The Metrolinx Regional Fare Integration Study studiously avoids one of the most important questions any new fare scheme must face: what is the effect for various types of riders?
As a starting point in examining what might happen, this article looks at some of the basics of travel patterns and fares to see what various Metrolinx schemes might imply. Note that this is not a definitive, accurate-to-the-nth-degree exercise, but a general discussion. The detailed work should already have been done by Metrolinx, but if it has, they are not publishing the results.
My apologies to readers in advance as this is an article more technical than political. Some of the calculations are unapologetically “back of the envelope”, and are intended as estimates, illustrations, not definitive results.
Updated Feb. 14, 2016 at 2:00pm: Comparative information about GO Transit fares has been added at the end of the article.
Updated Feb. 14, 2016 at 4:45pm: A further observation on the relatively low fare by distance paid by GO riders has been added at the end.
An All Distance-Based Fare Structure
For the very simplest of analyses, let us assume that the TTC (and by implication the rest of the GTHA) moves to a common distance-based structure. TTC accounts for by far the lion’s share of ridership and revenue, and so its stats are as good a starting point as any. Toronto also contains the largest number of riders who will benefit from or be hurt by changes in the existing fares, and so it is reasonable to use this biggest-slice-of-the-pie for a worked example.
For the year ended December 31, 2015, the total fare revenue collected by the TTC is expected to be $1,107-million for ridership of 534-million, or an average fare per rider of $2.073. [Source: January 2016 CEO’s Report pp 38 & 41]
From the Metrolinx Fare Integration background study, which in turn cites data from the 2011 Transportation Tomorrow Survey, we know that the average trip length is roughly 7.5km including trips that are captive to one mode or transfer between surface and subway, and trips that are local to downtown, to the rest of Toronto, or cross between these areas.
(Methodology: This number can be estimated from the average trip length charts by service type, or by estimating the trip type and length distributions from the chart below. I will spare readers the mechanics of doing this. What comes out of the process is that the average length of a “local” trip is about 4.2km, while a “rapid transit” trip is 8.9km. Averaging all trips gives a value of 7.3km. What does not make sense, however, is that the 7.5km average includes linked trips, journeys that use both modes, while Metrolinx segregates them. This is not the only problem with the Metrolinx data as we will see later.)
This number is in the same ballpark as one I calculated many years ago when the TTC was attempting to allocate revenue to its routes based on distance travelled, and average trip lengths could be worked out from the overall data. (Total passenger miles divided by total passengers gave average trip length.)
If the average trip is 7.5km, then the revenue per kilometre is $0.276. This number would move up or down depending on the value of average trip length we use.
On a purely distance basis, and with no reallocation for special fare classes, passes or other considerations, a 10km trip would cost $2.76, a 20km trip would cost $5.52. and a 30km trip $8.28.
Metrolinx subdivides trips by vehicle type with streetcars and buses classed as “local” and subways as “rapid transit”. The average trip length for “local” journeys peaks at a few kilometres, but most of the trips lie at or below 7km. It is no surprise that there are many very short trips, especially on a system where passes encourage hop on, hop off behaviour. Such trips have zero marginal cost for riders today, but that would not necessarily be true in the future.
The 7km limit is also interesting because it represents roughly the radius of the old “Zone 1” which was roughly the boundary of the former City of Toronto. A preponderance of trips within this range may be an effect of geography with the “old” city having a more compact route structure and generally better surface transit than the areas beyond.
The “rapid transit” trips tend to be fairly evenly distributed in length with a drop-off to the 10km mark, but a rise again at about 14km. Again, this is a function of geography because the subway termini are about 14km distant from the core, and the many trips taken from those points inward are actually part of longer journeys.
Distance Plus Base Fare
Even though “fare by distance” is often touted as the most “fair” way to charge for travel, in practice, the formula usually involves two components: a fixed base fare and a variable mileage-based component. GO Transit fares are alleged by Metrolinx to use this formula, but in fact the variable component goes down steeply as trips get longer. A journey to Kitchener from Union Station, for example, would be much more expensive on a truly distance-based scheme.
The question now becomes: how much of the fare should be a fixed cost, and how much should be variable. The chart below shows the effect of varying proportions of a base component plus a distance component.
All lines on the chart above converge on $2.07 (the average fare) at 7.5km (the average trip length). For the case where the distance component is 0% (a flat fare), the value of the fare stays at $2.07 regardless of the trip length. Depending on the percentage of revenue assigned to distance, the slope of each line changes, but they remain straight because there is no discount for longer journeys as there is on GO Transit.
Note that this makes no allowance for any elasticity effects such as revenue losses or gains thanks to people who ride more with very cheap, short journeys, or who ride less with expensive long ones. It simply gives an illustration of what happens depending on a basic assumption regarding the degree to which “distance” is part of the fare calculation while keeping the average fare and trip length fixed. If the effect of the right hand side of this chart is to be reduced or capped, then either a lower distance component is required, or someone has to subsidize those long trips.
The green 50% line shows what happens even with a “50/50” saw-off: the cost of long journeys quickly rises well beyond even the TTC cash fare, let alone token rate.
The chart assumes that all travel within Toronto remains under a common fare with no distinction between bus, streetcar, LRT or subway. However, Metrolinx is clearly trying to hive off the subway as a “premium” service (despite its crowding). This would almost certainly raise the cost of trips involving the subway, if only to paper over fare issues with SmartTrack and GO, while the degree to which bus and streetcar trips might become cheaper is unknown.
What is important about this chart is that it shows the effect of a straightforward, simplistic application of fare-by-distance, the sort of thing politicians and planners love to tout because it looks “fair”. What they do not acknowledge is that this actually has the greatest benefit for those who make very short trips, and penalizes those who travel further. This is inevitably “fixed” by bending those lines at higher distances to offset the real implications of linear increases by mileage, and this is exactly what GO Transit does in its tariff.
Motorists, by contrast, have a substantial fixed investment in the capital cost, insurance, parking (for owned home or assigned work spaces), and the price of a trip varies only with the marginal operating costs. For them, the cost vs distance equation only includes these marginal costs (fuel, tolls, mileage-based maintenance), not the full cost of ownership. Unbounded distance-based transit costs can run into problems for long trips where the perceived value of the competition, a car trip, can be lower than the fare. This can work contrary to transit goals that are not purely based on farebox revenue.
There are at least two gaps in this analysis:
- the omission of rides taken by passholders, and
- the proportion of rides taken (before adjusting for elasticity) at each trip length.
The Effect of Transit Passes on Average Fares
Passholders account for a big chunk of TTC revenue and rides. Unless passes are abandoned, including pseudo-pass equivalents such as capped daily, weekly or monthly charges, their trips and revenue do not participate in the “distance based” allocation. If someone is able to purchase any form of ride-at-will ticket, even for a limited geographic area such as one fare zone, they will no longer be paying by distance, and their incentive will be to take as many trips as possible at a fixed or capped price.
Pass users now represent 52.5% of all trips taken on the TTC network, although they are less than this percentage of all riders. They are the most prolific users of transit service, and their behaviour is encouraged by knowing that their monthly travel cost is fixed.
The TTC does not publish details of trip multiples (actual trips taken per pass), but a value of around 75 is not unreasonable. From this, it is possible to convert rides taken with passes to a count of passes sold, and hence an estimate of the revenue generated by passes. On this basis, the average cost per trip for pass holders is about $1.75. The remaining trips and revenue are for non-pass holders yielding an average cost of about $2.50. (This is an estimated number and it should not be cited as definitive.)
What we do not know here is the average length of a pass user’s trip versus the length for one paid with single fare media (typically tokens). Pass users have an incentive to take many short trips because these have no marginal cost. Single fare riders, by contrast, will avoid short trips, or they will bundle trip segments into a single “chained trip” notwithstanding the supposed ban on stopovers. This problem could be resolved, in part, by the adoption of a time-based transfer which would, in effect, be a time-limited pass such as that already in use outside of Toronto. A single fare would buy more travel, and some riders’ overall costs would go down even if the fare were not changed. Whatever happens, attempts to calculate trip length will always be frustrated by deciding just when a new trip begins, and how this relates to the payment for travel when the two are not necessarily linked.
As a general policy issue, it is much easier to sell transit use, at least on the level of short-to-medium urban trips, using fares that relate to a block of consumption, not to discrete trips. If a distance component is desired, this can be achieved by selling passes or single fares covering one or more zones, but retaining the concept of selling “time”, not “trips”. Even for so-called “regional” travel, nothing prevents the creation of a “Hamilton-Toronto” pass which confers unlimited riding on both local systems, GO and everything in between.
This is a fundamentally different way of thinking about “fare by distance” than the Metrolinx approach which remains fixated on the cost of individual journeys.
Trip Length Distribution
As for the distribution of trip lengths, if only a few riders make long trips (over 15km say), then most riders would face a maximum fare/trip of under $4 (see chart above where even the worst case crosses the $4 line between 14 and 15km). However, one must be careful with dismissing small proportions because even 10% of TTC trips represents about 150,000 per day, or as many as 75,000 riders depending on how many trips each of them takes. This is not a trivial number of riders (also voters), and the geographic distribution of home locations would be interesting to see. They are most likely not concentrated downtown.
Metrolinx does not break down mileages for journeys using more than one type of service. The chart above only gives information about each of three service types: local, rapid transit and regional. However, they do provide a table of trips by type.
There is a severe problem in comparing the chart with the table. In the table, “local” trips outnumber rapid transit by more than 2:1. However, in the chart the area under the “rapid transit” curve (blue) is larger than the area under the “local” curve (red) even though “local” has a taller peak by roughly the same factor. Rapid transit has more trips in total in the chart, but it has far fewer than local in the table. Something is awry in the Metrolinx data.
What is missing in the Metrolinx analysis is a simulation of how various fare structures would operate, and what this would mean for various types of riders. This should not be aggregated into averages which would be dominated by the more numerous short trips on local transit, but should be broken out geographically by location and major groups of destinations. Equally important to the concept of “fairness” will be a validation of a cross-border model that addresses the truly annoying short trips such as southern York Region to York University, while avoiding giving GO Transit riders an even lower cost/km of travel than they already enjoy compared to inside-416 TTC riders.
A common question I get is “so what would you do”? This is not easy to answer, in large part because the published data do not provide enough information, and modelling a fare structure is not a trivial undertaking. However, as basic principles, I would argue that simpler is better than complex, and transparent is better than obscure. A fare system must address the goals of transit overall including its social and political contexts, not just be a mechanism to ensure the most money goes into the pot.
“The Big Move” is not only about collecting fares, but of the need for transportation capacity and a network that will allow the GTHA to grow.
The cost of highways, for decades, has been treated as a general expense, not as a pay-as-you-go function with each mile of travel billed through tolls. Even gas taxes, which don’t pay the full cost of roads, are as much a tax on the style and efficiency of vehicle one chooses to drive as they are on actual travel.
Fare-by-distance is a blunt way of looking at fare structures and arises from the perceived inequity for certain types of cross-boundary travel in the GTHA. The implications are far more complex.
The concept of fare-by-distance sounds “fair” until one looks under the covers at its effects and the biases that inevitably exist in any fare system. Distance has a place in fare calculations, but it is not the only component to be considered, and in many cases it should be ignored both for simplicity and to encourage travel by transit.
Updated: Comparative Information About GO Transit Fares
GO Transit fares are substantially lower than average TTC fares on a distance basis. This is discussed in some detail by my article from December 2014 and by Sean Marshall’s article from November 2015.
The Mythology of GO Fare By Distance Pricing (December 2014)
Not so fair-by-distance: GO Transit’s problematic fare system (November 2015)
The fare per kilometre for GO Transit service declines precipitously from values well above TTC fare levels in the short distance range, but settles down to a range of values below $0.20/km for long trips. This is less than the average fare paid by TTC riders on a distance basis.
GO Transit fare increases consistently benefit those taking the longest trip. Although there is a “tiered” increase with an additional 40¢ to 60¢ per single ride fare, the upper tier applies to all fares from roughly Oakville or Brampton outward on the Lakeshore West and Kitchener corridors respectively. The result is that the percentage increase is lowest for the longest trip fares which, over many years, have risen far more slowly than those for short to medium length trips.
The equivalent of a monthly pass (40 Presto fares) to Kitchener now costs $545.70. If the fare by distance were raised to the level paid by riders from Oakville, the cost would be over $800. If the fare were proportionate to that paid by TTC riders, on average, the cost would be over $1000.
Any claim that GO is a true fare-by-distance system as a starting point for GTHA fare integration is bogus. There may be a valid reason to charge a lower fare to long trips, including the need to entice motorists onto GO, but this is just as valid an argument for local travel as for regional travel.
Updated Feb. 14, 2016 at 4:45pm:
Turning the idea of GO passengers paying at TTC rates on its head, if TTC riders paid at GO rates for longer trips (about $0.13/km), then the Metrolinx 14km “rapid transit” service territory would correspond to a fare of $1.82.
The idea that subway fares are somehow too low and deserve to be increased for this “premium” service does not hold up against the fares Metrolinx charges for its own regional network.
On a related note, there is no justification for charging SmartTrack fares at a higher level than TTC fares when this would place them higher than the distance rate offered to GO’s own customers.