« A Gift for the Reindeer Too | Main | Population Puzzler: Unwanted Pregnancies and Abortion Trends »

December 27, 2005

Driving Puzzler

Eric Pryne of the Seattle Times today provides a welcome but puzzling update on a project we’ve been watching for some time.

The gist:

The 400 volunteers in the Puget Sound Regional Council's "Traffic Choices" study have been paying virtual tolls since July. Devices mounted on their dashboards track where they travel and transmit the information to a central computer. Charges are deducted from prepaid "endowment accounts."

Those accounts are just play money. But if there's anything left in them when the experiment ends in February, participants get to keep it — in real dollars.

That's the carrot. They can save money by not driving as much, by choosing less-congested highways, or by staying off the road at rush hour.

As we’ve argued for years, paying-as-you-go for driving, rather than in occasional lump sums, ought to be a powerful incentive to economize on miles, trips, and fuel.

Here’s the puzzler:

[Study director Matthew] Kitchen says interim results indicate that, as a group, the study's 400 participants actually are driving a bit more than they did before the experiment started. But they're also paying a little less than they would have if the tolls had been in force when researchers first began monitoring their driving.

That could mean the volunteers are avoiding roads with the steepest tolls at times, even if it takes them out of their way, he says. "But we may find something very different when we do a more detailed analysis," he cautions.

What’s going on here? You put a by-the-mile price on driving and people do more of it? That’s unexpected!

Study results won’t be complete for months, of course, and these preliminary trends may be a seasonal fluke, a statistical error, or an effect of some unrelated factor such as a fuller employment.

But it could also be that drivers are much quicker to change roads than to change modes, or to reorganize their weekly travel plans. Stated that way, it’s less surprising. Past experience suggests that it takes some time for people to adjust where they go.

Still, by the time the study is done, I’ll have to rethink some assumptions if drivers haven’t reduced their total miles driven as well as their peak-hour trips on high-priced, congested roads.

Posted by Alan Durning | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Driving Puzzler:


From another point of view:

Gas taxes already make drivers "pay by the mile."

Therefore, it seems that by simply raising gas taxes then not only would drivers choose to use gas more prudently, the roads would also be funded. Plus, there wouldn't be any "spying" issues about using tracking devices.

Posted by: Michelle Parker | Dec 27, 2005 2:16:09 PM

I think the main thing is that it takes a lot longer than a couple of months for people to switch their driving habits. They really have to change jobs or move to a new house to see a marked change in their behavior.


This article from Slate is about gas prices, but the same concept applies. People are locked in to a certain amount of driving because of where they live, where they work, and where their kids go to school. Sure, there might be some movement around the edges, as someone decides not to take a rare "optional" trip somewhere.

But for the most part, it seems that most participants have not made any fundamental lifestyle changes, so they're just gaming the system by driving longer routes to save "money." In that sense, I bet the "Williamses" outnumber the "Tans" by a significant margin.

That's just a guess. Of course, it could simply be that all the urban planning in the world is no match for the fundamental irrationality of the American Commuter! :)

Posted by: Frank Bruno | Dec 28, 2005 9:58:43 AM

Good point, Frank.

And still, I wonder. Todd Litman at Victoria Transport Policy Institute has assembled an impressive body of literature -- much of it at www.vtpi.org -- to show that most drivers take a fair number of discretionary trips. Or take trips the length of which is somewhat discretionary.

And the PSRC charges quickly add up, more quickly than do fuel price increases.

The effect of fuel-price changes on distance driven is tempered by vehicle switching within households. That is, when prices rise, households take their more-efficient vehicles on more trips.

And fuel-price changes rarely amount to the same per-mile price signal as the PSRC pilot project. A $1.00/gallon increase in the fuel price works out to about 5 cents per mile for a typical, 20-mpg vehicle.

And 5 cents per mile is the bargain-basement bracket of price per mile in the PSRC pilot (except for between 10 pm and 6 am, which driving is free). It's the charge in the early morning and late evening on weekends on non-freeways.

Most driving is charged at higher rates: up to 50 cents per mile for rush hour freeway driving. That's like like paying an extra $10 per gallon of fuel.

So I understand why drivers are avoiding freeways. But I would have expected them to shed a few other trips, combine trips more carefully, or take shorter trips (to a local shop rather than a distant one, for example).

We'll have to wait for a while to learn more about this.

Posted by: Alan Durning | Dec 28, 2005 10:34:01 AM

Matthew Kitchen, who is overseeing this pilot project for the Puget Sound Regional Council, sent these notes:

"Interesting thread. There are lots of reasons why we (the study team) don't want to jump to any conclusions about these preliminary observations. It is certainly possible that the final results will be consistent with what we have observed so far, but it is unlikely. And here are a few reasons why:

1) At the end of the study we will begin analyzing individual driving behavior at the household level, by time of day, day of week, etc., and will be in a better position to say something meaningful about price elasticity of demand. Right now, what we observe lumps all data together without controlling for explanatory factors that may have been changing throughout the study so far.

2) Some non-trivial number of households have had travel changing events such as job changes, new residential locations, etc., all of this will need to be controlled for prior to serious analysis.

3) Our baseline data collection period was later in the year than originally planned, and overlapped with normal vacation periods (June). In part as a response to this issue we will be collecting control data at the end of the project as well. This will increase our ability to draw reasonable conclusions from the study data, but obviously means that we need to wait until the end of the study to produce "real" findings.

4) Some households were still getting equipped during the early weeks of the control period, thus skewing the aggregate measures when comparing control and experimental time frames.

None of the analysis we have done so far is intended to illustrate findings, only confirm that we are collecting a reasonable set of data upon which to perform the final analysis.

Having said all that it is of course possible that we will find that VMT may increase as people engage in diversionary behavior (remember that tolls are different on different facilities and some roads have no tolls at all). Even in this event we would expect that the implication is that vehicles are avoiding higher use facilities which could significantly reduce congestion, vehicle idling, drive time.

We'll have to wait a bit more before analyzing these issues.


Posted by: Alan Durning | Dec 28, 2005 11:28:43 AM

Good discussion.

There's one thing I see happening here, though:

Drivers are using just as much gas, if not more, to avoid idling on the freeway during rush hour, by taking less direct (i.e., non-tolled) roads.

Therefore, it seems that the main point of this computerized tracking system is to decrease traffic congestion rather than save gas.

I believe that raising the gas tax to the $10 per-gallon figure would both decrease traffic AND save gas. Additionally, it would greatly increase public incentive to invest in more efficient modes of transportation (i.e., fuel-efficient vehicles; mass rapid transit; etc.).

Matthew, I truly admire your desire to find solutions to traffic congestion. But I am wondering if the millions of dollars being spent on this project, as well as the huge amount of time analyzing the data and crunching numbers, might be better spent in developing mass rapid transit infrastructure alternatives, which could be paid for by significantly increasing the gas tax.

Raising the gas tax seems like a much easier and more direct solution here.

Posted by: Michelle Parker | Dec 28, 2005 12:28:39 PM

P.S. My above comment is said with the utmost respect for all parties involved (typed words can sound so harsh, sometimes!).

And I think it's absolutely wonderful that Matthew Kitchen sent NEW a note regarding this issue. I hope he continues to keep us informed. It sounds like he's learning interesting things about individual patterns of driving behavior.

Posted by: Michelle Parker | Dec 28, 2005 4:04:40 PM

First, I'm a huge fan of creating a more robust, efficient and effective public transportation system.

That said, does anyone know, more or less, what the optimum speed is for gas-efficient travel? (I think I have read that it is in the 30-40 mph range.)

If traffic congestion is relieved on the major freeways at rush hour and cars proceed at a more gas-efficient pace, would that reduce the aggregate total amount of gasoline consumed by all of the cars on the road? If so, getting some people to use a bit more gas by taking alternate routes might be a good strategy.

Posted by: Jed Mitchell | Jan 2, 2006 2:21:37 PM


Optimum speed depends a bit on the vehicle. Most of the numbers I've seen, though, lie in the 30-50 mph range.

Interesting hypothesis about a modicum of traffic diversion helping overall fuel-economy for all drivers.

One piece of counter-evidence, or at least, another way to think about it: One of the most interesting papers I've read on fuel economy was by the Australians Peter Newman and Jeffrey Kenworthy. It presented data to show that there's actually an inverse relationship between a city's per-capita gasoline consumption and its average traffic speed. Unfortunately, I don't have a copy of the paper, which was written in the 1980s, I believe.

The reasoning was this: vehicles operate more efficiently when traffic is flowing smoothly (though not zooming at full highway speeds). But smoothly flowing traffic tends to encourage more people to drive, which ultimately increases per-capita fuel consumption.

So traffic flow may improve fuel economy at the level of the vehicle but decrease it at the level of the city.

Another reason why it makes so much sense to concentrate less on vehicles, road design, or even transit than on building complete, compact communities.

Posted by: Alan Durning | Jan 3, 2006 9:57:43 AM

I think its important to keep speed limits in mind when we talk about fuel consumption. I mean, when I drive up to Snoqualmie Pass and the speed limit is 70, if its dry and not very congested, then I'm going 70-75ish. (My father is going to be mad at me after I write this!) This is much less fuel efficent, I would imagine, then going 50 miles an hour on a rural highway. I definitly think the regions transportation problems would be better solved by developing vehicles that are extremely energy efficent. This is because driving is an American pasttime. Personally, I love to drive. People are not going to want to give up their cars and sit next to some stranger on the bus they don't know unless forced to. At least that is my take on the situation.

Posted by: Gary Durning | Jan 5, 2006 12:31:11 PM