Does the travel-time index really reflect performance?
By Eric Sundquist and Bill Holloway
Last week’s release of the Texas A&M Urban Mobility Report, with its charts and lists, prompted the usual flurry of general interest media coverage. This year’s report, however, carries more importance than usual, as it comes at a time when FHWA is considering new performance measures under MAP-21, and one of the measures under consideration is the travel-time index, a staple of the UMR.
A benefit of the travel-time index is that it is understandable and fairly straightforward to calculate; it is the ratio of travel time in congested conditions to the travel time in free-flow conditions. But as a gauge of system performance, one that could eventually affect funding or trigger penalties, it is problematic.
Readers of this newsletter are probably familiar with some of the issues with the TTI (travel-time index, not to be confused with the Texas Transportation Institute, the A&M center that publishes the Urban Mobility Report), many of which are summarized in a 2010 report from CEOs for Cities. A key conceptual issue is whether the TTI measures the right thing. For example, an arterial that is dense with destinations may flow slowly, but with short travel distances, provide excellent access. Still, the TTI would indicate a problem. The same issue applies at the system level. An area that has “bad” TTI ratings may be better for travel than one with good ratings, depending on local road connectivity, availability of modal choices, convenience of land uses, and other factors.
One way to test the validity of the TTI is to hold it up against another rating—commuting time. For 100 U.S. urban areas, we compared A&M’s TTI change between 2000 and 2010 with the change in mean commuting time for residents who worked outside the home during the same period. If the TTI were an important factor in determining the quality of access, areas where the TTI went up should see an increase in commuting time.
This was not the case. In fact, the relationship was slightly negative – places with increasing TTI actually saw shorter commuting times on average – though these findings were statistically insignificant. (See Figure 1).
The 2010 CEOs for Cities report contains a similar analysis for 1990-2000 (p. 34; using minutes of delay rather than TTI, but these are equivalent for present purposes). It found a positive relationship, but one that again was statistically insignificant.
What this means is that if TTI, untethered from the system context, becomes a standard for congestion or highway performance, agencies may be pressured to spend precious resources on costly new urban lane-miles, even though improvments in TTI have little to do with the way the system as a whole provides travelers with access to their destinations.
Commute time, now measured annually for urbanized areas as part of the American Community Survey, would shed more light on system performance and allow decision-makers more leeway in applying policy levers for improvement. If this is too much of a departure from conventional practice, new measures should at least be normalized by some variable related to the total system. One such measure might be (∑(AADTsegment (lengthsegment/speedsegment)))/number of commuterssystem.
FHWA is consulting with stakeholders now on performance measures. Readers who would like to weigh in may find process details and contact information on the agency website.
Note on data.
The SSTI analysis employs data for 100 U.S. urban areas.
Travel-time index data are from the “Congestion Data for Your City” webpage on the Texas Transportation Institute’s Urban Mobility Information website.
Commuting time data for 2000 are from decennial Census Summary File 3 tables P31 and P33. Commuting time data for 2010 are from the Census Bureau’s American Community Survey 1-year tables B08013 and B08303. In each case, mean travel time was calculated by dividing aggregate travel time to work (in minutes) by the number of workers over 16 years of age who did not work at home.
Commute times and TTI for the 100 cities analyzed can be found here.