by Jeffery Memmott, Ph.D. and Peg Young, Ph.D.
Most urban areas of the country face rising congestion levels as increasing volumes of vehicular traffic exceed the capacity of the transportation system. The Texas Transportation Institute (TTI) reported that, in 2005, “Traffic congestion continues to worsen in American cities of all sizes, creating a $78 billion annual drain on the U.S. economy in the form of 4.2 billion lost hours and 2.9 billion gallons of wasted fuel.”1
Congestion Costs in Perspective
The 2.9 billion gallons of petroleum wasted in 2005 would have fueled U.S. daily transportation needs for nearly a week that year—6.1 days. If spread evenly across the entire U.S. population of 295.5 million in 2005, the 4.2 billion hours lost to congestion that year would have cost every man, woman, and child in the country about 14.2 hours.
Source: FHWA, Highway Statistics, 2005
At the national level, reducing congestion is one of the strategic goals of the U.S. Department of Transportation.2 But to reduce congestion nationwide, it is often at the local level where congestion must be addressed.3 Congestion patterns can vary depending on location and time of the year.
Drivers may notice seasonal changes in patterns of highway congestion in many urban areas of the country. These patterns, however, can be very different for individual cities. This report looks at congestion patterns over a 3-year period for three U.S. cities—Chicago, Los Angeles, and Houston (selected to illustrate geographic diversity)—by estimating the changes in monthly congestion during the year, the differences occurring in morning and evening congestion, and differences in weekend and weekday congestion.
Findings from the analysis of the 3 years of data are:
These data have the potential to provide a valuable source of detailed congested seasonal factors not previously available for the transportation sector. Different cities have different congestion patterns, and such information may prove helpful in developing future congestion reduction strategies.
Using traditional yearly average congestion estimates can obscure important variations:
An examination of monthly congestion data, in particular, could offer a view of the systematic intra-year movement or seasonal patterns of congestion.
The monthly average number of hours during the day when at least 20 percent of vehicle miles traveled (VMT) on the instrumented road network (road segments where detailed speed and traffic data are collected) are congested. For this measure, congestion is defined to occur when speeds on a particular highway segment are less than 50 miles per hour.
Source: Federal Highway Administration, Urban Congestion Reporting Measure Definitions.
The average monthly number of weekday congested hours per day for Chicago, Los Angeles, and Houston varied substantially among the three cities over the 3-year period from April 2004 to April 2007 (figure 1).5 The total number of congested hours per day for each city should not be compared to the other cities because the data collection system coverage and free-flow speeds for each city vary dramatically.
However, it is useful to look at the trends and variations in those data. There are some similarities in the trends for the three cities; for example, a tendency for lower levels of congestion near the end of each year. Because seasonality is often driven by periodic fluctuations in weather, vacations, or holidays, the declining congestion around December is not surprising. But there are variations in these patterns, both over time and among the three cities.
Seasonality is a pattern in time series data that repeats every year. This repetitive pattern in transportation can be attributed to a variety of causes. For example, the December pattern noted above is attributed to weather and holiday travel. To quantify the monthly seasonal factors (see box B), BTS uses a statistical procedure, called STAMP,6 to separate the seasonality from the time series data. STAMP reduces the seasonality in 3 years of data to 12 monthly factors and shows how the average monthly congestion deviates from the overall trend for each city.
Seasonal factors reflect the variations that repeat every year to the same extent, e.g., holiday effects, weather fluctuations representative of the season, and so on. A monthly seasonal factor represents the impact that a particular month of the year has on a particular measure relative to what that measure would be if the seasonal influence were removed – or “deseasonalized.”
Source: Bureau of Transportation Statistics
In addition to estimating the seasonal factors, STAMP can also calculate the underlying long-term trend, or trendline, for the congestion data. Figure 2 provides a graphical comparison of the monthly data against the trendline calculated in STAMP for Los Angeles’ weekday congestion. The deviation of the monthly data from the trendline essentially represents the impact of seasonality, and these deviations can be averaged to create the monthly seasonal factors.
The monthly weekday seasonal factors for the three cities are shown in figure 3. December congestion is lower for all three cities (between approximately ½ to 1 hour fewer congested hours per day). There are also some higher congestion levels for all three cities during September and October, months that may be affected by commuters returning to work after summer vacations and schools starting classes or by increased freight traffic transporting imports for the upcoming December holiday season. The other months show considerable differences among the cities. Los Angeles, in particular, has a unique pattern with much higher than normal congestion in February and March, and lower than normal congestion in April, May, and August. There are fewer congested hours in Chicago in both December and January, a time when adverse weather conditions occur frequently. Chicago has its highest level of congestion in September. Houston has the least variation over the year—December is the lowest congested month, with the most difference from the average, while September is the month with the most congestion.
The following provides an in-depth view of the congestion data for the three cities, including whether the monthly patterns for the overall weekday seasonal factors are replicated in the morning, evening, and weekend congestion patterns.
In Chicago, the largest seasonality factors are in December and January (figure 4). In December, the reduced congestion time is almost entirely due to a decrease in the a.m. period. In January, reduced congestion can be attributed to a drop during the p.m. period. During other months of the year, the a.m. and p.m. patterns are generally similar.
There are also differences between weekday and weekend travel in Chicago. For most months the weekday and weekend factors are similar in direction and magnitude, with the biggest differences in January and June. Weekend congested hours are much lower in January and higher in June
Los Angeles has a different seasonal pattern. The a.m. periods in August and December have lower than normal congestion levels. There is also a different pattern of congestion during December than in other months. In December, the p.m. period has somewhat higher congestion levels than normal, while the a.m. period is notably lower. In contrast, congestion during the p.m. periods in April and May is lower than normal, while congestion during the a.m. periods for those months is about normal (figure 6).
The weekday seasonality and weekend seasonality in Los Angeles, in contrast to Chicago, have several months with factors in opposite directions. In Los Angeles, the greatest variability is in the weekday periods, with the highest congestion levels in February and March, and the lowest in August. Weekend congestion is highest in June and lowest in September (figure 7).
Congestion levels in Houston in the a.m. periods in July and December are lower than normal. But the estimated July seasonal factor may not be indicative of a seasonal effect because two of the three July values in the series were signaled as outliers – in opposite directions. In other months, there were only small variations in Houston congestion levels (figure 8).
Overall weekday congestion in Houston in December is lower than normal. Weekday congestion is highest during September. Weekends show very little variation during the year (figure 9).
For the three cities under study, key similarities and differences were found in the dataset on the monthly average number of congested hours per day from April 2004 through April 2007:
Weekday congestion (a.m. and p.m. combined):
The weekday patterns are not always mirrored in the morning and afternoon periods. The two time periods do not always have the same degree of congestion for each month, possibly confounding the monthly weekday congestion estimates:
Morning (a.m.) congestion:
Afternoon (p.m.) congestion:
Congestion patterns change during the weekends, when fewer commuting trips are made. But there are still differences among the three cities.
As more congestion data are gathered over the next few years, the estimates of these seasonal factors will improve, thereby allowing more in-depth statistical studies of the seasonality of congestion.
1 David Schrank and Tim Lomax. September 2007. The 2006 Urban Mobility Report, Texas Transportation Institute, College Station, TX;
2 U.S. Department of Transportation, Strategic Plan, Fiscal Years 2006-2011;
3 See U.S. Department of Transportation, Research and Innovative Technology Administration, Bureau of Transportation Statistics, Compendium on Congestion: Issues and Analyses Across Modes, May 2007, p. 16.
4 These data are collected and compiled for use in the Federal Highway Administration’s Urban Congestion Report, various months, using detailed traffic flow and speed information from the Intelligent Transportation System (ITS) installed on the major freeways most affected by congestion in each of the subject cities. The data are compiled by the Texas Transportation Institute and Noblis, Inc.
About this Report
This report was prepared by Jeffery Memmott and Peg Young, members of the Bureau of Transportation Statistics (BTS) Trending and Forecast Team. BTS is a component of DOT’s Research and Innovative Technology Administration (RITA). The estimates in this report were developed from monthly Intelligent Transportation System congestion data collected and compiled for the Federal Highway Administration by the Texas Transportation Institute and Noblis, Inc.
For related BTS data and publications: www.bts.gov