Induced demand or induced traffic – modelling vs monitoring

Although I agree that the statement in the draft National Policy Statement for National Networks that “evidence that development on the network leads to induced demand is limited” is misleading, it is not necessarily wrong. Hear me out...

DALl-E generated image of “A traffic jam in the style of Renoir”

DALl-E generated image of “A traffic jam in the style of Renoir”

Any contribution on induced demand is worth reading – is there any other topic in transport planning and modelling that gets people as excited? OK, probably LTNs and 15 minute cities….

So I read with interest the article “Induced traffic: yet again a worryingly overlooked dimension in crucial road planning and appraisal policy” by Lisa Hopkinson and Phil Goodwin (LTT 871).

These topics will feature in our Modelling World 2023 programme when we meet in Birmingham on 2 and 3 November. Save a place now...

And although I agree, that the statement in the draft National Policy Statement for National Networks that “Evidence that development on the network leads to induced demand is limited” is misleading, it is not necessarily wrong. Hear me out.

I first misread the statement as that the evidence is that induced demand is limited. And I have some sympathy for that – it’s all about definitions. I first mentioned this problem in my contribution to the Modelling World 2022 showguide: “Modelling an emotive topic – dealing with induced demand”.

Assessing induced traffic by just monitoring traffic flows is not enough

Basically, we must separate out induced (or generated) TRAFFIC from induced DEMAND. The former is the extra (road) traffic that occurs and is enabled by road capacity expansion (or reductions in generalised costs for roads generally), composed of (long distance) rerouting, mode, destination and/or departure time shifts.

The latter, on the other hand, is composed of purely new demand for travel, not previously made using other modes, going to other destinations, or travelling at other time periods.

Models that follow TAG guidance should model generated TRAFFIC (TAG Unit M2.1: “ there should be a presumption that the effects of variable demand on scheme benefits will be estimated quantitatively unless there is a compelling reason for not doing so”).

The guidance is a bit more coy about modelling induced DEMAND: “It is … possible to assume that … certain trip purposes will change their frequency in response to changing travel costs, for instance trip frequency changes may be modelled for leisure trips but not for commuting trips”.

I have developed and used models that both have, and have not, included trip frequency models responding to changes in generalised costs, and then only for discretionary purposes.

Although it’s attractive to use a simple rule of thumb for induced traffic, like the one that Hopkinson and Goodwin mention (“An average road improvement, for which traffic growth due to all other factors is forecast correctly, will see an additional 10% of base traffic in the short term and 20% in the long term”), all studies considered in the literature show such a large variation in traffic generated that for me, and when the impact really matters, modelling is inevitable.

A second reason is that in appraisal we need to know where these extra car kilometres originate from: other modes, other destinations or other time periods, for example, as there will be both positive (decongestion elsewhere) and negative (reduced farebox) consequences. A study in Amsterdam found that mainly departure time changes had occurred, whereas a later study in Manchester (both described in my showguide article) identified destination choice as the main source of generated traffic.

That’s why assessing induced traffic by just monitoring traffic flows is not enough.

There is one important reference missing in the Hopkinson and Goodwin article, and that’s “Disappearing traffic? The story so far”, by Sally Cairns, Stephen Atkins and (again) Phil Goodwin, from 2002. I found an online copy. If road improvements generate traffic, then capacity reductions should lead to traffic evaporation – if induced traffic exists, so does disappearing traffic.

The authors find for 70 case studies that, following a road-space reduction measure, the average drop in traffic flow (in the study area or crossing a cordon) is around 22%, but the figure ranges between studies from a more than 100% reduction to increases (yes, increases) of 25%. It’s an article from 2002 – there should be many more case studies now.

Of course, even though variable demand models reflect most or all of the mechanisms that lead to generated or disappearing traffic, the modelled responses may not be correct sizewise, and that’s why monitoring and evaluation exercises such as those by Hopkinson and Goodwin are important, challenging model parameters, and strengthening the evidence base behind the modelling of both induced traffic AND induced demand.

Returning to the draft NNNPS. Although technically the point that evidence that road network development leads to induced DEMAND is limited is not wrong, it’s the induced TRAFFIC that matters – filling up new capacity, increasing miles travelled and continuing to commit people to a car-based lifestyle.

We understand induced traffic better than induced demand, and we can model it. This is what the text, and government road policy should really focus on.

Tom van Vuren is the Policy Director at the Transport Planning Society and a Visiting Professor at the Institute for Transport Studies at the University of Leeds. He recently joined Amey as Strategic Business Partner