Do new cycling facilities generate new cycling demand?

And if so, how much of that demand is truly induced versus being drawn from other modes? Two new reports throw some light on the question...

A new cycle lane in Manchester

A new cycle lane in Manchester

I have a ‘thing’ about induced demand, generated traffic. I generally find that it is misunderstood, accidentally or purposefully, by both proponents and opponents of road expansion. 

It undeniably exists (the Jevons paradox - make something cheaper and the demand will increase) but it doesn’t appear from nowhere: much of that generated demand is a redistribution from other modes, other time periods, other destinations or from routes quite far away. And we can model this - these mechanisms are included in the 4-stage models that we tend to use for larger transport infrastructure projects.

There will only be induced demand if latent pent-up demand exists, for example in congested urban situations, where road capacity expansion could be expected to at least partially fill up with unmet demand that was making detours, travelling elsewhere or outside of the peak periods.

The same argument can be applied to cycling: if there is latent demand for cycling that is currently not met because of poor standards of cycling infrastructure, then we can expect this to be released as induced demand when segregated cycle lanes, or cycle superhighways are introduced. 

But in contrast with car travel, such generated cycling trips will generally only lead to increased benefits: health benefits for the new users, and decongestion benefits for those remaining in their cars and buses as some users have shifted to an active alternative.

So whereas induced demand is often presented as a reason not to invest in more roads (as the new capacity will only fill up), the opposite would be true for modes where generating additional demand would be considered a good thing. For example, induced cycling demand from investing in high-quality bike facilities. This is why two recent publications are interesting – and the first that I have come across that provide insights and analytical methods to calculate these effects:

1.     Bikeability and the induced demand for cycling (Mogens Fosgerau, Miros?awa ?ukawska, Mads Paulsen and Thomas Kjær Rasmussen (April 2023) https://www.pnas.org/doi/10.1073/pnas.2220515120

2.     Cycling Diversion Factors, Rapid Evidence Assessment, Summary Report (Transport for Quality of Life, University of Westminster, Sustrans and UWE Bristol (July, 2022)  https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1150061/cycling-diversion-factors-summary-report.pdf

The first Danish study estimates that 60% new cycling trips and 90% more cycling kilometres were induced by the implementation of cycling superhighway style measures in Copenhagen. How much of that demand is truly induced versus being drawn from other modes is unclear. 

When applied elsewhere, this would be obviously context-specific, depending on the level of latent demand as well as the attractiveness of the alternatives.

This is why we build mode choice models and also why the use of diversion curves as in the second publication is an attractive but ultimately quite simple rule of thumb. Accepting that caveat, the numbers in tables 1 and 2 of that report suggest that of 100 new cycling trips 11-28% will be attracted from cars, 45-58% from public transport, 19-26% from walking, but 17-28% would not have travelled before, which I consider truly new, induced trips. 

The good news is that these calculations can be used to make a first order estimate of induced demand from cycling infrastructure:

1.     If you don’t have a mode choice model available, your top end guess would be an increase of 60% in trips and 90% in kilometres if a network of high quality cycle superhighways as in Copenhagen would be introduced. The effects of individual schemes could be expected to be smaller, although that might be counterbalanced by sizeable long-distance rerouting and substantial pent-up demand.

2.     If you do have a mode choice model available, you can argue that, on top of the modal shift that is calculated from other modes, a further uplift of 17-28% in cycling flows is defensible as purely induced demand.

The bad news is that the second report on cycling diversion factors also implies that (in the absence of a mode choice model telling you otherwise) approximately half of any new cycling trips can be expected to be extracted from public transport, and only a quarter from the private car. 

Tom van Vuren is Regional Director for the UK and Europe at Veitch Lister Consulting and a Visiting Professor at the University of Leeds. He is also the Transport Planning Society’s Policy Director