Rail – does it pay?

Public transport is all the rage amongst the left of centre. It has a community feel to it which gives us a warm inner glow in these days which are heirs to the plummeting of social capital. It yields benefits in many forms. It typically generates less pollution than cars and it keeps cars off the streets reducing traffic congestion. So it’s a no brainer right?

Well maybe not. One of the new fronts of the culture wars is an attack on this kind of thinking coupled with the attack on the ‘inner city latte sipping elites’. (I liked Leunig’s cartoon about the person who went to her local TAFE to enrol in chattering classes but I digress).

The Demographia push is all about this – a group of people who are arguing, quite persuasively I think, that land regulation of various kinds is killing the affordability of housing. Michael Duffy is a fan and has run the issue hard in his columns and radio program. It’s a much more promising angle for him than greenhouse denialism IMO.

In any event, over the fold is a summary circulated on a Demographia email list I’m on of a study which appears to have integrity – it’s not by ideologically driven hacks – which finds that rail systems of public transport are almost invariably not cost effective even when you include all those externalities. It’s pretty interesting. The original study can be downloaded from the net here (pdf). It says it shouldn’t be quoted (so I won’t). But the summary is over the fold.

Brookings Scholar on Rail Transit in America

Rail transit projects in most U.S. urban areas arouse controversy. Many taxpayer advocates argue against them, contending that they produce very small benefits in exchange for very large taxpayer costs. But transit advocates justify the projects as essential to reduce congestion and improve air quality, by “getting people out of their cars.”

Brookings Institution’s transportation scholar Cliff Winston has taken a quantitative look at this issue. His new paper is titled “On the Social Desirability of Urban Rail Systems,” co-authored with Vikram Maheshri, an economist at the University of California at Berkeley. It appears in the Journal of Urban Economics and is available online at www.sciencedirect.com.

The purpose of the paper is to estimate the contribution of U.S. urban rail systems to social welfare. The authors define the net benefit of a rail transit system as the difference between its benefits, broadly measured, and its net cost to taxpayers. If this difference is positive, it means that the dollar value of the rail system’s benefits is greater than its net cost to taxpayers (i.e., the difference between what the rail system’s customers pay as fares and the total cost to build, operate, and maintain the rail system).

On average, rail transit systems cover about 40% of their operating costs from farebox revenues and none of their capital costs, according to figures in the National Transit Database. That means their net taxpayer subsidy is large, given the high capital costs of rail.

Winston and Maheshri construct an elaborate econometric model to estimate the “consumer surplus” of 25 rail transit systems. This is economists’ term for the benefits to users, over and above the fares they pay. The large systems (New York, Washington, DC, San Francisco’s BART, etc.) all produce significant consumer surpluses. But most of the smaller ones do not; those with “negligible” consumer surpluses are Miami, St. Louis, Sacramento, San Jose, Pittsburgh, Denver, Buffalo, and Newark. Most of these are relatively new light rail systems.

Next, the authors compare the consumer surplus of each system with its net taxpayer cost. On this measure, every single one of the 25 systems has negative net benefits¢â¬âi.e., the annual value of the benefits to users is less (usually much less) than the annual cost to taxpayers. Surprisingly, this is true even for the massive New York City rail transit system, which by itself accounts for two-thirds of the nation’s rail transit passenger miles.

But what about larger benefits to the metro area? Rail systems are advocated not just to benefit their riders, but because they are expected to reduce traffic congestion, reduce air pollution, save energy, etc. So the final step in Winston and Maheshri’s analysis was to estimate the value of these “externality” benefits. They first conclude that the only one of these purported benefits large enough to make any difference is congestion relief. Given rail transit’s low load factor (less than 20% during all the hours of the day these systems operate, generating costs), neither the energy savings nor the emission reductions are significant.

They do quantify the congestion reductions, which are significant because the rail systems attract riders during rush hours, when marginal reductions in cars on the road can make a meaningful difference in the level of congestion. Adding the congestion savings to road users to the consumer surplus gives the total benefits of rail transit. When this total is compared with the net taxpayer costs, only San Francisco’s BART produces net social benefits (though the Chicago Transit Authority system comes close).

All 23 other U.S. rail transit systems are net losers. The net social cost of some of these systems is as follows:

Miami-Dade Transit: $141 million/year

St. Louis light rail: $171 million/year

Dallas DART light rail: $457 million/year

Sacramento light rail: $106 million/year

San Jose light rail: $210 million/year

Pittsburgh light rail: $135 million/year

Denver light rail: $279 million/year

(I checked the whole paper – no Houston)

This means each of those urban areas is poorer by that amount each year.

Winston and Maheshri anticipate that some rail advocates will protest that these systems offer other benefits that are not accounted for in their calculations. For example, rail stimulates some development around rail stations. “But case studies have yet to show that after their construction transit systems have had a significant effect on employment or land use close to stations and that such benefits greatly exceed the benefits from commercial development that would have occurred elsewhere in the absence of rail construction.”

And there is also the claim that rail systems increase the mobility of low-income residents. But the authors point out that the median annual income of rail users in 2001 exceeded $50,000, which was greater than the median income of the general population in that year. So rail’s primary market is not the poor (unlike bus transit).

Overall, then, the authors conclude that rail transit is erroneously believed by the public to be socially desirable, because “supporters have sold [rail systems] as an antidote to the social costs associated with automobile travel, in spite of strong evidence to the contrary.” [emphasis added] They conclude that, in fact, rail transit is “an increasing drain on social welfare.”

This entry was posted in Uncategorised. Bookmark the permalink.
Notify of
Newest Most Voted
Inline Feedbacks
View all comments
david tiley
15 years ago

I can’t imagine New York actually operating without the subway. It must mean the metrics are stuffed, surely.

David Rubie
David Rubie
15 years ago

I can imagine Sydney without rail. At least without CityRail. Buses are more flexible. Smaller loads of people that are run far more often, requiring more operators with less skill (think train crash vs. bus crash, both are bad admittedly but the potential loss of life of a bus crash is much smaller).

Perhaps Sydney should ditch the train system and install busways through those corridors (and allow emergency services through them with the buses) but exclude private cars. I think everybody would appreciate a system that had a bus every minute vs. a train every twenty. Put those big bendy buses in there and power them off natural gas to offset the diesel particulates and you’ve got a winner. Even better, the off-hours services that benefit the disabled and disadvantaged would be cheaper to provide than an enormous, inefficient, empty train.

Buses are also cheaper to replace than the expensive system of building new train systems. It’s much easier to phase in a new range of buses than build new trains.

Mind you, when I lived in Sydney I hated CityRail, so this suggestion may just be revenge speaking.

Ken Parish
Ken Parish(@ken-parish)
15 years ago

Then there’s the probably fairly imminent tipping point/s on global warming, not to mention peak oil quite likely coming some time soon with petrol prices shooting up thereafter. Does it really make sense to promote urban sprawl without public rail transport infrastructure in those circumstances? (which is what the Demographia mob are advocating, not to mention the Institute of Public Affairs and a paper they published by a chap by the name of Alan Moran). At least with rail you can reduce or eliminate CO2 emissions by cleaning up the power stations; it’s much harder if our cities are totally dependent on cars and buses, and we’ve failed even to reserve rail corridors let alone actually build rail lines. How would the figures for rail networks look, I wonder, with twice or three times as many rail passengers when petrol costs twice or three times as much as now, and urban road congestion worsens as the populations of large cities continue to grow?

Moran also reckons it would be a great idea if urban development was completely deregulated and developers were allowed to build subdivisions without paved roads or curbing or guttering or footpaths, like they did back in the 1950s when land was cheap and everyone was much happier and nicer. Is this a good idea too?

Demographia/IPA also appear to assume that all cities are the same. But Sydney, which is the large city I know best, is hemmed in by the Blue Mountains to the west, and built on mostly hilly terrain interspersed with steeply sloping jagged estuaries that make development costs unavoidably higher than flatter places like Melbourne (or Houston). Moreover, if anyone really thinks that Sydney isn’t already suffering from major urban sprawl, you haven’t been looking very closely. Leaving aside the shrinking green belt national parks round Port Hacking and the Hawkesbury, what we see is pretty well unbroken urban development from ocean to mountains for a distance of way over 300 kilometres, from Kiama in the south to past Newcastle in the north. Try driving it one day and see if you reckon even more urban sprawl is a great idea.

There are a couple of directions the sprawl could be extended, though – to the north-west past Wiseman’s Ferry or the south-west past Picton towards Bowral. Architect and planner Elizabeth Farrelly makes a convincing case in today’s SMH that this would be a particularly stupid idea:

The north-west growth centre runs from Mulgrave to Bidwill, with a major urban centre at Rouse Hill; the south-west growth centre runs from Kemps Creek to Harrington Park, centred on Leppington. Between them, they will receive 160,000 new dwellings and 180,000 new jobs. Fair enough, you might think. Sydney is expecting maybe a million blow-ins over coming decades (though figures are revised downwards all the time) and they have to live somewhere. But here? In Sydney’s vegie basket?

Sydney agriculture generates $1 billion annually; one-eighth of the state’s vegies from just one-hundredth of the land area. This food production provides much needed jobs in rural fringe areas, often sustaining sizeable immigrant populations and, perhaps more importantly, fresh coriander and strawberries for our delectation. And yet government and council policies, across the Hawkesbury-Nepean basin, will plough hundreds of farms under brick and asphalt. …

But might isn’t necessarily right. It isn’t necessarily bright, either. As Phil Dunesky, of the Pitt Town Residents Group, says, “Without agriculture, there’d be no rural in rural living.” Plus there’s the sheer ugliness, moral and visual, of turning the Hawkesbury into one big housing estate. Not only will the new estates be unavoidably car-based. If we exile our market gardens to the rain shadow beyond the mountains, we’ll have to truck back anything that grows there, or fly it in from south-east Queensland, while those pretty embroidered fields blossom grotesquely with Kellyville max.

Moran and the Demographia mob are dickheads of the worst sort, substituting blind reliance on statistics for observation and commonsense. What we need are more and better rail networks, with “hubs” and “spokes” so that commuters can use the train to get to work right around urban areas not just in the CBD. That’s the real major reason why rail only attracts the more prosperous passenger demographic – most rail lines have been built radially out from the CBD, so it’s effectively impossible to use it unless you’re a middle class office worker commuting to and from the CBD. Most current urban rail networks are useless to factory and suburban retail workers and customers. Some of these needed lateral links could be light rail, some could even be electric or hydrogen-fuelled buses. But to argue for completely rejecting rail transport while promoting unrestricted urban sprawl is almost criminally stupid IMO in a world afflicted by global warming and rapidly depleting oil resources.

David Rubie
David Rubie
15 years ago


Have a look at the cost of replacement trains. Further, the electricity they run on doesn’t appear magically – just because it is spewed out as greenhouse gases in the Hunter Valley doesn’t mean it doesn’t appear. I find the whole “centralised emissions” argument specious in that it doesn’t address the inflexibility of the long (decades) replacement timetable of their expensive, outdated and increasingly inefficient motors. They are 18th century anachronisms kept alive by misty eyed historical revisionists on the right and illogical greenies on the left. The massive improvements in diesel technology in the last 10 years has been astonishing, both in it’s progress and (sadly) lack of adoption.

The corridors we already have are massively underutilised, and poor planning has ensured that no new corridors have ever been allocated. We can and should do better with our public transport and smaller, more flexible units of transport that could cope with missing corridors but thrive on the existing ones are part of that puzzle.

However, I think Australia’s biggest city is the greatest disaster of the age – the efficiency it should have produced in terms of concentration of resources is actually shouldered by the poor bastards who are forced to live there. There are plenty of regional centres in Australia who would be more than willing to spread the concentration of population across a wider, more sustainable area, with less aggregate pollution. We would be far better served with larger numbers of smaller cities that are easier to get around.

15 years ago

I am in favour of urban rail!

If nothing else I use urban rail almost daily, as one of those (low) middle class professionals, and I would hate to drive to work. I think Ken’s last paragraph, shorn of its final sentence, is quite sensible.

All that said and agreed with, I think peak oil is one of the most preposterous chimeras masquerading as an economic theory that I have ever heard of.

As much as I love urban rail, and as much as I have thought for 10 years now that Melbourne needed an ‘outer loop’, planning must accomodate cars, because cars are king.

15 years ago

At best, the paper provides an argument against expanding rail networks, especially in small cities.

On a quick reading, the paper doesn’t acknowledge the existence of sunk costs, so it’s pretty much worthless in terms of assessing the viability of the existing suburban rail networks.

Ken Parish
Ken Parish(@ken-parish)
15 years ago

There’s also a certain irony in the timing of Nicholas’s post, given this morning’s reports of the imminent collapse of the Sydney Cross City Tunnel builder-operator.

15 years ago

relatedly, via instapundit:

November 16, 2006
IN THE MAIL, AN INTERESTING BOOK ON TRAFFIC: Ted Balaker and Sam Staley’s The Road More Traveled: Why the Congestion Crisis Matters More Than You Think, and What We Can Do About It. They argue that traffic congestion does much more harm than is generally appreciated, and that municipalities’ programs aimed at making traffic worse in order to encourage people to use mass transit are deeply mistaken. They also argue that fixing traffic problems is easier and cheaper than is popularly thought. I’ve read the first several chapters and it’s very interesting; I hope it gets a broad readership. Interesting tidbit: If you exclude New York, America has more telecommuters than mass-transit commuters.

I’ve had some related thoughts on this topic myself, here.