Monday, October 10, 2005

Cole v. Cote Round 3

> No love for tunnels
> By Rick Cole

...
> > The moment the ribbon is cut on new mega-projects, sprawl and
> > "induced demand" start filling them up.

Mr. Cole needs to read the literature and history of this particular
urban myth in addition to rereading the Hansen study above.

Sure. It's easy to induce demand for a subsidized anything. Just tie
another porkchop to the ugly kid and you'll get more dogs to play with
her. Induced demand is a misunderstood term. The Sierra Klub has a new
rant on the subject;

http://www.sierraclub.org/sprawl/transportation/seven.asp

They are so desperate they also use the Hansen/Huang SoCal study.

Induced demand is not diverted traffic.

Induced demand is not natural growth.

Induced demand is not unmet transport capacity.

Induced demand is not time shifting nor increased flow.

Then a few years ago some UC researchers revisited Hansen:

[This is from the professional journal "Transportation" (ISSN
0049-4488), Volume 29, No. 2, Dated May, 2002 - published by Kluwer
Academic Publishers of the Netherlands - I read from the paper copy of
the journal, NOT off the Web.

The abstract (only) can be found at URL
http://www.kluweronline.com/issn/0049-4488/current - you have to pay
to read the whole paper.

The title of the paper is "Revisiting the notion of induced traffic
through a matched-pairs study."

Authors are Mokhtarian, Samaniego, Shumway and Willits]

The study used 18 pairs of roads in California as the basis
for analysis.

Here is the fatal paragraph of the paper near the end:

"There are clearly factors that have induced the California public
to drive more, on a daily basis, over the period we have studied.
These factors include population growth, demographic changes (such
as the entry of more women into the workforce), economic changes
(rising incomes, falling real costs of fuel), shifts to the
automobile mode from other modes of travel, and land use changes.
The impact of this inclination to drive more seems, however, to be
distributed quite evenly across improved and unimproved highways. Our
study finds no support for the claim that capacity expansion generates
traffic disproportionately on account of the expansion itself. The
increased traffic on California highways may have more to do with
factors like population growth and the life-style changes of
Californians than it does with whether or not a couple of lanes are
added to a particular highway."

Estimates of this effect vary. One study showed that, over
time, a 10 percent increase in road capacity led to a 9 percent
increase in travel, while other research finds that these changes
in demand may have a smaller effect.

Where the reference is:

Robert Cervero. Road Expansion, Urban Growth, and Induced
Travel: A Path Analysis, Journal of the American Planning
Association, Vol. 69, No. 2 (2003).

But actually the Cervero paper is a survey of the literature and you
guessed it, the 9% induced demand for a 10% increase in capacity is the
old Hansen and Huang paper that upon real analysis actually reveals
that increased roads capacity in the areas studied decreases demand.
Given that they looked at SoCal this is entirely believeable. By
reducing congestion trips can be chained and equivalent trips take less
time thereby allowing drivers to optimize route selection.

Let's look at the 3 largest regions of California (1996):

population Lane Miles DVMT
per 1000 per
persons person

Los Angeles 12.2 mill 2.1 21.6
San Franscisco/Oakland 3.9 mill 2.3 20.8
San Diego 2.6 mill 2.3 21.7

Ave of regions >500k pop 123.6 mill 3.3 21.4

See the problem? California urban areas have 50% fewer lane miles per
person than the AVERAGE urban area. Can you say latent demand? Notice
also that there is little difference in DVMT. Not as car crazy as most
think eh?

My take on the entire subject;

IIRC the range of non-consensus was from 3% to 10% of measured VMT on
new capacity could be called "induced" but that this number went lower
when it was realised that the predictions weren't good enough to
account for 3%-5% accuracy in the model. Calling the component that
came from unknown or modelling errors "induced" is not any more
accurate than calling it anything else. Subtracting out the 3-5%
unknown that frequently gets added on to "induced traffic" generation
takes that 3-10% number lower. How much lower? Who knows, that is
exactly the point.

"Induced traffic" isn't as important as many seem to believe; not
because they don't see more traffic but because they can't identify
which traffic is induced anymore than someone could predict fluid flow
by looking at a single molecule of water. Modelling or measurement
don't work that way, they breakdown when the phenomena being observed
approaches the error level of the measurement. .

1 comment:

SmellyPogoStick said...

First! Murst! Liverwurst!