My recent Planetizen column, Smarter Congestion Solutions in 2014, discusses best practices for traffic congestion evaluation. It describes the methods recommended by international experts for measuring congestion costs and evaluating potential congestion reduction strategies. Since then, Thomas A. Rubin and Fatma Mansour released their report, Transit Utilization and Traffic Congestion: Is There a Connection?, which fails to reflect these practices. Let's look at what they do wrong.
My column pointed out that commonly-used indicators, such as the Texas Transportation Institute’s Travel Time Index, the INRIX Traffic Scorecard, and TomTom’s Traffic Index only measure congestion intensity, the degree that traffic declines during peak periods. Such indicators do not account for exposure, the amount that people must drive during peak periods and therefore their total congestion costs. Intensity indices are useful for short-term decisions, such as how best to cross town during rush hour, but are unsuited to strategic planning decisions that affect the quality of transport options or land use development patterns, and therefore the amount that people must drive. For planning purposes, the correct indicator is per capita congestion costs.
For example, a compact, transit-oriented city may have a 1.3 Travel Time Index (traffic speeds decline 30% during peak periods), 60% automobile commute mode share, and 6-mile average trip lengths, resulting in 34 average annual hours of delay per commuter; while a sprawled, automobile-dependent city has a 1.2 Travel Time Index, 90% automobile mode share, and 10-mile average trip lengths, resulting in a much higher 45 average annual hours of delay. Intensity indicators imply that the compact city has worse congestion due to greater peak period speed reductions, although its residents experience lower total congestion costs because they drive less during peak periods.
Rubin and Mansour base their entire analysis on the Travel Time Index, which is a fatal flaw. It would be accurate for them to say that their analysis indicates that congestion intensity is positively correlated with transit ridership, but it is inaccurate to say that it is positively correlated with congestion costs.
To be fair, in an appendix (page 73) they do acknowledge controversies concerning the TTI’s methodology (they even cite my report, Congestion Costing Critique: Critical Evaluation of the ‘Urban Mobility Report’, although they misspell my name), but they do not discuss the issues of debate or implications for their analysis, for example, how their results might change if the analysis were based on congestion costs instead of the Travel Time Index ratings.
Another fatal flaw is the study’s failure to account for confounding factors, particularly city size. Urban region size, development density, transit service, transit ridership and traffic congestion intensity all tend to increase together, so it is unsurprising to find positive correlations between any two. It is unfair to compare congestion intensity between 74 urban areas without accounting for such factors.
The report also fails to disaggregate by transit type. Public transit is intended to achieve two very different: basic mobility for non-drivers and efficient urban mobility. Some transit services are oriented primarily toward one or the other of these goals. For example, providing basic mobility requires operating buses at times and places with low demand, while efficient urban mobility requires high quality (frequent, fast and integrated) service on major corridors. To understand how public transit can reduce congestion the analysis should measure the impacts of high quality services, such as subways, light rail and bus rapid transit, on congested urban corridors, excluding basic bus service.
Public transit helps increase urban transportation system efficiency in other ways besides reducing congestion: it reduces parking costs, provides consumer savings and affordability, reduces chauffeuring burdens, reduces per capita crash costs, and provides a catalyst for more compact, multi-modal development, impacts that should be considered in analysis.
There is an extensive body of good, peer-reviewed research which indicates that high quality public transit can help reduce traffic congestion. Rubin and Mansour discuss political advocacy documents but provide no citations of the extensive technical literature on this subject, such as:
Md Aftabuzzaman, Graham Currie and Majid Sarvi (2010), “Evaluating the Congestion Relief Impacts of Public Transport in Monetary Terms,” Journal of Public Transportation, Vol. 13, No. 1, pp. 1-24. Also see, “Exploring The Underlying Dimensions Of Elements Affecting Traffic Congestion Relief Impact Of Transit,”Cities, Vol. 28, Is. 1, February 2011, Pages 36-44.
Michael L. Anderson (2013), Subways, Strikes, and Slowdowns: The Impacts of Public Transit on Traffic Congestion, Working Paper No. 18757, National Bureau of Economic Research.
Sutapa Bhattacharjee and Andrew R. Goetz (2012), “Impact Of Light Rail On Traffic Congestion In Denver,” Journal of Transport Geography, Vol. 22, May, pp. 262-270.
Changchoo Kim, Yong-Seuk Park and Sunhee Sang (2008), Spatial and Temporal Analysis of Urban Traffic Volume, 2008 ESRI International User Conference.
Todd Litman (2004), “Impacts of Rail Transit on the Performance of a Transportation System,” Transportation Research Record 1930, Transportation Research Board, pp. 23-29.
Todd Litman (2007), “Evaluating Rail Transit Benefits: A Comment,” Transport Policy, Vol. 14/1, January, pp. 94-97.
Shih-Che Lo and Randolph W. Hall (2006), “Effects of the Los Angeles Transit Strike On Highway Congestion,” Transportation Research A, Vol. 40, No. 10, December, pp. 903-917.
Wilbur Smith (2008), Traffic & Transportation Policies and Strategies in Urban Areas in India, Ministry of Urban Development.