It is impossible to browse the news online or in print without seeing an article about transit ridership. Will people ever go back to work in office buildings? Has work-from-home destroyed transit demand? What is the future of cities in general and our town, Toronto?
I am not going to attempt to peer into that crystal ball as there are too many possible futures. However, I want to throw out a few ideas that should be part of any debate or prognostication.
The transit market is not a monolith, not in Toronto and certainly not from city to city, system to system. There are different markets each with its own demand patterns and riders, and recovery of these markets will not all happen at the same time.
The TTC has a particularly broad reach in that, in “before days”, there was strong demand not just for downtown commuter trips, but around the suburbs. Many types of jobs do not have the work-from-home option. They are keeping us fed, alive, and supplied with an unending delivery of goods we once picked up from a local store.
Moreover, many trips are not “work” trips but are for other purposes.
- School trips for all ages are an important market, and these are often overlooked as part of transit demand.
- People make trips for shopping and other personal errands on the TTC, especially if they do not have a car (or if the only one in the family is being used by someone else).
- Finally, there are leisure trips, broadly speaking, for outings to sports events, theatre, movies, an afternoon on the beach or a walk through the cherry blossoms.
Office commuting will resume when employers and employees feel safe gathering together. Yes some prefer working at home, at least some of the time, but others are just as happy to escape to the alternate universe of their work space and colleagues. There were already moves to reduce the space per office worker and design more for “hotelling” to make better use of expensive space, but there is still a demand for office space, at least in the core area. The creation of new space may halt or slow for a time, but “downtown” still has its allure.
We have already seen that workers in critical industries and in many settings where there is no online alternative are straining the transit service despite claims that crowding is rare.
School traffic is not going to disappear either, but it will return under different circumstances and at a different pace. Much depends on when it is really safe to resume gatherings on that level, based on actual health science, not on political pressure to reopen at whatever cost.
The last to return will be the leisure trips because here has to be something “there” to generate the traffic.
This affects the TTC in a different way from GO Transit which is almost entirely dependent on one market: downtown commuters. GO’s future is tied to these riders much more than is the TTC’s largely because it is the only market they pursued for most of their existence. They are highly dependent on parking lots and personal cars for “last mile” feeder service.
GO has tried to market itself for off peak traffic and group travel, but that demand is trivial. It is viewed as a loss-leader to get potential new customers onto GO who would try weekday service after they use it on the weekend. Of course, someone bringing their family in for a ball game might not actually work downtown, and GO has little service to other destinations.
GO’s daily ridership dropped much further than the TTC’s and it remains in single-digit percentages of its former level with some trains carrying loads that would fit on a bus.
The TTC is different as its stats show. Ridership (equivalent to fares paid, or “linked trips” in planning parlance) are at 25-30% of former levels across the system, with higher values in some areas. The budget plans for a growth to about 50% by fall 2021, but whether this is achieved depends a lot on the perceived success of the vaccination campaign and a big drop in future infection rates. In the first quarter of 2021, the TTC expected to see the first glimpse of a recovery but, thanks to political bungling, we got a third wave of infections and another “lockdown”.
That hoped-for growth starting in September may turn out to be wishful thinking, and that does not bode well for a transit system that cannot be kept on financial life support forever. The 2022 budget assumes a fairly healthy growth in demand, although not back all the way to pre-pandemic levels.
“Boardings” or “unlinked trips” count each leg of a journey separately. Any transfer between vehicles (except on the subway) counts as a new boarding. While all modes suffered a downturn through the fall and winter, the bus network did best of the three showing its relative importance to riders who continue to be on the TTC.
Crowding complaints come overwhelmingly on the bus routes, in part because there are so many more of them. The bus fleet has automatic passenger counters that can report real stats, but these are not yet widely available on the streetcar fleet.
In the near future, the TTC will provide a crowding level feed to two apps: Rocketman and Transit App. This will allow riders to see how crowded approaching buses are, although it will of course do nothing to prevent a bus, once boarded, from filling up later. This information, however, will make an interesting adjunct to real time and historical tracking analysis because it will allow crowding and service reliability to be compared.
The TTC’s analysis above fails to show the real situation riders face because it lumps every bus trip on every route all day together. Many routes are never crowded. Routes with chronic bunching might only be crowded on the “gap” bus. Routes with highly directional demand will have a lot of lightly-loaded counterpeak trips. Having “only” 5% of trips show up as “crowded” will understate the degree of the problems when and where they actually exist.
Moreover, there are more riders on a crowded bus than on an empty one, and so the “average” experience will be that more people to see crowding. A good analogy might be that someone trying to board the subway southbound at Bloor in the AM peak as opposed to eastbound at Broadview at the same time will have a very different experience of subway crowding. The same is true for bus routes.
With route level crowding stats, the TTC should be able to provide “hot spot” reports rather than simply averaging the entire system.
Clearly they are doing some of this already because service plans for coming months, including the May schedule changes, will reduce service on some less-loaded routes to be redeployed on other busy routes.
An important improvement would be to include the hot spots/hot time list in the Daily Customer Service Report so that riders can compare their experience with what the TTC thinks happened. The next challenge is to make the service run reliably, a frequent topic on this blog.
Is Ride Sharing Siphoning Off Riders?
At its April 14 meeting, the TTC passed a motion from Commissioners Carroll and Bradford, as amended. The original version sought to have management commission a study on their own, but this was modified to take into account work already done by the City of Toronto and its access to private data from ride sharing companies that would not be sharable with an outside consultant.
The TTC Board direct the Chief Executive Officer to work with City of Toronto Municipal Licensing, Transportation and Planning staff and Standards to study the impact of ride-hailing services on public transit, which builds on the City’s 2019 report “The Transportation Impacts of Vehicle-For-Hire in the City of Toronto” and specifically studies:
a. Rides lost annually to ride-hailing since 2014
b. Ride loss projections based on
i. Anticipated growth of ride-hailing and
ii. Changing rider concerns during and after the pandemic.
c. The corresponding impact on fare revenue and, therefore, the TTC operating budget.
d. Changes to traffic congestion, vehicle kilometres travelled (including commuting, cruising, on route and in-service time) by PTC vehicles and corresponding effect on TTC surface transit caused by the growth of ride-hailing apps.
e. Strategies and solutions to remain competitive in the mobility ecosystem when coming up against ride hailing companies.
f. The potential synergies with ride hailing companies that could drive mutual economic benefit.Sources: Notice of Motion and amendments taken from the video record of the meeting
Frankly I am not sure that the data exist to support this scale of retrospective study. The real question going forward will be whether ride sharing use will go up, and what market(s) are most likely to be affected. Equally important are the questions of why people choose ride sharing over the TTC and what the TTC might do to counter this.
The TTC has a long and rather sad history of assuming that most of its problems originate from external forces and looking there for some way to counter riding losses, among other issues. Rarely do they look inward to ask whether what they are selling is what riders want to buy.
Anyone familiar with the club district in pre-pandemic times will know that Ubers prowl everywhere including driving into TTC curb lane passenger waiting areas. Some people take them rather than waiting for the King streetcar. The economics are simple: for a short trip with multiple riders, some of whom may not be regular transit users (i.e. they don’t already have a pass) the combination of cost and speed is hard for the TTC to beat especially if the journey might involve a transfer or a walk at the destination from a transit stop.
However, most people going to work or to school or to shop are not going to call up a ride share. Those who do make trips this way are quite willing to pay a premium for speed and convenience, and they will be difficult to woo onto transit just as auto drivers will (at best) chortle mildly at the suggestion they might use the TTC.
Rides lost to sharing companies are not new: conventional taxis have been doing this for decades, and the more enterprising cabbies would troll along busy streetcar routes hoping to pick off riders tired of waiting for a streetcar to show up. The difference now is that ride hailing is greatly simplified, notably that one can actually “hail” a ride online rather than hoping a cab will pass by, and the cost for many trips is lower than with a regular cab.
Even I, a transit advocate, have taken cabs (not ride shares) when the circumstances favour it: late at night, bad weather, carrying luggage, a shared ride. Taxis are a form of transit that addresses its own market. Of course it costs more to come home from a night out by taxi than by TTC, but when the occasion demands it, that’s part of the cost of the evening. I have that option because I live downtown and can afford an occasional cab ride. The same cannot be said for those making longer journeys, or of lesser means, and certainly not on a regular basis.
As the pandemic recedes, there will be a period while the desire to ride in close quarters will probably lag the actual effectiveness of vaccines and the likelihood of infection. We have been through enough false recoveries to avoid leaping onto a packed transit vehicle soon.
In my own travels, which are much less frequent than pre-pandemic, I know that riders are very sensitive to crowding and particularly to anyone who travels unmasked. Even if only 1 in 20 (5%) do not mask, that means a typical half-full bus (25 riders) will have at least one unmasked rider.
TTC looks at the stats from the point of view that almost everyone does mask, not from the likelihood that at least one rider per vehicle does not, and worse that such riders might behave in a way that would concern others.
Pooled ride sharing faces the same dilemma, and Uber Pool is currently not available. This increases the cost per trip making ride sharing less attractive. The decline in work available both to conventional taxis and to ride sharing is well documented in an era where travel is down, and the desire to use a shared vehicle, even sequentially, is also dampened.
Nibbling Around The Edges of Service
A common political attack on transit arises when times are tough, when “taxpayer dollars” are the rallying cry for trimming public services. As a starting point, it is important to understand what the TTC’s Service Standards call for today. The following standards affect service quality especially at the margins where demand is lower than normal. (Everything here refers to prepandemic conditions.)
All-Day Every-Day Service
TTC’s hours of service are nominally from 6 am to 1 am, except on Sundays when service begins at 8 am. In practice, many routes operate outside of these hours notably to match to the end of subway service. The last outbound trains leave Bloor-Yonge just before 2 am, and as they move outward, connecting last buses leave their stations. Routes with less late-evening demand end at 1 am.
It is currently TTC policy that service is an “all or nothing” decision rather than having different hours of service tailored to each route, with rare exceptions. The intent is that the network is the same from a rider’s viewpoint for the entire service day and week.
Overnight service is provided on a subset of routes generally on half-hourly headways, although the TTC does not use protected time points in its schedule where vehicles must hold for their times to guarantee reliable connections.
Headway standards only constrain the most lightly used routes. For the subway and RT network, the maximum headway is 6 minutes while for local bus and streetcar routes it is 30 minutes.
Productivity standards dictate the minimum demand (counted as “boardings”) expected from various classes of service. If demand falls below these minima, this would trigger a service reduction. Note that 10 boardings per vehicle hour for off-peak bus service sets a floor below which service would not be offered.
A further metric is the net cost per boarding which is expressed as:
(Route Operating Cost/Boardings) – Average Fare per Boarding
This can be calculated on an all-day basis, or for individual periods of service. The challenge, one that we saw during Mayor Ford’s attempt to save money by culling unproductive routes, is that the marginal saving is less than the average cost per hour or kilometre of vehicle operation might imply for various reasons:
- There is no marginal vehicle-related saving in off-peak service because the TTC already owns them. They require garage space and must be serviced whether they run until 1 am or only to 9 pm, for example. No capital cost is avoided by using buses for fewer off-peak hours per day.
- Running less off-peak service can reduce the opportunity for spreading driver costs over full shifts, avoiding the premium cost of short blocks of work which includes the time for garage trips allocated over fewer in-service hours.
- Off peak service tends to run at higher speeds reducing the number of vehicles and drivers required to provide service.
All of these factors combine to set a standard for when and where the TTC would provide service. Some routes and periods of operation will be more “efficient” or “productive” and we see this even on the subway network.
However, the money to be saved on marginal, off-peak services is generally less than the average net cost/rider on the system.
The question, then, is whether an alternate method of providing service would deliver the standards we expect as a city but at a lower net cost. This brings us to ride sharing and alternative providers.
Uber Contemplates Transit
Uber produced a study, Transit Horizons: Toward a New Model of Public Transportation, of the applicability of their service as a replacement for or supplement to conventional transit. Numbers in the study are based on data reported by systems in the USA.
The short version:
The fundamental conclusion of this report is that ride sharing is not competitive for the vast majority of travel, certainly on large urban transit systems. There is little for Uber to chase within those systems, but by nibbling around the edges, Uber establishes a presence and the marketing presence to politicians who think there is always a way to make transit “more efficient”. This might be true in some parts of the GTHA especially if one regards transit as a service run for low volumes of riders and staffed by drivers making low wages. But it is no answer to how the vast number of potential transit trips now taken by private auto might be redirected to some form of transit.
It is clear that Uber’s real aim is to become a hub and a broker for multiple services including conventional transit, ride shares and other modes such as cycle rentals. Transit represents a lucrative revenue stream, and Uber hopes to supplant conventional fare collection.
The long version:
Reading this report has an odd feel as if there are two separate authors with their own separate takes on the issue.
In the first half of the study, we learn that across a wide range of US cities and transit systems, Uber is not competitive with fixed route transit unless that service carries fewer than 10 trips per vehicle hour. In this context “competitive” means that if a city subsidized Uber trips so that riders paid regular transit fares, the city would be better off just running their own buses provided there was enough demand. The situation is worse as long as Uber Pool is shut down because its economies of scale are lost.
In the second half, the tone shifts and Uber reveals its vision for a transit service it would, in part, provide, but with an important additional role: Uber would integrate trip planning for both the conventional transit system and its own services tailoring trips to the optimal way they might be served. It goes without saying that fare collection would be through this central system, and it would displace some or all of an agency-based system like Presto. This appears to be the real market Uber is chasing, one where it provides service management, allocation and most importantly revenue handling.
The chart below, with calculations by Uber, show how few bus routes on large systems fall into the range where savings with ride sharing are possible. Note that these charts treat each route as one data point regardless of how many riders it might carry.
The 10 trips per hour value is comparable to the TTC’s Service Standards floor and below their all-day averages. Time-of-day breakdowns only show up in reviews of service periods where a route falls below acceptable performance set by TTC Service Standards. This also ties in to the standard of all-day service. A route might fall below the loading standard late in the evening, but the all-day numbers could be much better. Conversely, the marginal saving from dropping late evening service will be lower than the average cost implies. The situation varies from route to route.
The TTC last published statistics for its surface network in 2018. I restated these numbers to show the riders per hour value in What is a “Low Performing” Transit Route [table]. With the recent change to use of Automatic Passenger Counters and real time reporting, we will get a very different and detailed view of transit demand.
None of the TTC routes fell below the 10 boardings per vehicle hour threshold, not even the Downtown Express buses. This means that on an average basis, none of the routes fits into Uber’s range for effective replacement by ride sharing. The situation, however, could be different at the margins depending on demand. This gets tricky very quickly, however, for many reasons including:
- The marginal cost of providing transit service is lower than the all-day average for a route as discussed above making the cost per boarding lower and with it the “break-even” point.
- Demand varies hour-by-hour, day-by-day and depends on many factors. A route that could appear to be “unproductive” on Monday could do quite well on Fridays. It is not practical to schedule service to deal with variable ridership with regular buses on some days and ride shares on others.
- Weather affects both transit demand and ride share pricing, and the last thing a transit rider needs is a service that vanishes because ride shares are not available, or because the transit system will not pay their premium pricing during surge events.
- Riders need to know when regular bus service is offered and when they should call up a ride share. This has implications for how service is provided and managed that I will return to later.
Uber plays fast and loose with claimed savings by lumping a wide range of services together in their analysis. Even then, the savings are not huge:
We estimate that ~1-6% of bus trips could be replaced by ridesharing at a cost reduction for agencies (based on Uber Pool pricing and pre-COVID ridership levels). These trips represent ~5% to 25% of all bus routes. Agencies can expect cost reductions per trip of ~15-30% on inefficient routes that are replaced with Uber Pool (assuming ridership remains constant).
Using a sample of 10 agencies of different sizes from the National Transit Database, we estimate cost per trip reductions of between ~13% to 70% from leveraging ridesharing services like UberX for ambulatory demand response trips. In markets where Uber Pool is available, we have found the cost reductions per trip could be higher than 70% on average.Uber Transit Horizons p12
The reallocated trips might represent “5-25%” of bus routes, and the saving on affected trips might be from “13% to 70%”, but on how many trips out of a system’s total? The table below shows how the vast majority of trips across US systems are carried at a cost with which ride sharing cannot compete according to Uber’s calculations. Large transit systems like the TTC will have trip profiles in the upper portion of this chart with only crumbs for ride sharing.
There is a further policy question of the type of service a transit rider-share might offer and of just what “last mile” service means. In the context of replacing private auto trips, a ride share is door-to-door. Transit, by contrast, is stop based and even a conventional bus dispatched on a demand-responsive basis is not going to pick up and drop off riders at their destination unless it happens to be at a bus stop.
Should a “transit” ride-share be offering a substantially more convenient ride, in effect a taxi rather than a bus, on the very marginal routes and locations where conventional transit is considered to be uneconomic? When we talk about “microtransit”, just what do we mean, and what service quality are we prepared to pay for?
I have omitted here a fundamental problem with ride shares: the “digital divide” that isolates riders who do not have smart phone technology through which ride shares could be easily booked, especially on an anticipatory basis before the portion of a journey on a trunk bus route reaches the “transfer point” to ride share service.
The idea that riders would call into a dispatch centre assumes that everyone either has a phone, or that there is a pay phone at every stop where a ride share might be needed.
Uber looks to nibble around the edges of a transit system at marginal routes during marginal times, but that is not really the basis for a robust business.
Uber’s vision is clearly shown early in their paper:
Consumers are demanding higher quality and more dynamic services together with integrated digital experiences. Agencies meanwhile are expected not to leave anyone behind. Adding to the challenges, the COVID-19 crisis has decimated ridership, funding sources have evaporated, and the recovery is progressing slowly. Against this backdrop, the need for public transportation agencies to evolve and innovate on their service delivery model has never been higher.
At the highest level, we see public transportation systems transforming from decentralized networks, where different modes can often operate in silos, toward a system that is truly integrated, connected, and optimized in a highly agile way.
On the rider side, users will plan, book, pay for, and access their trips across any mode – both public and private – via Mobility-as-a-Service (MaaS) apps. Rewards, bundles, and mobility subscriptions will follow shortly thereafter.Uber Transit Horizons p3 and p5
Uber has a confused attitude to the way public transit costs actually work, and they start with the claim that the costs is largely fixed:
The cost structure of public transportation, similar to airlines, is largely fixed. Whether they operate at 90% or 5% capacity, agency-owned vehicles have a cost per supply hour that varies little, if at all.Uber Transit Horizons p8
Elsewhere, however, Uber touts the premise of “peak shaving” by trimming vehicle requirements. They are confused about saving peak vehicles that would operate when the ride share model cannot possibly compete on a cost/trip basis as opposed to off-peak savings when the only possible cut is in fleet utilization, not size.
Vehicles have costs that are partly fixed and partly variable. Fixed costs include:
- Capital cost of vehicles and garage space
- Routine cleaning, servicing and maintenance of buses
- Facilities maintenance
- Major overhauls such as the TTC’s mid-life rejuvenation of its buses
- System management
Variable costs include:
- Mileage based maintenance due to wear and tear
Labour costs are driven mainly by hourly wages but also by the service design including the amount of dead mileage for garage trips and premiums paid for very short or very long shifts (minimum crew value in one case, overtime in the other).
The cost per supply hour depends a great deal on vehicle utilization, and some costs cannot be avoided simply by running a bus less and replacing it with another service.
Capital costs are a special consideration because, depending on the agency, they may not appear as part of the cost base. If buses appear “out of the air” thanks to capital subsidies, then there is an incentive to maximize capital (replacing buses) rather than maintaining them, usually an operating cost. The TTC is already well down this road.
In the ride sharing model, many of the costs are borne by the vehicle operator who hopes to recoup the investment (or ongoing lease payments) through revenues. Ride Share trips are all an “operating cost” to the transit agency which benefits mainly through the much lower effective labour rates paid to drivers, a “benefit” they would be unable to obtain through contract negotiations with their own staff.
Subsidies in the US (and much of the world) are considerably higher than in Toronto. The chart below shows heavy rail with the best cost recovery no doubt because of its economies of scale.
When subsidies are plotted against ridership per vehicle hour, it is no surprise that agencies with light demand have the highest costs. Note however that many agencies show a negative subsidy/trip for buses despite the mode’s overall low cost recovery in the chart above. This suggests there is something wrong with Uber’s analysis below.