The Importance of Service Frequency to Attracting Ridership

A few months back, I gave a presentation at the Canadian Urban Transit Association in Halifax. I examined the relationship of service frequency and ridership through a look at two very successful transit systems in suburban Toronto that have taken somewhat different approaches to transit service in recent years. Brampton Transit has established a grid of high-frequency routes that cover nearly the entire built-up area of the municipality. Meanwhile, York Region Transit has focused on two BRT corridors in the southern built-up part of its service area.

The presentation offers some arguments for transit agencies that will enable them to advocate for public investment in improved service.

The comparison demonstrates that adding service can have dramatic positive effects on ridership. Over the medium term, increased ridership can also make up most of the financial cost of additional service. Even though Brampton is mostly typical North American cul-de-sac suburbia, it has seen dramatic increases in ridership in recent years as it improved the frequency and service span of its key routes.

Here is one example of a route that saw spectacular improvement when service frequency went from about every 45 minutes to consistently every 30 minutes on a Sunday. The growth is striking because on Saturday, which saw no improved service, demand growth was minimal. This offers something of a natural experiment given that any external factors such as new development or demographic change would have affected ridership just as much on Saturday as on Sunday.

We can also clearly see that the ridership increase occurred following the service improvement, and the ridership growth tapered off in subsequent years as service stayed the same. Clearly, reliable transit on predictable schedules can have a very strong positive effect on people's willingness to take transit.

The presentation concludes with a simple regression that examines the effect of service on ridership. It compares the transit mode share in adjacent census tracts in the TTC and York Region Transit service areas. They are intended to act as proxies for transit service, since TTC offers a considerably more extensive service than YRT. Even controlling for population density and household income, there is a statistically significant positive relationship between being located in the TTC service area and transit mode share. In other words, the regression suggests that transit service is a key factor in determining transit ridership, and not merely population density and income.

An Easy Way to Give East New York a New Subway Stop

I just wrote an article for Urban Omnibus about the opportunity for an affordable extension of the 3 train in East New York.

"Imagine there were the possibility for the Metropolitan Transportation Authority to extend a subway line to a major concentration of new affordable housing — and a neighborhood with some of the longest commutes in the city — without building a single foot of new subway track. 

"That chance exists right now in East New York, where the 3 train’s tracks continue nearly a half-mile beyond their current terminal at New Lots Avenue to the Livonia maintenance yard, near the Gateway Center mall, Spring Creek Nehemiah, and several large public housing projects. Using these tracks for passenger service would significantly enhance transit access to a major development area at a low cost, and spread the city’s current subway expansion program beyond Manhattan."

The O-Train Model: Affordable Rail Transit for North America

Source: JVL Photography (www.jvlphoto.com)

We’ve become accustomed in recent years to the idea that even modest rail transit projects cost hundreds of millions or even billions of dollars. But there’s a model right here in Ontario that shows that high quality rail transit can be developed for a small fraction of that cost. Ottawa’s O-Train uses off-the-shelf German railcars to provide fast and reliable trains every 15 minutes. It’s 8km long, has five stations, and is grade-separated so that it never has to wait for turning cars or red lights. And all of it was built for $21 million dollars, the cost of less than a kilometre of Toronto on-street LRT.

How did Ottawa manage to develop a high quality rail transit service for so little money? It starts with using an existing lightly used freight rail corridor that provided a useful north-south route across the city through Carleton University. Another key was defining it as a “pilot project”, so it was exempted from out-dated regulations and didn’t require years of multi-million dollar studies before construction could begin. Rather than buying custom-made trains like most projects in North America, no matter how small, they bought off-the-shelf Talent trains that had been proven in widespread use in Germany. Finally, rather than building elaborate multi-million dollar stations, they either adapted existing Transitway stations or built a simple asphalt platform with an ordinary bus shelter. This focus on simplicity, ease of construction, and off-the-shelf equipment allowed the project to be built in a fraction of the time and at a tiny fraction of the cost of normal North American rail transit projects.

The possibilities opened up by the order-of-magnitude lower price per kilometre of an O-Train-style project are immense. Smaller cities and even rural areas can enjoy quality rail rapid transit if they have an existing rail corridor that provides a useful route. For example, Cambridge—currently excluded from the Waterloo Region LRT project—could be included with a route from Fairview Mall through Preston to Galt along the existing freight spur for a cost that would be a rounding error on the overall LRT project. With modest track work, Windsor could have a line from Devonshire Mall or Walkerville through the University of Windsor to Lasalle. These are just a few examples of small and mid-size cities that could easily afford rail transit if it comes at the price of a bus lane.

The O-Train model is also relevant for major cities on peripheral or suburban routes. In New York, this approach could be used to activate the Rockaway Beach branch for a fraction of the projected cost, or even (with some additional challenges) to build the TriboroRx. It can then be upgraded over time, as demand growth requires. The O-Train should be a model that is emulated throughout North America.

CityRail: Getting Toronto Moving

CityRail: Getting Toronto Moving was published in Urban Toronto on January 28, reintroducing the CityRail concept. It has generated considerable interest and feedback. Best of all, it has inspired people to build on the concept with maps and presentations. Designer Iain M. Campbell created an extraordinary graphical presentation that effectively explains what CityRail is and why we need it. There is also an excellent new map of the concept.

CityRail now enjoys a permanent home on its new Facebook group. Only a week old, it already has over 70 members and is growing quickly. There is also a CityRail Twitter account for immediate notification of all updates: @CityRailToronto.

I hope that the momentum of the CityRail concept and the campaign for rapid transit on our existing rail corridors will continue to build. There will be many articles to come going into greater detail and explaining the benefits that CityRail will bring.

Neptis Report Response

This post is in response to Michael Schabas, who very kindly responded to my review of his report on Metrolinx's "Big Move" prepared for the Neptis Foundation.

I greatly appreciate you taking the time to respond to my review. I found your study of Metrolinx’s plans to be very informative and well thought-out. I also felt that your inclusion of real regional rail in the study was an excellent choice.

I definitely agree that loco-hauled trains have a purpose, but I think they’re best used on longer distance, limited stop trips where their acceleration time penalty is less important. That’s generally how they are used in Germany: Dostos operate many RegioBahn and RegioExpress services, while S-Bahns in major cities are all-EMU. I’m definitely a supporter of two distinct levels of service, like in Germany, with CityRail providing rapid transit-style service roughly out to the inner edge of the Greenbelt, and a Regional Express service, perhaps locomotive-hauled, out to places like Barrie, Kitchener, and Niagara Falls. It could make limited stops within the GTA and operate less frequently, perhaps hourly or twice-hourly. Bilevels would be less of an issue on those trains as the bilevel loading and unloading time penalty would be less problematic than on the high-frequency CityRail services. They would be a good place to use the existing GO rolling stock for the rest of its service life.

While I am obviously a big supporter of real regional rail, I think that the DRL is a very important project precisely because the corridor can’t be very well served by regional rail. Connections from surface routes, including the Eglinton Crosstown, to the Richmond Hill line, which is the closest parallel to the DRL, would be extremely awkward given its deep valley location. Its winding route makes it slower than other regional rail routes and the risk of flooding also poses problems in an era of concerns about “resiliency.” I agree that connections at Main Street could be designed to be as painless as possible, but it is still not an ideal setup and would impose a time penalty. The DRL is a case where, unlike the other regional rail corridors, the existing corridor isn’t adequate to meet the area’s needs. A new corridor is needed.

The DRL is also incredibly useful because it would be able to provide a very fine-grained service to some of the fastest-developing parts of the city. It’s hard to argue that stations in Leslieville, Liberty Village, Cityplace, the East Bayfront, or the West Don Lands would not be very well used. While I agree that some of the DRL’s future riders are currently be using the east-west streetcar routes, those lines provide a pretty poor service over long distances, which is why their ridership has been dropping steadily for two decades. Better service could attract back many people who have given up on the 501. As it extends up Don Mills, which wouldn’t be possible for a regional rail route, it would serve the heavily developed area along that street as well as effectively connect to a number of very busy bus routes. This would also dramatically improve service for many existing riders and could attract people to transit who currently find the car more attractive than a long bus ride to the Yonge subway. I am not certain how well the TTC’s ridership modelling for the project accounts for all of these potential new sources of ridership. Even if it would not necessarily add as many new riders as other routes, as your report suggests, I think that the dramatic improvement in service quality for thousands of existing transit riders makes the project very worthwhile.

It makes perfect sense that Bombardier would be much more receptive to a contract cancellation if they got a comparably large contract as a replacement. That would be a very good way to reduce the cost of cancellation, though this decision would need to be made almost immediately.

Fare Integration: The Essential Next Step

This article is shared with Urban Toronto.

Regional fare integration is the critical reform that would transform Toronto from a transit city to a transit region. Despite the opportunity presented by the introduction of the Presto fare card, it has been at most a subject of minor discussion, overshadowed by debates over capital spending priorities. For the region as a whole, however, it is far more important than any individual subway or light rail project.

My earlier post about the effect of feeder buses attempted to disabuse readers of the notion that density is the only determinant of transit system success. Yet even simply considering population density, it is clear that there is no fundamental reason for transit ridership on one side of Steeles (or the Etobicoke Creek, or the Rouge) to be vastly lower than in virtually identical neighbourhoods on the other. The lower demand is a result of poorer service provision and an unattractive fare structure. The Steeles bus in Toronto runs better than every 10 minutes until after midnight, while the 2 Milliken bus a handful of blocks away in Markham runs about every forty minutes for most of the day and stops entirely before midnight. Equally important is the fare: while a rider can travel from northeast Scarborough to Long Branch for the price of a token, twice that fare is required for a trip of a few kilometres from Denison down to Sheppard, for example. There is no rational public policy argument for why some trips should cost multiples of other longer trips simply because they traverse an invisible jurisdictional boundary.

Fare integration offers even more transformative potential when it is applied to regional rail. The earlier post on density has demonstrated how vital convenient and free transfers are to building a high-quality transit network. A high-frequency service on the current GO corridors, like the CityRail proposal, can only succeed when passengers can transfer freely to connecting bus, subway, and LRT lines, like they do from the subway to TTC buses today. It's far more important than any physical infrastructure project for the success of CityRail. The benefits would be immense, greatly increasing ridership on the GO corridors to mitigate for the inevitable reduction in fares for some passengers. It would also reduce the need for parking at GO stations. Perhaps most importantly, it would bring rapid transit to countless areas even within the City of Toronto that currently have none. Weston residents know that their community is one of the most challenging in Toronto to reach by transit. A ride on the 89 bus from Weston can take three-quarters of an hour in rush hour just to get to the subway at Bloor. Yet this bus is crowded, while comparatively few people ride the GO train a block away that could whisk them to Bloor in a few minutes, and all the way downtown in less than twenty. Why? The insistence on providing service only for 9-to-5 commuters to downtown is certainly an important factor. But even if trains ran every ten minutes, all day, few people would choose to ride them if it would mean paying for a $4.50 GO Train ticket on top of their TTC fare. Fare integration would bring rapid transit to an array of neighbourhoods just like Weston, without the need to spend billions on tunnels and other mega-projects.

Suburban Toronto subway stations are well-used because of the large number of people transferring from connecting buses. If fares for buses connecting to the new York Region subway extensions are not fully integrated and passengers are forced to pay a transfer fare or even a double fare, they will be far less useful than they would otherwise be. Toronto would be ignoring the most valuable lesson that makes its existing system so successful.

How could fare integration be implemented? In North America we are accustomed to the idea that different agencies and municipalities should operate as independent fiefs with virtually no provision for connection between them, but such an approach would be seen as downright bizarre in much of the world. In Germany, for example, “traffic unions” administer fares and schedules so that transfers are seamless between modes, and riders would never know that they are actually riding vehicles operated by a variety of different agencies and even private companies. Equally importantly, they allocate revenues fairly to the participating agencies. This approach might work better in Toronto than a mega-merger of all GTA transit agencies, and Metrolinx could be an excellent body to administer such a union.

An integrated fare system could not, of course, simply extend the flat TTC fare to the entire GTA; some kind of zone fare would be required. While it is certainly a matter for more detailed study, there are a few basic approaches that are worthy of consideration. The first option would be for large concentric zones. In Berlin, for example, the metropolitan area is divided into three zones: zone A is the inner city, zone B is the remainder of the city proper, and zone C is the outlying suburbs and rural areas. Normal tickets are always for two zones, so that crossing a zone boundary does not result in a sudden fare increase and any rider can travel within two adjacent zones for the base fare. This approach has the benefit of simplicity and means relatively little change for most riders. For example, an area roughly approximating the old City of Toronto could be zone A, the remainder of the City of Toronto could be zone B, while the 905 suburbs could comprise zone C. A basic AB ticket would be just like an existing TTC fare, permitting a rider to cover the entire City of Toronto. The BC ticket, however, would be revolutionary for the inner suburbs. A person living in Thornhill and working at Yonge and Sheppard would no longer be penalized with a double fare, and would pay a similar fare to what someone travelling the same distance within the City of Toronto would pay.

While the aforementioned approach has the benefit of simplicity, it does not necessarily provide the fairest system. Covering the entire 905 with one zone is not necessarily practical; a more radial approach might be required. Using modern smart card technology like Presto, it is even possible to implement true fare-by-distance so that a rider’s fare very closely correlates with the distance he or she has travelled. This produces a far more equitable fare system, though it would be more challenging for riders to plan how much he or she will have to spend for an occasional trip. While trips from Scarborough to Downtown, for example, might see an increase in price, trips from a person’s Scarborough home to Scarborough Town Centre could become dramatically cheaper, balancing out the effect for most riders. A major reason for the significant drop in ridership on downtown’s east-west streetcar routes over the past two decades has been the unreasonably high fares charged for a short trip. With lower fares for short distances, many downtowners would likely return to transit. 

There is no question that there would be some cost, at least initially, to implement these proposals. The fare collected from passengers crossing municipal boundaries would be lower, though the dramatic increase in ridership produced by a fairer system would likely mitigate most of the revenue loss, especially over the longer term. For GO Transit, the change would be more fundamental. It would need to entirely transform its mentality from one of a commuter parking shuttle to a true rapid transit system. In the short term, revenue per passenger would drop, but that’s why this kind of reform would best be combined with a plan like Cityrail that would be simultaneously redesign the network into a rapid transit system that would be able to accommodate the inevitable influx of new riders.

Toronto has been unfavourably described as “Vienna surrounded by Phoenix.” The reason the GTA has developed this way is that the high-quality transit service operated by the TTC, especially the subway, was restricted to the old Metro Toronto. Once the suburbs of the city crossed into the 905, transit service was limited to comparatively infrequent buses and GO Transit’s parking-lot-to-downtown commuter shuttle. A region can’t be expected to have transit-oriented development when it doesn’t provide high-quality transit. But rectifying the fare imbalance with the 905 and transforming the GO system into true rapid transit that is as seamlessly connected with buses and subways as TTC buses are to the subway today would go a long way to upgrading Toronto from Transit City to Transit Region.

Ultimately, the type of zone fare system is far less important than the fare integration itself. A transit system which removes artificial jurisdictional boundaries from its fare structure, charges riders purely on the distance they travel, and allows riders to choose the transit option that gets them to their destination most quickly and reliably would lead to an explosion in ridership in the GTA that would place the region at the forefront of global transit metropolises.

Reacting to the Neptis Big Move Transit Review

This article is shared with Urban Toronto.

Debates over Metrolinx's Big Move returned to the headlines this week with the release of a detailed new report prepared by planner Michael Schabas for the Neptis Foundation. While some observers may dismiss it as “yet another” study of transit in Toronto, we can never have too much information about a plan that will, after all, cost tens of billions of dollars. The report brings some useful and occasionally provocative suggestions to the table and also effectively criticizes some of the weakness of the GTA’s transit planning process. This article will examine some of Schabas’ conclusions.


Regional Rail and Fare Integration
I was extremely pleased to read a detailed assessment of the enormous benefits for reasonable cost that would be produced by real regional rail in Toronto. As Schabas effectively argues, Metrolinx’s GO electrification study was flawed as it concentrated on a mediocre and arbitrary 30-minute frequency that research demonstrates is inadequate to generate the massive ridership increase that comes from passengers not needing to rely on schedules, and because it insisted on the retention of massive 10-car bi-level trains. Both of these assumptions greatly limit the potential benefits of electric multiple unit operation. More importantly, the study did not consider the huge ridership that could be gained by allowing riders to pay the same fare to ride regional rail and local transit.  (More on fare integration in an upcoming article)

Schabas’ solution is to maintain a separate fleet of locomotive-hauled bi-levels for peak period service while using smaller electric multiple unit trains to maintain high frequency off-peak. While certainly a better approach than 30-minute infrequent bi-levels all day, it does not correspond with best practices on most real regional rail systems. They manage with a single fleet for peak and off-peak even with far higher ridership than Toronto. Many German S-Bahn systems, for example, move far more people than GO in the peak periods with single-level multiple units. They accomplish this through high frequency, just like the subway, which also moves far more people than GO. Bi-level cars may seem like a reasonable solution to add capacity, but in fact they are one of the biggest causes of capacity limitations on the system: because they take so long to unload at Union Station, headways are severely limited. With a better platform layout and EMU trains with no stairs and more doors per car to speed loading and unloading, frequencies of five minutes or better would be possible. Such a system would be able to move far more people in the peak period than even GO’s massive 12-car trains.

The benefits of EMUs go far beyond shorter headways and reduced emissions; a cutting-edge regional rail multiple unit like the Stadler FLIRT or comparable models from Bombardier would provide dramatic acceleration improvements over existing GO trains. A FLIRT making all stops from Hamilton, for example, would be as fast as existing GO trains from Hamilton running express after Oakville. This would permit the addition of more stops for rapid-transit-style service without sacrificing travel times. (The benefits of electrification will be examined in greater detail in a future article.)

Loco-hauled trains could be retained if necessary for longer distance trains to outlying cities like Kitchener or Barrie, which would likely be limited-stop services where acceleration is less important.

CityRail Plan central area, cartography by Craig White

The Downtown Relief Line
The Neptis report goes a bit astray when it examines the Downtown Relief Line. There is no question that real regional rail (not “GO Trains,” as has been reported in the media) would provide significant relief to the subway system. However, that is only one of the many benefits of the DRL.
The DRL would be extremely useful even in the context of CityRail because it would provide service to riders in the few areas that aren’t particularly well served by regional rail. Schabas suggests that riders on the Danforth line could easily transfer to regional rail at Main Street, but that connection is in reality quite awkward. Given that passengers would be required to walk considerably further than the Spadina station connection between Bloor and YUS lines, it is likely that relatively few people would choose it over continuing to transfer at Bloor-Yonge.

The effect on the Yonge Line north of Bloor would also be limited. Most riders on the line, which will become increasingly overcrowded as it is extended north, transfer from connecting bus routes from the east. These riders would switch en masse to a Don Mills extension of the DRL, dramatically reducing congestion on Yonge and providing much better service to riders in that part of the city. Unlike the Georgetown corridor, for example, regional rail in the Richmond Hill corridor would not connect very effectively with surface routes because of its deep valley location.

Finally, the DRL serves some of the fastest-developing parts of the city. The waterfront, East Bayfront, Portlands, Cityplace, and Leslieville areas are all seeing massive growth and development. Furthermore, it serves areas where existing transit service is slow and unreliable. It could reverse the significant ridership declines that the east-west downtown streetcar routes have suffered over the past two decades.

The Downtown Relief Line, map by Christopher Livett

The key problem with the cost/benefit case for the DRL as evaluated in the Neptis report is the extraordinarily high cost estimate provided by Metrolinx. The most striking feature of almost all transit planning reports over the past decade is the complete absence of attention to cost control. Very few studies include an examination of different approaches (i.e. underground vs. elevated) or routes and the cost implications. This is in stark contrast with earlier reports, such as the original 1985 Downtown Relief Line study, in which cost was the primary factor being considered when different routes were evaluated. That report concluded that the most economical routing would be along the rail corridor from Bay Street to the Don River, where vacant land is available for a subway. The cost savings would surely be dramatic since virtually no new infrastructure beyond tracks and surface stations would be required in that segment. It would have the added benefit of running right through the heart of the rapidly developing East Bayfront and West Don Lands areas. Such a route does not appear to have been considered in the contemporary DRL reports. The Don Mills segment, as well, is planned to be built entirely underground even though an elevated routing through that area would clearly be feasible and would likely produce enormous cost savings (See "The Rising Cost of Rapid Transit Construction" for more detail). Underground construction costs in Toronto are becoming increasingly out-of-line when compared with peer cities both in Canada and Europe.

Schabas uncovers a particularly striking case of inattention to costs in the Scarborough rapid transit Benefits Case Assessment:
 “TTC seems to be requiring a fairly elaborate and expensive yard. The BCA (which was prepared by consultants) notes, ‘The cost of a Vancouver facility with comparable capacity was roughly $200m lower, although the yard alignment and maintenance practices differ from the TTC’s.’ If Metrolinx thinks there may be the opportunity to save $200 million, surely it should give this more attention than a short footnote?” (60)


Automated Light Metro
Certainly one of the most provocative elements of the Neptis report is its advocacy for automated light metro. While the technology has been extremely popular and successfully implemented around the world, in cities as varied as Copenhagen, Paris, London, Vancouver, Tokyo, Moscow, and Madrid, debates about transit in Toronto have remained rigidly within the subway vs. light rail framework.
Part of the blame for the technology falling out of fashion in Toronto comes from the Scarborough RT, which is the only example of automated light metro in the GTA (though the TTC chooses to have an operator in the cab). This is a poor example of the technology, however, that should not discredit an entire approach to transit. Automated light metro does not need to rely on a proprietary technology like the RT with its complicated and sometimes problematic linear induction propulsion system. At its simplest, automated light metro is just a driverless, fully grade-separated train that is lighter and quieter than the subway to facilitate elevated operation when desired.

The Neptis report illustrates a number of significant benefits to the Automated Light Metro technology. Its operating costs over the long term are significantly lower than non-grade-separated light rail since it does not require a driver. The lack of a driver also makes it possible to run shorter trains at maximum frequency all the time for no additional cost. This is a major benefit on less busy routes where the need for a driver could result in unreasonably long waits for passengers. Vehicle costs, in many cases, are also lower than for LRT at a given passenger capacity. Schabas makes the case fairly effectively that automated light metro would be suitable for the Eglinton line, providing faster trips, better frequencies, and greater reliability. Altogether, he persuasively argues that it would provide a significantly better cost/benefit ratio over the long term than the existing LRT plan.

The Neptis report also underplays the enormous importance of transfers from feeder buses in providing the high ridership that makes Toronto’s existing suburban rapid transit so successful. The large majority of riders at Toronto’s suburban subway stations don’t walk to the station from the surrounding neighbourhood; instead, they arrive by bus. There has not been much study about whether bus riders will transfer to a surface LRT that only offers, according to Metrolinx, about 25% faster trips than a bus. This is of critical importance on Eglinton, since if passengers on north-south bus routes decide to stay on the bus until they reach the Danforth line, rather than transferring to the Eglinton Crosstown, it would make the justification of the multi-billion dollar project much weaker.

Vancouver's Canada Line, image by Michael Berry from Wikipedia

The Canada Line in Vancouver is an excellent example of an automated light metro line with a comparable capacity and length to the Eglinton Crosstown line (more on this subject here). It is, however, completely grade-separated and so will offer a considerably faster and more reliable trip than a surface LRT that faces obstruction from traffic lights. It is also fully automated, permitting higher frequencies and lower operating costs, particularly off-peak. Built as a public-private partnership, it cost governments $2.5 billion—less than half of the Eglinton Crosstown—and was completed in time for the 2010 Olympics as planned.

Of course, all of these points assume that redesigning the Eglinton line yet again is desirable. Certainly, it would do nothing to dispel the image of disarray that has surrounded many recent Toronto transit projects. It could also bring significant cancellation costs, though they might be reduced if Bombardier is retained to produce the vehicles for the redesigned line. While it is difficult to argue with Schabas when he says that this project will be with us for decades so it should be built right, there are significant costs to halting and redesigning the project yet again and the risk of the useful project falling through entirely is very real.