Traffic congestion is a frustrating daily experience for many of us. It results in wasted time, fuel, and money for residents and businesses. Many of us have also seen shocking headlines that estimate the costs of congestion to our region, recently pegged at $8.2 billion in 2010. That total corresponds to an average cost of $1,568 per peak period automobile commuter, the highest of any American metropolitan area.
While we may "know it when we see it," congestion remains a difficult phenomenon to measure, as are its associated economic or social impacts. Many studies attempt to quantify the magnitude and cost of congestion at the metropolitan scale and often rank cities from most to least congested. Although they study the same topic, these reports produce differing results, and varying interpretations of these results tell different stories. Further, mainstream media coverage tends to focus on the city rankings rather than the root causes and effects of congestion.
To help clarify these issues, this Policy Update explores the methodologies commonly used in congestion studies. A second Policy Update will describe common strategies to mitigate traffic congestion and explore the nature of traffic congestion more broadly.
Survey of Congestion Studies
Perhaps the best-known ongoing congestion study is the Urban Mobility Report (UMR), published annually by the Texas Transportation Institute. The report analyzes 439 metropolitan areas in the U.S. and provides a number of congestion metrics from 1982 to the present. It provides statistics at the national scale for groups of metropolitan areas (e.g., large metropolitan areas with more than three million residents) and for individual metropolitan areas. The UMR methodologies have been revised over the years, and today the report also provides estimates of the congestion benefits of highway operational improvements and public transportation. A detailed description of its current methodology is available on-line.
One of the UMR's chief metrics is the Travel Time Index, a ratio of the time required to complete a trip under congested conditions to the time required to complete the same trip under free-flow conditions. For example, a Travel Time Index of 1.25 indicates that congestion would increase travel times 25 percent compared to free-flow travel; a 20-minute trip would require an extra five minutes.
The Travel Time Index offers many advantages and is a common congestion metric across the transportation industry. As an index, it is comparable across scales, locations, and time periods; it can be applied to specific road segments or, as in the case of the UMR, to entire metropolitan areas. Additionally, the Travel Time Index does not presuppose a policy objective. Other performance metrics such as travel speeds may reflect local preferences (e.g., setting lower design speeds) rather than the presence of congestion.
Although the Travel Time Index is not the sole measure of congestion used in the UMR and related studies, its basic tenet – a comparison of free-flow travel speeds to congested travel speeds – underlies almost all common congestion metrics. For example, the UMR estimates total hours of delay and total gallons of wasted fuel in the same way and derives its signature cost estimates by applying assumed values of time and fuel costs to those wasted hours and gallons.
The 2011 UMR incorporates travel speed data from a private company, INRIX. This partnership is relatively new and represents an improvement compared to past UMRs, which relied on modeled speed data and daily traffic volumes. In contrast, INRIX aggregates real-time traffic data from millions of GPS-enabled vehicles and mobile devices, traditional traffic sensors, and other sources. Its data is also available at a fine level of detail.
INRIX has produced a Traffic Scorecard since 2007 featuring its own INRIX Index, a metric that is essentially the same as the Travel Time Index. Each point on the INRIX Index corresponds to a percentage point increase in travel times above free flow (i.e., an INRIX Index value of 30 corresponds to a Travel Time Index of 1.30). Similar to the UMR, the INRIX Traffic Scorecard provides metropolitan-scale statistics. Unlike the UMR, the INRIX Scorecard also identifies the most congested highway corridors in the country. Given the localized nature of congestion, this analysis of specific highway corridors is particularly compelling.
The IBM Commuter Pain Index is a composite of ten variables, both objective (e.g., commute times, time stuck in traffic) and subjective (e.g., belief that the price of gas is too high, agreement that driving causes stress). This composite is based on a survey of over 8,000 respondents located in 20 cities across six continents. Unfortunately, more detailed information on the study's methodology is not included in its 2011 report.
The following table summarizes each study's parameters and how the Chicago region performed in each study's most-recent report.
Click image for larger version.
Critiques of Congestion Metrics
Congestion is difficult to measure and compare across metropolitan areas – it can vary substantially within a metropolitan area and also varies widely over time for any given location. A single metric cannot fully capture the variability and nuance of urban traffic congestion, and comparisons across metropolitan areas must be taken with a grain of salt. Indeed, the UMR provides a suite of performance measures and discourages its readers from focusing on a single measurement. However, some congestion studies provide only one or a few metrics, and media coverage of more detailed studies such as the UMR tends to focus on only one or two key measurements from any study.
Further, these congestion studies tend to rank metropolitan areas, and those rankings are often prominently featured in the popular press. For some metrics, there is very little difference between cities' rankings: Several are clustered at or around the same data point, and a small shift in the value of the metric can correspond to large shifts in the city rankings. In fact, the UMR encourages its readers to focus on the cities' actual values rather than their rankings.
A more fundamental critique of regional congestion metrics is that they cannot reflect the experience of individual travelers. As described in Rethinking Traffic Congestion by Dr. Brian D. Taylor, professor of urban planning and director of the Institute of Transportation Studies at UCLA, commonly used congestion measures tend to focus on the conditions of the transportation network, rather than the experience of individual travelers. In fact, the per capita measurements simply divide the total hours of delay, wasted fuel, or costs by the number of peak-period automobile commuters. In reality, some travelers may experience crushing congestion, while others experience little or no congestion. In addition, some travelers have alternatives to congested highway travel (changing their mode, time, or destination of travel), while other travelers have no such alternatives.
A 2010 CEOs for Cities critique of the UMR also stresses this point. The report notes that the Travel Time Index fails to account for the distances traveled during peak periods, which would better reflect drivers' actual experience of traffic congestion. A driver in a compact city may experience high congestion, but only for a short distance and thus a short period of time. In contrast, a driver in a lower-density city may experience moderate congestion but over greater distances and thus longer periods of time. Traffic congestion is arguably less onerous for the average driver in a compact city, but the Travel Time Index would suggest the opposite.
Another fundamental critique about the various statistics produced by these congestion reports – Travel Time Index, total delay, costs of congestion, etc. – is that they are predicated on a comparison of congested to free-flow travel times. While this comparison may seem reasonable at first glance, several writers have criticized this "zero-delay" baseline. In his 2004 book Still Stuck in Traffic, Anthony Downs of the Brookings Institution argues that the free-flow baseline is unrealistic: The capacity required to allow all peak-period auto commuters to travel at free-flow speeds would be prohibitively expensive, consume vast quantities of land, and impose significant environmental and social costs. As such, metrics that compare actual travel speeds to an impossible free-flow speed tend to overstate the magnitude and costs of congestion.
It is important that transportation system metrics be characterized not only in terms of an index of delay, but also by other measures as mode choice, travel measures, travel time reliability, and detailed evaluations at the corridor and even at the intersection level. This is the approach taken at CMAP, where we calculate many measures to understand transportation system conditions and trends (visit CMAP's Congestion Management Process or MetroPulse for more information) in addition to travel time indices.
Costs of Congestion
To take a step back, we can conceptualize the importance of congestion across three dimensions: its impacts on the average traveler, on the entire transportation system, and on society at large. The performance measures used in the UMR and other studies address the first two issues. They do not, however, effectively address the broader economic and social costs of congestion. The UMR's cost estimates simply look at the value of wasted time and fuel and apply national assumptions of the value of time for commuters and trucks to monetize these estimates. This approach ignores substantial variation in the value of time across individuals, industries, metropolitan areas, days of the week, and times of year. Further, some critics have argued that the assumed values of time are too high, which overstates the costs of congestion. The UMR does not produce more sophisticated estimates of congestion's effects on productivity, such as delayed deliveries and shipments, higher inventory costs, and missed transfers. While it does include estimates of the total value of commodities shipped by truck in an urban area, it does not explore how those commodity values are affected by congestion.
This focus on the economic impacts overlooks the real social impacts of congestion – the costs of stress, aggravation, and foregone social and economic activities due to traffic congestion. While the IBM study attempts to measure these issues, its index does not disaggregate the various metrics. Its geographic scope is also limited, and the report does not present time series data. Finally, these studies do not quantify the environmental costs of congestion, a substantial effort in its own right.
Conclusion
Congestion has a profound impact, particularly so for our metropolitan areas. Congestion affects individual travelers, increasing their congestion and time costs and also diminishing overall quality of life. It affects the regional transportation system, reducing its efficiency. And it impacts society at large, creating economic inefficiency and imposing costs on the environment.
While the popular congestion studies have methodological limitations, they do illustrate two important points. First, the costs of congestion are substantial, likely reaching into the billions of dollars each year. Second, the costs of congestion are felt most heavily in large metropolitan areas, as reflected in the fact that lists of most-congested cities are generally filled by the likes of New York, Los Angeles, and Chicago. While these rankings should be taken with a grain of salt, the studies do suggest that appropriate policies to address traffic congestion should be adopted at the metropolitan scale and be consistent with urban contexts. Indeed, the UMR estimates that congestion costs would have increased by $2.39 billion in the Chicago area in 2010 if operational improvement and public transportation service were discontinued. The next post in this two-part series of Policy Updates will review potential congestion mitigation strategies in more detail.