Is There an Optimal Supply and Demand for Transit?

On September 16, the Canadian Urban Transit Association (CUTA) released a study on the optimal supply and demand for transit in Canada.  Although I may have slept through the local press coverage, I don’t think that there was much if any as other issues crowded out the story.  One might ask why I’m bothering with it now, but I think this is worth talking about even though I don’t agree with the premise of the report.

Only the Executive Summary and Backgrounders are available on the CUTA website, and as a courtesy for copyright, I will not post the full version here.  You will have to get it from CUTA if you really want it.

The principal conclusions as highlighted on the CUTA site are:

  • The economically and socially optimal level of transit supply in 2006 would have required an estimated 1.7 billion vehicle-kilometres of transit service, or 74 percent more service than actually supplied.
  • In 2006, capital investment of $78.1 billion would have been required to bring the supply of transit into line with the optimal conditions of supply in that year.
  • Results of the analysis conclude that Canada is clearly underinvested in urban transit.
  • Bringing transit to the optimal level of supply would produce several positive economic and social benefits – more than two thirds of these benefits constitute the economic value of reduced roadway congestion.

There is no question that higher investment in transit is required across Canada.  However, there is a danger with any calculated “optimal” value that this will be taken as an upper bound.  Moreover, the methodology of the study does not address future needs, only the situation that existed in 2006.  It is based on average relationships between several economic variables taken on a national basis that almost certainly misstate the micro-level effects in urban areas.

Governments today have a fetish for economic studies that, among other things, could establish whether services should be delivered in the private sector.  Such studies are fraught with the possibility of error because there simply isn’t a way to foresee how a specific implementation might work out in the long term.  Assumptions made in a study may not even apply soon after the outset of a private contract, let alone ten or twenty years in the future.

CUTA’s study does not speak to privatization, but the large infrastructure deficit invites alternative schemes for transit financing.  Indeed, this is a topic that Metrolinx has clearly stated would be part of their investment strategy.  (We will see exactly what they propose when the strategy is released to the media on September 23.)

The concept of an “optimal” investment assumes that one can calculate the point at which the overall socioeconomic benefit of added transit spending falls below the spending itself.  In simple terms, we spend a dollar, but don’t get a dollar’s worth of benefit in return.  The problem with this scheme is that it depends on the imputed value of many soft variables such as the value of someone’s time sitting in traffic or the potential economic stimulus of a new transportation corridor, not to mention linkages between variables that may not be well-understood over a wide range.

Economists refer to these linkages as “elasticity”.  There are a few examples we see every year in debates about transit fares and service:

  • If fares go up, riding tends to go down everything else being equal.  The factors used in CUTA’s study are minus .25 for peak periods and minus .60 for off peak periods.  This means that for every 1 percent increase in fares there would be a .25 percent loss of riding during the peak, but a .6 percent loss during off peak.  This is easy to understand because other factors bear on the decision.  Peak trips are more difficult to change to auto travel because the cost in time and money is greater.  Therefore, peak trips should be easier to hold through a fare increase.
  • If service goes down, so does riding.  The factors used in the CUTA study are .50 for peak periods and .20 for off peak.  This means that a 1% drop in peak service should produce a .5% drop in riding. The lower elasticity for offpeak trips suggests that service cuts in that period do not have as much of an effect.

The important thing about these factors (and many others used in the study) is that their behaviour is not well understood for large changes.  There is no base of past experience to calibrate the relationships and guarantee that they are linear (no matter how big or small the change in factor A, the proportionate change in factor B is always the same).

We know both from service and fare debates that this is not true.  There has always been a desire (rarely actually implemented for political reasons) for small, frequent fare changes that would be small enough to be absorbed in the “background noise” of other price and convenience issues and limit the impact.  Large jumps in fares produce a greater impact, proportionately, because they are far more visible and may be perceived as being unreasonable especially if service is also less than ideal.  People hate to pay for poor service, but paying more for less service is really bad marketing.

Using a number of assumptions that I won’t detail here, the CUTA study suggests that a 74 percent increase in vehicle-km of service would have generated at least 50 percent more demand, or higher if there was already a backlog of unmet demand on systems due to overcrowding.  This scale of change is so far beyond any real-world experience that the calculated outcomes are meaningless. 

Moreover, there is a big difference between a 74 percent increase on an uncrowded hourly service and the same relative increase on a crowded 10 minute service.  Frequent services attract riders simply by existing, whereas infrequent services get only those customers who can afford to make an appointment for their transit trip and suffer the impact of off-schedule service and missed connections.

As an aside, I should note that the elasticities for auto travel with respect to transit pricing and service are either zero or very low.  A huge increase in transit service (or lower fare) is required to cause a miniscule change in auto travel using the factors in this study.  This reflects the general situation in Canada where auto users do not regard transit as a viable alternative except in markets where there is good service, and even then it has to be really good to overcome the benefits (real or perceived) of driving a car.

In this context, it is odd to see a conclusion that over two-thirds of the benefit of increased transit investment would be the value of reduced road congestion.  Note that I say “the value” because this is a calculated number derived from all of the assumptions in the study.  Whether there would be a real reduction in congestion, and where this would occur, is beyond the scope of CUTA’s report.

We already know from the Metrolinx White Papers that even a large investment in new transit will not have much effect on the modal split in 905-to-905 travel because these trips generally don’t fit well with a classic transit network.

Where does this leave the CUTA paper?  For me, it’s a curiosity, one of those reports that sits on my shelf until its value as an historic reference falls below the ever rising pressure for space. 

As a political document, however, it is very dangerous because it gives the impression that one can make national policy for transit spending based on a hodge-podge of assumptions and factors.  Hard costs change (for example the relative cost of owning and operating a car) and imputed social values (improved mobility, lower emissions) may be hard to translate into real budgets.

Transit policy in major urban areas will always be different from that in the rest of Canada because the market for transit is inherently strong in cities and particularly in large, dense cities.  The behaviour of auto and transit riders in those cities and on those transportation networks will be very different from that of their country cousins.  Even within large areas like the GTA, there is a huge variation in our ability to attract travel to transit at reasonable cost.

Transit needs more money, as do hospitals, schools, housing and roads.  Let us justify spending based on real needs, not on a formula whose derivation is suspect and open to challenge.

I salute CUTA for raising the issue of the national transit funding shortfall, but have little faith in the way CUTA made their case.

4 thoughts on “Is There an Optimal Supply and Demand for Transit?

  1. “For me, it’s a curiosity, one of those reports that sits on my shelf until its value as an historic reference falls below the ever rising pressure for space.”

    Good thing I only sent you my report as a soft copy, then, innit? 😉

    But seriously — without the benefit of the full report, all I’m seeing from the executive summary put forward by CUTA is a load of figures pulled out of thin air to attempt to justify pseudoscientifically what we already know empirically and want to see confirmed scrupulously.

    This presentation does an especial disservice to those of us amongst the Green South Park Republican nomyenklatura out here in The New Centre of the Universe who already know Pasha Weyrich’s arguments chapter and verse and who want to find some way, ANY way, to motivate Stefka to give a wet slap about mass transit as a strategic investment in urban mobility in this new century — his sort have to be rigorously convinced that the higher initial capital investment in mass transit is well repaid in lower operating, maintenance, fuel, and repair costs relative to spending on roads and automotive infrastructure.


  2. For a totally different (tongue-in-cheek) take on what constitutes a “higher investment in transit” in other parts of the (subway) world, see the Globe & Mail’s delightful “Underground tourism” article (Section T1 Saturday, September 20, 2008) and don’t forget to do the “Subway Know How” quiz.

    I scored 8/15 on the quiz and was rated “Average. You are not afraid to venture underground in a new city, but still prefer the bus.”

    I want to know if Steve will score 15/15? or somehow flub one answer and only get 14!~ 😛

    Steve: I got only 8 too. Contrary to popular opinion, I don’t know everything.


  3. One of the things that has concerned me recently is that many/most/perhaps-all the estimates in reports to date do not appear to have taken into consideration the change in usage patterns/volume caused in the past year or so by the price of gas and other economic factor changes. (I understand that most report would not have been able to, since such reports take time to generate.) It should be interesting to see who reports what (and when) about this issue and how it should be factored into budgets, etc. for the future.


  4. I got 11/15 on that quiz. Pretty good for a suburbanite and a perceived “car advocate who probably never took a subway”. Perhaps you guys should take some lessons from me, huh? (I’m only kidding)

    Your remark about the percentage of fares lost for every percentage of fare increase caught my attention. Consider the loss of free parking: assuming one person per car, that is a $6 surcharge and pretty much a 100% increase in fares. By those calculations, only 25% of those fares would be lost. Unfortunately, that is not my impression as a petition that came to my apartment regarding the reinstatement of free parking had mentioned in several polls that “upwards of 85%” of drivers (and their carpoolers) who rely on the parking lots would rather drive downtown instead of forking out the extra money. I would probably rely more on an upwards quadratic (parabolic) relationship rather than a linear relationship. For small fare increases, I can understand the results, but for large and very significant fare increases, it is much more serious and with a greater impact. This is despite my previous 7-8 persons in a minivan example.


Comments are closed.