Learning from TTI

Michael Lewyn's picture

In a recent post, Todd Litman criticized the Texas Transportation Institute's Urban Mobility Report.  In this post, I'd like to do something a little different: assume that TTI's congestion estimates are more or less reliable, and try to learn something from them.  So here are a few observations:

1.  Congestion isn't (consistently) rising anymore.  Out of the nation's 15 largest urban areas, five (Seattle, San Francisco, Atlanta, Los Angeles and Detroit) benefitted from decreased traffic congestion (measured by hours lost per driver) between 1999 and 2009.  On the other hand, congestion continued to rise in some places- for example, by 38% in Houston (from 42 to 58).

2.  Highway-building didn't seem to help.  Two of the five congestion-reducers built enough highway lane-miles to increase lane-mileage per person (San Francisco and Detroit).  But three others did not. In Altanta, the number of lane-miles increased by 7 percent from 1999 to 2009 (from 2350 to 2520).  But population increased from 3.7 million to 4.2 million (a 13 percent increase).  So road-mileage per person actually decreased, yet congestion decreased from 49 miles per driver per year to 44. In Seattle (where congestion hours went from 52 per motorist in 1999 to 44 in 2009), a 20 percent regional population increase was not matched by road growth: freeway mileage increased from 1600 to 1855 miles, only a 9 percent increase.  In Los Angeles, freeway mileage and population were almost evenly matched.  Regional population increased from 12.3 million to 13 million (about a 6 percent increase) while freeway miles increased from 5400 to 5610 (a 4 percent increase).  Yet congestion nosedived from 76 hours per person/year to 63. 

What happened in regions that failed to beat congestion?  In four of our nation's fifteen largest regions (Chicago, Dallas, Philadelphia, and Houston) hours lost to congestion increased by over 20% between 1999 and 2009.  In three of those, freeway mileage increased faster than population.

3.  Density was of limited importance.  This is a difficult issue because there are so many ways of measuring density: central city density, urbanized area density, and even attempts to combine the two. But it seems to me the most rational way of dividing large U.S. metro areas is between "transit metropolises" (places where at least in the central city more than 20% of commuters ride transit) and "car cities" (in varying degrees, everyplace else).  There are six large "transit metropolises" in the United States: Washington, Boston, Chicago, Philadelphia, New York, and San Francisco.   Did they have more congestion than other megaregions like Dallas and Houston?  Sometimes.  On the one hand, Washington and Chicago were nos. 1 and 2 in congestion.  On the other hand, the other four metropolises were somewhere in the middle, ranging from No. 5 among the 15 largest urban areas (New York) to No. 11 (Philadelphia).    Thus, the relationship between urban form and congestion does not appear to be enormous.

Michael Lewyn is an assistant professor at Touro Law Center in Long Island.



Proximity and capacity

This is frustratingly complex stuff.
Of course a valid criticism has been made of these figures by some - these figures involve the difference in travel time between peak and off-peak. Some cities may actually perform better on actual travel time measures than other cities, yet have worse "time lost to congestion" scores.
And what you are doing above, Charles, looks at "change" in a city over time. This is all interesting. Some cities must have started from a lower (or higher) "base" of road lane miles per person.
I think the most important factor for transport, must be "proximity" between homes, jobs, and other common destinations. Urban form, density, transport modes, road lane miles; are all not as significant as proximity.
There is no guarantee that mixed uses of land per se, will result in improvements in actual proximity between real life households and their common trip destinations. I suggest that lower overall urban land costs will correlate with better "proximity", because there is less "pricing out" effect.
I think it is almost certain, though, that stronger monocentricity of urban form and inflated urban land prices, if present simultaneously, will correlate with the worst possible outcomes. But inflated urban land prices themselves cause "decentralization" unless the region's planners have the power to reduce it, and exercise this power. Some cities planners are in the process of killing their cities economically today, by doing exactly this.
Cities evolve, and their types of employment change, and the location of different industries evolves over time. The number of workers per household changes over time, affecting how "optimum" household location decisions can be.
I believe that as cities decentralize, there are "low hanging fruit" gains that can be made in reducing congestion, where roads with only one or two lanes in each direction become congested for the first time. The reductions in congestion that are possible through capacity expansions in these situations, are high. The arguments that apply to huge radial freeways with many lanes, simply do not apply here.
The decentralization of employment in cities, is a natural efficiency-gaining mechanism, and planners should learn to expedite this process. Besides increasing the capacity of "suburban" roads experiencing congestion, flow improvements like interchanges, grade separation, bridges and tunnels should be provided. Ringroads are a good idea - the criticism that has been made of new ringroads that immediately fill up with traffic, misses the point that these have been seriously underprovided, while all the attention was focused on radial freeways. We have never really tried to "build our way out of congestion" other than with radial freeways, and I believe that addressing the "dispersed congestion" that accompanies decentralization, is far more likely to succeed. It is surely far more difficult to "induce" traffic when "doubling" capacity of "suburban" roads all over the region.
In "Planning for Cars in Cities: Planners, Engineers, and Freeways in the 20th Century" By Jeffrey R. Brown, Eric A. Morris, and Brian D. Taylor; we learn that many cities back in the 1950's had plans to increase "suburban" road capacity, but were over-ruled by the Federal focus on freeways. I am extremely interested that "decentralization" of employment has occurred anyway - the planners of those days were right and should have been listened to.
We have a serious problem with mass transit under these conditions, of course. Transit-focussed planning is the ultimate example of the tail wagging the dog. Improved "proximity" is the ultimate decider of the efficiency of the urban transport economy, yet the "mixing" and decentralization that this implies, is inimical to mass transit. We simply need to move on to new solutions. Mass transit really is "19th Century".

Todd Litman's picture

Need to Clearly Define Congestion Cost Units

Thanks Michael. I appreciate your contribution to this discussion.

I believe that it is important to be very clear how congestion costs are evaluated. All the statements you make above appear to reflect congestion costs per peak-period motorist, not per peak-period traveler, commuter or capita. I believe that 'per capita' is the best unit for most evaluations.

For example, in the last paragraph you write that Washington and Chicago ranked 1 and 2 in congestion among U.S. cities, but that only reflects delay per peak-period motorist. It ignores the fact that a smaller portion of Washington DC and Chicago commuters drive compared with Dallas and Houston, and therefore ignores much of the congestion-reduction benefits provided by high quality public transit systems that attract travelers who would otherwise drive. In fact, according to the 2007 Urban Mobility Study (the most recent I analyzed), Washington DC and Chicago residents had fewer annual hours of congestion delay per capita (30 and 25 respectively) than Dallas and Houston (34 and 33), as discussed in my report "Rail Transit In America" (www.vtpi.org/railben.pdf ). I'm not suggestion that 'per capita' is the only metric that should ever be used, but I think it is important to specify the units and to let people who use the analysis know how the results can be affected by the units selected.

Todd Alexander Litman
Victoria Transport Policy Institute
"Efficiency - Equity - Clarity"

Is there an areawide traffic congestion measure?

Traffic congestion is "spotty" in time and space. In the city where I live, many roads are uncongested all day while some major roads are regularly congested in certain directions at specific times. Generally, these roads are congested because of the lack of competitive alternative routes. Often, failing intersections could be eliminated with modest capacity improvements.

Using a single number to represent city-wide congestion may be thought-provoking, but I don't see how it is useful for local decisionmaking. It is akin to using body temperature as a measure of the well-being of a human being and expecting a surgeon to know exactly what to do based on that one number.

Bill Barker, AICP

on measuring failure

Right, let's try to measure why US ground transportation systems are a colossal flop. First you have to get in the nearest time machine and fast backward to April 1957 when the US congress vetoed proposed legislation to include mass transit provisions within the average 100m ROW's for the emerging 48,000 mile interstate highway system. Thank the powerful trucking/highway lobby for that one.

While still in your time warp device you should also visit the US congress a few years later in the 60's who consistently failed to provide R&D funding to Amtrack for a national mass transit grid.

Had either of these actions been positive we wouldn't have to find jobs for "planners" in the 21st century to grind out endless "studies" regarding why no can get anywhere anymore.

Only in America, where folks still believe that enshrined ,unrestricted free market capitalism and intelligent growth planning can actually co-exist.

Wow, what a fcked up system.

Michael Lewyn's picture

Here's a more sophisticated analysis

Prepare for the AICP* Exam

Join the thousands of students who have utilized the Planetizen AICP* Exam Preparation Class to prepare for the American Planning Association's AICP* exam.
Starting at $245

Essential Readings in Urban Planning

Planning on taking the AICP* Exam? Register for Planetizen's AICP * Exam Preparation Course to save $25.

Wear your city with style!

100% silk scarves feature detailed city maps. Choose from five cities with red or blue trim.
Wood necklace with city map

City Necklaces

These sweet pendants are engraved on a cedar charm with a mini map of selected cities. The perfect gift for friends and family or yourself!