Building more roads

[jack]

I've recently read several posts like, http://www.politics.co.uk/blogs/2014/12/01/george-osborne-s-15-billion-road-to-nowhere, which say "if there's congestion, building more roads will cause more car travel and more congenstion and won't really help".

I was inclined to agree with the general conclusion, but I didn't actually understanding the reasoning.

Eventually I read the 1994 DFT report http://persona.uk.com/A21Ton/Core_dox/P/P14.pdf, which was a bit of a slog, but made it a bit more clear to me.

The report itself

One thing that struck me, is how rarely I read reports like that, and how competent it was: clear what it was saying, what it was based on, who endorsed it, and what they thought DFT should do. I realised how much government process goes on in reports like this that I know NOTHING about, and I wish evidence-based policy making was a lot lot lot more high profile!

More roads

The gist of the report is, DFT evaluation of the benefits of new road schemes usually rest on the assumption of, "assuming the same volume of traffic, how much benefit will there be to this extra road building". And extra traffic was assumed to be a small effect that was usually not significant. But the report says that's not true: demand is elastic, so better transport means extra traffic (both from short term effects like people choosing to drive, and long term effects like people building new companies in places with good transport links)

I think what I'd not understood was why that mattered. I'd unconsciously assumed that more cars had an inverse relationship to travel time, so more roads would translate into better travel -- either quicker journeys for existing travellers, or making possible journeys for people who previously couldn't travel or travelled a different way. But I couldn't see why it mattered which.

However, I think what the report is trying to say -- which if I'm right, I'm puzzled wasn't explained more clearly in a summary at the start -- is that my unconscious assumption was wrong. It's more like, up to a certain point, more cars don't really slow anyone down at all (a motorway can handle two cars travelling at 70 just as easily as one). But past the point where you start to get congestion, it slows down LOTS for EVERYONE. So congestion doesn't just mean "too many cars", it actually makes the travel time WORSE.

Even, maybe, something like "if everyone tries to travel at once, it takes four hours for everyone, but if half the travellers set off immediately and half in two hours, it would take two hours for everyone". I'm not sure if that's true, but it seems what is trying to be said. But I'm not at all sure I'm understanding this right -- is that actually right?

If so, "more roads means more cars means more congestion" specifically makes sense: roads reach a level of congestion people stop travelling on them at, and if you build more roads, you get more people, but everyone's journey still takes twice as long as it would on empty roads. (Until you reach the point of enough roads for everyone in the area who wants to commute at once, which is what new road building schemes imagine, but the report says is basically impossible.) And the report says, if DFT new-road evaluations allowed for that, a lot fewer of them would be evaluated as worth the cost.

Is that actually right?

Disclaimer

I would prefer it, both for myself and the country as a whole, if public transport were so convenient that everyone preferred to use it all the time, not relegated to second-class-status. But I want to actually understand the evidence for things, not just assume the evidence that gives the conclusion I hoped for is always right.

Comments

[forestofglory]

Written like someone who has not spent hours and hours sitting in rush hour traffic :). Around here one actively plans trip so as not to get stuck in traffic. So for example R sometimes goes to a game on the other side the the hills. If he leaves an hour early it will take him an hour to get there but if he leaves and hour and a half early the driving time will only be half an hour -- so he frequently goes early and shops or something before hand.

Anyways I'm haven't read the report but your summary matches what I know about roads and congestion. Another name for this is induced demand.

[seekingferret]

Yeah, I think that's right. I learned this concept from Robert Caro's The Power Broker, about how Robert Moses reshaped New York's roads as public works commissioner... The conclusion from Caro's analysis is that more roads did mean more and better access to more places, but it did not solve the problem of congestion, for all the reasons you state. The solution Caro pushes for, generally, is for roads and public transportation to be given coequal status in urban planning, rather than for public transport to replace roads altogether. For example, he is sharply critical of Moses's failure to construct a rail line to Kennedy Airport on the median of the Van Wyck Expressway when it was being built- the later efforts to bring rail to Kennedy were much more expensive than they would have been to do it all at the same time.

[jack]

Written like someone who has not spent hours and hours sitting in rush hour traffic :)

Well, sort of. I'm really fortunate that living and working in Cambridge, I've never had to regularly experience rush hour traffic. But when I was going to visit Liv in Stoke, driving up along the A14 and M6 straight after work, it was pretty painful, because it was hard to predict the traffic, but also hard to say "I'll get there some time between 7pm and 11pm" :(

Like, there's a difference between "a shortage" where there's just not enough to go round and "tragedy of the commons" where more people make things worse exponentially, not just by dividing the same benefit amongst more people. In both cases, more people make things worse, but the first is "how to share it out fairly" and the second is "if you share it out, the total amount actually gets better, but if you don't, it actually gets worse". It seems this was obvious to everyone else, but simply sitting in it told me "it was bad" but didn't teach me that the delays were not proportional to the number of cars, but exponential in the number of cars...

your summary matches what I know about roads and congestion. Another name for this is induced demand.

Right, that was the term the report was using, though I didn't previously know it. It also makes we wonder if there's any way of tackling congestion specifically, not just reducing car use in general (although both would be desirable). Which is something people who already know about urban planning know something about, but I don't :)

[forestofglory]

Well one thing some places are doing is peak pricing. This works on tole roads. (Though I guess British rail also uses this concept.) The idea is that you make in more expensive to travel at peak times (communing hours in most places)so some people chose to travel at other times. London as some type of congestion based pricing, but I'm not remembering all the details.

In the US it is quite common to have a carpool lane, were only cars with at least 2 or 3 people can drive. This does encourage carpooling which reduces congestion, but there is most likely a lot more the local governments or even large companies could do to improve carpooling.

And of course getting more people to live closer to were they work and shop reduces the amount of traveling for those people do which helps reduce congestion.

[highlyeccentric]

I have heard of city councils offering incentives to businesses to offer incentives to *employees* to either start work before 8 or finish after 6 (with corresponding adjustments at the end of the day) in order to reduce peak traffic and put a bit of pressure on flexible working hours at the same time (eg in theory a working couple where one parter works 'early' shift and one 'late' can sort out school dropoff and pickup more effectively than two 9-5 office workers).

I know Sydney public transport does off-peak returns, but I think at one point there was a long-term pass which was heavily discounted if you promised to travel early or late. Canberra had - but scrapped, more fool them - *free bus travel* for anyone with a bicycle, in order to encourage people to cycle+ride rather than drive across town.

[jack]

Yeah, I've heard all of those things suggested, but not very seriously expected. I think what's difficult to me is that economically it would make sense to charge more for using roads, until the congestion went away, because that would be better in aggregate. But it would be massively unpopular with everyone for whom it was inconvenient (which I can live with, and it's getting better as toll-mechanisms get more sophisticated), and also, it would disproportionately hurt people who can less easily afford to pay :(

AIUI, the London congestion charge was controversial when it was introduced, but only covers an area about three miles across in the centre of London where I can't easily imagine ever really wanting to drive anyway. I think it mostly covers business hours, but I'm not sure.

[redbird]

I think another piece of it is that more roads mean people are less likely to do things like carpool. They'll make separate trips to see a friend, check out a restaurant in the town where the friend lives, and go to a jeweler or a second-hand bookshop that's sort of on the way; fewer roads mean someone is more likely to combine a couple of those things, or buy things closer to home.

Some of the "adding roads doesn't reduce congestion for very long" comes from a natural experiment: the last major SF Bay Area earthquake (the World Series quake) took out pieces of the local road network. They had a good idea of previous usage of those roads, and a lot of those trips weren't displaced to the surviving bridge and other roads, a nontrivial number of them just vanished. (I am working from memory here, and don't know how many of them came back when the Bay Bridge was fixed.)

Possibly related, once I was able to buy a monthly transit pass rather than paying per trip, I became more likely to stop off partway home (or even make a side trip that took me slightly out of the way), because the extra stop might still cost me a few minutes, but it didn't also cost money. It's less appealing to hop off the train for a loaf of bread when you're paying $2.40 to the bakery and another $2.00 for the extra subway trip; a cheap lunch in Chinatown isn't cheap if I have to pay for the round trip on the subway.

[This userpic shows the Bay Bridge partway through reconstruction; mostly I use it for posts/comments where it is more metaphorical.]

[gerald_duck]

One elephant in the room is that demand is also a function of population (though not linear in population: the average journey gets shorter as population density rises). If we had fewer people in the UK, the roads would be less congested.

People understand that in relation to other resources, at least enough to hold some quite unpleasant views about immigration.

We might be able to hold Malthus at bay indefinitely. But why try? I can't help feeling that some policies encouraging people to have fewer children, later in life wouldn't go amiss, perhaps even incentives to emigrate. Possibly also discouraging immigration, once we've taken those other steps so that doing so needn't be xenophobic and hypocritical.


Here's another complicating factor, though: if we take a barrel of oil, put it in a car's fuel tank and use it to limp slowly along a congested and circuitous road, the carbon in that oil is released into the atmosphere as CO₂ emissions. If, on the other hand, we turn it into a tarmac road, that's zero-emission.

[jack]

I can't help feeling that some policies encouraging people to have fewer children, later in life wouldn't go amiss

Hasn't this already happened? I thought in most developed countries, numbers of children were starting to fall below replacement, apart from immigration. Is that wrong?

There's still lots of problems with that: it means we have to survive when an increasing proportion of the population are retired, and it won't be solved for the world as a whole until every country is as developed as Japan... But the population may naturally peak (even if at a high level)...