to push for more commuter rail.
Thanks for writing this, it clarifies a lot. The "Americabrain" point is realy insightful.