Forecasting as Voodoo

Post by Steven Dale

There’s nothing more common and consistently wrong in the transit planner’s toolbox as ridership forecasting and projections. It’s like voodoo: 90% of the time it doesn’t work, and the 10% of the time it does no one knows why (hint: it’s not because of the voodoo).

So here comes Tom Rubin, a veteran transit consultant saying if Los Angeles had forsaken its program to build streetcars and light rail and instead “run a lot of buses at low fares, they could have doubled the number of riders.”

Meanwhile, quoting the LA Times article above, Jarrett Walker echoes this philosophy stating that “if you really want a transformative boost in transit ridership, the single most effective thing you could do can be done entirely with paint and signs: converting traffic lanes or parking lanes to bus lanes.

It would be great to see Tom Rubin (and to a lesser extent Jarrett Walker) prove his claim. How can he know that Light Rail directly decreased ridership and that bus ridership would have doubled the number of riders? How can he make such a sweeping prediction?

He can’t.

There’s no way to make that claim unless Rubin has access to a time machine capable of visiting an alternate universe and reporting the results back to our current universe. And if Rubin did have such a machine, why is he wasting his time as a transit planning consultant?

If you read the LA Times article closely you notice four things:

  1. Rubin  makes clear that the initial decrease of transit ridership in 1985 was due to an increase in fares. It’s a bait-and-switch. First he attributes the decrease in ridership to an increase in fares. He then tries to pin that on Light Rail (because the subsidy used to artificially keep bus fares low was shifted to rail).
  2. Rubin notes that traffic congestion continues to rise throughout the region and uses that as evidence of rail’s ineffectiveness. It’s a correlation versus causation error: Just because rail was built at the same time that transit ridership decreased does not mean one can attribute the latter to the former. Meanwhile, during the same period of time, LA opened one of the longest and most heavily used Bus Rapid Transit lines in North America. Why is rail to blame and not BRT?
  3. Rubin conveniently ignores the fact that transit ridership has returned to pre-1985 levels in Los Angeles.
  4. Rubin focuses on running “a lot of buses at low fares.” His argument in favour of buses is dependent upon them having low fares. The same argument could be made for running “a lot of streetcars at low fares” or “a lot of ponies at low fares.” Rubin’s argument should be rephrased as low fares increase ridership not buses increase ridership.

Generally speaking, I’m not the biggest fan of LRT because it’s rarely implemented properly. Nevertheless, I wouldn’t go so far as to say that it was the cause of decreased transit usage in Los Angeles, especially when the logic underpinning such an argument is completely suspect.

I also wouldn’t go so far as Jarrett Walker does to say that any one technology or technique (bus in particular) is the single most effective means to boost transit ridership. That’s a pretty big claim to make especially without any statistics to back it up.

For any technology-specific advocate, the stakes are high. Transit contracts are some of the most valuable in the world, costing billions of dollars. It shouldn’t, therefore, surprise us that some industries play fast and loose with facts and truth. Is it right? No. But just because it isn’t right doesn’t mean we should blind ourselves into believing it doesn’t happen.

Cities, meanwhile, are continually struggling to increase transit ridership. So if a certain group of technology enthusiasts can make a specious claim that their technology can do that, maybe their technology will win more contracts and their consultants and planners will get more work. It’s a self-fulfilling prophesy that’s (strangely) rarely fulfilled.

Selling one transit technology as the be-all-and-end-all savior of transit is irresponsible. Damning another technology using incredibly faulty logic worse still.

Note to Tom Rubin: If you do have the aforementioned alternate-universe-time-machine handy, could you please tell me who has my copy of Jane Jacob’s Dark Age Ahead? I really love that book and I have no idea who I lent it to. Also: Whose going to win the 2014 World Cup? And: What would I look like with a mustache?

Want more? Purchase Cable Car Confidential: The Essential Guide to Cable Cars, Urban Gondolas & Cable Propelled Transit and start learning about the world's fastest growing transportation technologies.

Want more? Purchase Cable Car Confidential: The Essential Guide to Cable Cars, Urban Gondolas & Cable Propelled Transit and start learning about the world's fastest growing transportation technologies.


  1. Good points Steven, Also, in terms of PT funding (especially rail based) in North America, I am aware that forecasts are generally inaccurate/optimistic as cities compete with each other for federal and/or state funding. I may be incorrect here. But if thats the case, you see the incentive for local professionals / city authorities to 'cook the books'. When there is a finite pool of money, a competitive tendering process for it, a large benefits for the receivers - the conditions are rife for twisting information to make it look better. If you don't, as a city, your 'shooting yourself in the foot'. In New Zealand, there has been shifts away from quantitative analysis (especially case by case analysis) as it has proven to be so inaccurate when compared to the end results, both underestimated and overestimated. For these reasons, feasibility or CBA's are now starting to be looked at as a component of decision making only. Unfortunately, this justification has seen a wave of funding go into roading based infrastructure because of "wider external benefits". You can't win either way! Survival the fittest (i.e. best sales person) will always apply in the transport infrastructure world. Ryan
  2. Being a common reader of Jarret's blog, Human Transit, I think he meant more along the lines of buses, being cheap to operate, allow you to operate them at a higher frequency (still not LT1M) and although they are still very compromised when it comes to speed and reliability, you can marginally increase their reliability and speed very cheaply (paint and signs), and this can increase ridership... where effective is the ratio between the increase in ridership and amount of money invested. Since you're blog is about cable, you may argue that cable would be far more effective by that definition. It could be Jarret, being a consultant in current best practice on making transit work, is unfamiliar with this emerging technology.
  3. Jeff, I'm a frequent reader of Jarrett as well and respect his work deeply. I just don't happen to agree with his stance in that post. It isn't a case of me being for cable and him being for bus/BRT. I simply don't believe in making blanket statements such as those that were made by Tom Rubin and Jarret. To say that the best, cheapest, fastest way to increase ridership on transit is to merely create bus lanes is folly. Especially without considering the vast number of variables involved in such a scheme. Will those lanes be fully dedicated across the entirety of the route? If not, reliability will be severely compromised which (in theory) should reduce ridership. However, to create fully-dedicated lanes across the entirety of the route is incredibly expensive. The same holds for Light Rail. What increases in traffic will dedicated lanes cause? How long are the buses? What is the fare structure? What's the socio-political culture that will allow for something to be built? What's the level of municipal corruption in the city in question? Now throw in the myriad of variables that could arrive in the future that we have no capacity to predict: Economic crisis? War? The $1,000 car? Whether it be bus, light rail, subway or gondola, there are just too many variables involved to simply say "my technology is best and will generate the largest ridership." Ridership forecasting, meanwhile, has demonstrated itself to be an almost complete failure. I think it would be great if everyone in transit were simply to admit "you know what? Beyond next week, we really have no idea how many people are going to ride whatever we build."
  4. I agree it is frustrating that there is so much "my favorite technology is better than your favorite technology" debating going on, as opposed to people just seeing different transportation technologies as different tools in the toolbox. I also agree the current system for generating ridership projections is flawed, and maybe the very idea is ill-conceived. But one nit: one of the nice things about BRT is that you don't necessarily need to have dedicated lanes along the whole route to get most (if not quite all) of the possible reliability and speed gains available. For example, if you can ascertain specific parts of the route in question where there is congestion, a bottleneck, or so on, you can potentially target just those specific problems with dedicated lanes and get a lot of gains. The flip side of this concept, though, is that these may be precisely the same places where the most harm to mixed-use traffic flow will be caused by dedicating lanes to BRT. After all, something is causing the congestion/bottleneck/etc. in question, and dedicating a bus lane to be used as a bypass may actually make that situation worse for the remaining traffic. And as I see it, that is where something like gondolas may come in handy--they may be able to bypass these congested/bottleneck areas on surface routes without making the situation worse for the remaining traffic.
  5. Brian, I love the toolbox analogy and wish more people would look at it that way. Hammer's are great. But so are screwdrivers. And vice grips. But only if used properly and in the right context.