Transportist: November 8, 2022: In defense of induced demand

Newsletters, now with footnotes.

This is a post for paid Transportist subscribers. It will eventually be liberated from it’s paywall, but not today.

Can’t afford The Transportist ? If you’re a student, or poor, or cheap,  or giving greatly to important causes and don’t have $0.14/day, drop me an email and we’ll work it out. Leave no transportist behind.

Some thoughts on Mastodon

I am now at Mastodon @transportist@mastodon.social . Given the direction things are going with the bird site, you may want to Join Mastodon [though admittedly servers have been slammed lately]. Follow me and I will follow back.

A brief note on Mastodon.

As all you cool kids move over — be patient — but do move over. With outages, Mastodon can sometimes feel like Mastodoon, a mysterious social network that only reappears periodically. There is a learning curve on top of servers being slammed and a Major software upgrade happening simultaneously. That it works at all is outstanding. Remember the fail whale back in the day on Twitter.

[Sidebar: How much would it cost ($/month) to have enough servers to handle the Twitter diaspora on Mastodon. Could Bezos’s AWS, or Brin’s Google do this as a public service to stick a metaphorical knife into Elon, Bezos and Brin each have reasons to hate the man?]

Mastodon and the Fediverse is the revenge of Web 1.0. The switch to a new social network platform is like Jubilee. All the old relationships are broken and new ones formed. We should do this every 7 years. I posit that in a newly viral social network, follow everyone who follows you. The most interesting people are early adopters. Most early adopters are interesting.

People for whom Twitter is a broadcast medium (People who Tweet but don’t read) will be slower to move off to Mastodon than those for whom it is two-way because they aren’t monitoring the feedback. It might be a useful fork, separating the celebrities from the cerebral.

Also, those of you who are academics should list your Orcid ID as one of your links in your profile. This will help the open web.

There are some great articles about this new great turn from centralised to decentralised social networks: Twitter is our Future by James Fallows, and Hope for a post-Musk net by Jeff Jarvis.

Transport News

• TuSimple Fires CEO Xiaodi Hou Amid Federal Probes

• Cruise Expands Service Area to Cover Almost all of San Francisco

• Trouble in Tram Paradise by the Antipodean Light?

Research by Others

• Introdução à acessibilidade urbana

In Defense of Induced Demand

The idea of “induced demand”¹ has become more or less mainstream in the transport community.

Briefly induced demand is just an increase in the quantity demanded of travel when the cost curve for travel reduces. Some of this induced demand may simply be a transfer from one location or time to another, and not produce a net increase in total travel (or may result in a net decrease of travel, e.g. if a short-cut route is added, reducing distances of existing travelers more than it induces additional travel distance from new travelers). Travellers may switch routes, times-of-day, modes, destinations, or choose to travel when previously they tele-commuted. New development may be built to take advantage of the new access.

My 1992 Master’s Thesis is essentially about one approach to model induced demand in the classic urban transport planning framework (`strategic planning model’ for my Aussie friends, `four-step model’ for the Americans). The idea was that there needed to be an equilibrium between the output travel time that results from the route assignment stage of the model and the input to trip distribution and mode choice. This was achieved in my thesis (published as Integrating Feedback into the Transportation Planning Model: Structure and Application) through a feedback of the output times from an iteration of the route assignment model into the input, recomputing trip distribution and getting a new trip table, and then, running the next iteration of traffic assignment, which updated the output travel times matrix. Repeat until convergence. I doubt this is the most efficient way to achieve the end, and I cannot prove mathematically in a complex modeling system that there is only one equilibrium,² but we did get an equilibrium. Which means that the model considered induced demand in a way that assuming a fixed travel demand and changing the network, and assigning it, and looking at the outputs, WITHOUT REVISING THE TRAVEL DEMAND, would not.

Now, if the change is small, the effects on demand will be small, and perhaps de minimis changes can be neglected in the model. But we should not be using this kind of regional planning model for de minimis changes³

But the term, and the idea, has been misinterpreted. First, there is a conflation of travel and traffic with congestion. So compared with the before case, adding capacity “induces demand”. That is, more people are traveling more distance by whatever mode added capacity, assuming that additional capacity actually reduces travel times. That is, if the mode had previously been congested, capacity additions at critical bottlenecks would reduce travel time. Capacity additions upstream of a bottleneck (or downstream of one) may have no effect on travel time at all. This time savings may induce traffic, but this doesn’t mean that adding capacity increases congestion, or makes travel time on the expanded segment (where capacity was added) worse.

Alternatively, the new capacity may reduce travel distance, when the new infrastructure serves a market more directly than before,

However when travel is induced on a particular link that is expanded, there are what economists might call`pecuniary externalities’ (though in this case `temporal externalities’ might be more apt). Pecuniary⁴ externalities are when the purchase of something changes prices in related markets. They are distinct from technical externalities.⁵ Travel is also induced on upstream and downstream links that may not have been expanded, as travellers move from an origin to a destination. That may increase congestion or crowding on those links, but the total travel time for the average traveler (between an origin and destination) should not increase because of the expansion, an increase in supply should not increase price. Though some travellers may be worse off though as pecuniary externalities change the travel demand pattern.