What everybody gets wrong about induced demand

Induced demand is isn’t talked about enough. This one simple phenomenon explains not only traffic congestion but the history and development of cities ever since cities began.

But most people miss the explanatory power of induced demand because of the limited, popsci, midwit version of it they’re usually taught or told.

If it were understood fully, it would completely up-end the way we build transport infrastructure - for starters, we would never build a freeway again, and probably even tear down most existing ones.

So let’s look at what’s missing.

The usual induced demand explanation

A lot of people, especially politicians, believe that we need more roads to “reduce congestion”. Induced demand is typically raised as an objection to this - and correctly so.

This recent Sydney Morning Herald article on traffic jams explained it as follows:

Forever spending billions building more roads or improving existing ones can do surprisingly little to ease congestion in the long run – eventually traffic, like water finding its natural level, catches up. Think of Sydney’s Parramatta Road or Melbourne’s Hoddle Street: however many new lanes they add, both arterials are full to bursting come rush hour. Moscow’s infamous Garden Ring, around the Kremlin, was once up to 18 lanes; now a modest 10 at its widest, it is still often gridlocked. “Roads are invitations to drive on them,” notes Schreckenberg.

In 1962, US economist Anthony Downs called this effect “induced demand”. Downs was among the first of a new breed of scientist-engineer-economist types who have made it their business to study traffic and deduce what might be done to make it flow more freely.

“The opening of a new expressway reduces peak-hour congestion on many previously existing streets because large numbers of commuters shift onto the new expressway,” Downs observed in the journal Traffic Quarterly. “At first, they are able to make much better time on the expressway. However, word of [this] soon spreads, and even more commuters shift from other routes onto the expressway. Gradually, the time required for commuting on the expressway rises as peak-hour congestion increases.”

This is exactly what happens - to the extent that additional road capacity speeds up journeys, they are more attractive to drivers, more people drive (“I used to take the tram to work but now I drive because it’s faster”, “I used to avoid peak hour because of the traffic jams but now I don’t need to”), and the traffic congestion is soon back to square one (at best) or worse.

Possibly the most viral piece of transport planning content ever is this clip from Australian TV show Utopia, which demonstrates the futility of adding more car capacity:

It’s a great bit of TV - very funny, while deftly explaining a a real problem.

Limitations of the usual induced demand explanation

But there’s something unconvincing about this as a complete explanation of the impossibility of ending traffic congestion. If all that happens is a one-off increase after new road capacity opens, then building even more lanes would probably work. Particularly in cities where most people already mostly drive, there’s only so much more demand you can induce.

Certainly a new road will create mode shift for cars, and that’s bad, but from a perspective of a politician or planner who does want a car centric city, it seems like building more lanes will eventually work if you commit to it - as long as you build enough to overcome that initial spike.

More cars are getting moved about, and moved about faster. Sure, there might be more congestion, but there’s more throughput.

With only this part of the explanation, the case for More Roads is at least arguable. Again, this is the part of induced demand that everybody talks about - it’s the conventional midwit understanding of induced demand.

Short term vs long term induced demand

But this aspect of induced demand is secondary to the main effect, which is longer term. It involves not only roads, but real estate. It’s about where people move to.

When a road is built, people move to the other end of it.

Short term, the behavior change is “great, now I can drive down the new road and get to work faster”. Long term, the behavior change is “great, I can move to the other end of that road and drive back down that road to get where I need to go”.

The other end of the road is typically further out of the city where housing and land is cheaper. Not only does the new road induce congestion, but it induces sprawl, which induces longer term congestion.

Instead of the original point of congestion-busting, faster journeys, the behaviour change does the opposite. Longer journeys are caused in two ways:

• Spreading out - things generally being further from each other

• Mismatch - the destinations people move between being in different streets/suburbs/towns rather than the same one

More concretely, somebody might move from an inner city neighbourhood which they are finding expensive to a cheaper housing development on the outskirts of the city, relying on the new faster freeway into down to drive back into the city for their job, to shop at Costco, to their old neighbourhood to visit friends and family, whatever.

Induced demand from the ground up

Note that induced demand is not specific to cars, and is not even specific to transport.

It is simply the basic Econ 101 observation that when the cost of something is lower, people buy more of it.

In the context of transport, we mean cost in the holistic sense of not only financial cost but time, comfort, safety, etc.

This aspect of induced demand is not merely a theory on why freeways are bad, but the mechanism by which cities develop. Once train travel became faster and more convenient, London spread out in a starfish pattern along the various train lines.

Today, people still plan their house purchases around transport, perhaps moving next to a train station for convenience, ensuring they have enough parking for their cars, or choosing to live close enough to the city to bike to work.

More roads means more cars, more train lines means more trains, more bike lanes means more bikes. All that is obvious - although many get the causality backwards. It is often claimed, for example, that “Australians love their cars”, and that is why Australian cities prioritize cars.

In fact, the causality fundamentally runs the other way: Australians cities prioritize cars, which makes people want cars because they are the only feasible/convenient transport option. The demand for cars is… induced.

The futility of freeways

Considered from this longer term dimension, the complete uselessness of freeways and faster roads is clear. These roads don’t ease congestion or help deal with sprawl, they create it. They are worse than useless. If only our politicians could reckon with this simple fact.

Faster roads don’t mean faster journeys, they just mean longer journeys. That’s why all the long term data shows that commutes haven’t got faster in decades even as road speeds have gotten faster and faster.

Of course, long train lines move people further from the city too - but train lines have an enormous capacity that can cope with it, and produce denser clusters around the stations rather than widespread sprawl. This is why we don’t characterize outer-suburban transit-oriented neighbourhoods and towns (like in the Netherlands) as sprawl the way we do with Australian neighbourhoods.

Cars are uniquely spatially inefficient and throughput-inefficient. That is why in spite of the generality of induced demand as a principle, it’s uniquely a congestion-causing problem for cars.

Induced demand, car parking, and the YIMBY movement

It’s not just about the car journeys themselves, but parking at the other end. Here too, more available parking at the destination means more people driving. And the flipside, if we want to reduce car dependency and sprawl, a great way to do that is to remove car parks, or at least, not build new ones.

Understanding of this is slowly spreading through government, and as such there has been some progress recently on things like reducing car parking minimums in new developments.

But this doesn’t go far enough. It’s still very standard, particularly in higher end developments, for each new apartment development to have a large basement car park with at least 1 car park per apartment.

The increased density a new apartments brings with it should add walkability and bikeability to the neighbourhood and reduce congestion - but with parking garages as standard, the opposite occurs.

This reduces quality of life for existing resident, and fosters NIMBY sentiment, undermining the case for further housing being built in the area.

Many NIMBY objections are actually about traffic and parking, sometimes thinly disguised. “Too many people” actually means too many cars, “reduced amenity” actually means “too hard to find a car park on the street. And when these apartments do induce more car activity by building new basement car parking, these objections are valid.

If the NIMBY movement is to genuinely foster support for more density, rather than simply supporting short term rubber stamping of what developers want, it needs to support not only the elimination of car parking minimums, but car parking maximums. Developers like Melbourne’s Nightingale have shown that apartments without car parking can be cheaper and very popular.

(It should also support things like mandatory footpath upgrades and pedestrianization of roads next to developments, which ought to get better rather than worse when a new development happens).

Which demand shall we induce?

Nothing is set in stone in cities. We can induce whatever demand we so choose. Sydney ripped out most of its light rail and built freeways, inducing car demand, now it’s building back new light rail and new metro lines to induce demand the other way.

A new bike path across the Harbour Bridge will induce thousands of new bike trips that otherwise wouldn’t have happened.

Induced demand builds cities, and the demand we induce is up to us.