2.0 Project Summaries
This chapter describes the eight congestion pricing projects that were examined as part of this study. For each project, the overall objectives are summarized, along with the travel and environmental analysis methodologies and findings. The information presented in this chapter is strictly descriptive, that is, a summary of information contained in the reviewed literature. As such, the “findings” discussion within each project summary presents the findings reported in the literature, including any reported observations or conclusions pertaining to the strengths and weaknesses of the methodologies. Chapter 3.0 compiles the key project findings presented in this chapter and synthesizes the state-of-the-practice, including an assessment of collective strengths, weaknesses and gaps associated with those practices.
The methodologies used in before-after evaluations of congestion pricing projects are the focus of this study, rather than the impacts themselves. However, the project summaries that follow as well as some of the summary information in Chapter 3.0 include findings (impact) information because the information was generally readily available from the same literature that was reviewed for methodologies.
Table 2‑1 presents basic characteristics of each of study project. Additional information on each project is presented in the summaries that follow.
This project summary is based primarily on a 2007 report prepared by the project sponsor, the Oregon Department of Transportation.5
The Congestion Pricing Strategy and Objectives
The Oregon Department of Transportation conducted a simulated field demonstration of a time-of-day and area-based congestion pricing scheme. The congestion pricing investigation, the “Road User Fee Pilot Test,” was part of a broader project—the Oregon Mileage Fee Concept—which examined the general notion of mileage-based charges in lieu of traditional gas taxes.
The congestion pricing demonstration was conducted in 2006-2007 and involved providing volunteer (general public) drivers of instrumented vehicles feedback on how much they would have paid for their travel under the traditional gas tax approach versus under a per-mile charging scheme in which charges were higher for travel during peak periods and within congested zones (within the Portland region). The drivers did not actually pay the congestion charges but were asked to make their travel decisions as if they were actually being charged. Feedback on what the mileage fee charges would have been was provided to drivers when they refueled at special, participating gas stations. Drivers were also updated monthly by program administrators as to the status of their charges.
This project was concerned only with understanding travel behavior changes and did not assess how those behaviors translated into traffic congestion or transit ridership changes. This project did not assess any environmental impacts.
This project examined the following congestion pricing-related travel performance measures:
Changes in total VMT, time-of-day changes, and route changes were collected using the in-vehicle global positioning system (GPS) devices. Mode shifts were assessed qualitatively via traditional (non-travel diary) surveys of participants.
The project included both a baseline (no pricing) and experimental phase. During the experimental phase, drivers were broken into three groups, a control group where no mileage-based or congestion pricing elements were introduced, a “VMT group” where flat, per-mile (not congestion) pricing was introduced, and a third group where time-of-day and area pricing was introduced. Changes in the performance measures were determined by comparing among the various phases and experimental groups: the baseline phase (no pricing), the experimental phase control group (no pricing), the experimental phase VMT group, and the experimental phase time-of-day/area pricing group.
Analysts concluded that the premium charged in the peak periods motivated participants to change the timing of their trips, seek alternate routes outside the congested zone, or use transit more. Participants in the time-of-day/area pricing group reduced their total VMT by 10 percent relative to the baseline (no pricing, pre-deployment) condition, with a peak hour VMT reduction of 13 percent. Study data also showed that the congestion pricing charges could impact mode choice during peak hours, with distance to transit influencing the extent of impact. Those who live closer to transit stops were more likely to use it during peak periods as an alternative to driving.
This project summary is based primarily on a 2008 report prepared by the project sponsor, the Puget Sound Regional Council.6
The Congestion Pricing Strategy and Objectives
In 2002, the Puget Sound Regional Council (PSRC) received a grant from FHWA to conduct a pilot project to see how travelers change their travel behavior in response to variable charges for road use (variable or congestion-based tolling). Global positioning system tolling meters were placed in the vehicles of about 275 volunteer households. From July 2005 until February 2007, the project observed participant driving patterns before and after hypothetical tolls were charged for the use of all the major freeways and arterials in the Seattle metropolitan area.
The primary aims of the Traffic Choices Study were to:
This study considered only travel behavior changes. It did not examine how travel behavior changes may translate into traffic or environmental impacts.
The study recruited a sample of volunteers and, after establishing their baseline “before-tolling” driving routine, began imposing hypothetical charges (levied against an endowment account as a financial incentive) for access to selected roadway facilities at particular time periods in the day. The study monitored driving behavior of participants for an average of approximately 18 months per household.
The study goal was to determine how travel behavior (distance, timing, and number of vehicle trips) was affected by a range of other factors (e.g., household income, size, location; number of vehicles; availability of transit; day or time of travel; congestion price charged). The study formulated statistical models to explain how measures of travel demand, across households, vehicles, and workers (the dependent variables) are affected by changes in the generalized costs of travel (tolls, out-of-pocket costs and time costs), while controlling for household demographics (income and number of drivers), seasonal factors, and a measure of transit viability (the independent, explanatory variables). The study estimated the elasticities of travel demand with respect to changes in the price of travel (tolls).
Characteristics of the study included:
The billing system provided detailed physical and financial information on trip activity:
Trip purpose and traveler demographic attributes were appended to trip information:
GPS tolling meters were placed in vehicles of 275 volunteer households:
The Traffic Choices study resulted in a number of changes in aggregate travel demand. Under the tolling policy established for the study, the changes included:
The participating households altered the nature and amount of vehicle use in response to hypothetical tolls that increased the costs of travel but did not result in improved travel times.
Many households made notable changes in their travel practices. Households that modified their travel did so in many different ways: taking fewer and shorter vehicle trips, choosing alternate routes and times of travel, or linking trips together to reduce vehicle use altogether. Some households altered their routine travel practices. On the other hand, other households had very limited opportunities to avoid using high demand roads during peak travel times.
The results were consistent with the study team’s expectations. Researchers concluded “paying tolls that reflected the costs of congestion caused many travelers to change aspects of travel behavior, some more than others, depending on the usefulness and convenience of the opportunity for change.”
A conservative analysis of the benefits of network tolling in the Puget Sound region indicates that the present value (2008) of net benefits would be $33.6 billion over a 30-year period. The implementation and operating cost of the system in present value was estimated at $5.5 billion. This provides a benefit/cost ratio of 6.1 for a variable toll network in the region.
Not all aspects of a road network tolling system were fully demonstrated, but the core technology for satellite-based (and whole road network) toll systems were mature and reliable. The tolling system performed as expected and met basic system operating requirements. Further work on system refinement and design of enforcement and billing systems would be required prior to any full system deployment.
The Congestion Pricing Strategy and Objectives
“Commute Atlanta” refers to a multi-faceted, multi-phase program. The portion of the program of interest to this study is a field study of mileage-based value pricing that was carried out between 2004 and 2006. Participants in the overall Commute Atlanta Program include FHWA, the Georgia Department of Transportation and the Georgia Institute of Technology.
The mileage-based value pricing strategy evaluated the effectiveness of a cash incentive to reduce the vehicle miles traveled by volunteer households. For each household, an incentive account was established based on the household’s actual number of miles traveled in the equivalent quarter of the previous, baseline year.7 The amount of the account varied quarterly and was determined by multiplying a dollar amount (which varied from quarter to quarter over the study period, starting at $0.05 per mile and concluding at $0.15 per mile) by the number of miles traveled in the baseline quarter. During the nine-month experiment, households were eligible for a payment from their account each quarter, with the amount of each payment dependent on how many miles they had driven. In the first quarter, $0.05 per mile driven was deducted from the payment. So, if a household had traveled 1,000 miles in the first quarter of the baseline year, and 900 miles in the first quarter of the experimental period, their total potential payment would have been $50 (1,000 miles x $0.05) but their actual payment would have been $5 ($50 – [900 miles x $0.05]). Households that traveled as many or more miles than in the baseline period received no payments.
The evaluation of the mileage-based value pricing strategy used a before (no incentive) versus after (with incentive) methodology and focused strictly on household travel behavior. No investigation of environmental impacts was performed. However, as part of a separate study of the potential for congestion pricing in the Metropolitan Atlanta Area, an air quality evaluation framework was proposed and that proposed approach is summarized here.
The evaluation of the mileage-based value pricing component of Commute Atlanta considered the following travel-related performance measures and variables:
Household travel behavior data were collected using volunteers’ vehicles instrumented with GPS and data loggers which monitored travel parameters (position, speed, etc.) from which data could be accessed remotely. These data were collected from 95 households that constitute a subset of a much larger group of households who have participated in other aspects of the overall Commute Atlanta program data collection over a number of years.9
The collection of household socio-economic data was performed using longitudinal (panel) surveys and supported a detailed, case study analysis of each of the 95 households to understand the relationship between demographic changes over the course of the study and their impact on travel behavior. Participating households were surveyed monthly over the course of the study on household demographic parameters. It is important to note that although the household surveys included measurement of parameters that are believed to be closely linked to travel behavior, such as work status changes, the surveys do not appear to have specifically queried participants on the motivations behind travel decisions and changes.
The overall Commute Atlanta program included additional data collection, including two-day travel diaries and surveys focusing on employer commute incentives, but it does not appear that those data were used in the analysis of the mileage-based value pricing demonstration.
The air quality impacts of the Commute Atlanta mileage-based value pricing demonstration were not investigated. However, as part of a separate study of the potential for congestion pricing in the Metropolitan Atlanta Area,10 an air quality evaluation framework was proposed that focuses on two air quality performance measures:
Had vehicle emissions been studied, the plan was to calculate them using emission rates (grams per mile) from the EPA MOBILE emission factor model multiplied by observed link vehicle miles traveled. Localized CO levels were to be calculated using the EPA CALINE4 model, which utilizes user supplied traffic, meteorological, and topographic inputs coupled with EPA CO emission rates to estimate CO concentrations at specific modeled locations associated with roadway intersections.
The data collection associated with the mileage-based value pricing demonstration included trip lengths, trip durations and other performance measures. However, the literature that was reviewed included only findings related to vehicle miles driven and the influence of household demographic factors on miles driven. The later findings are especially important as this study delved deeply into these particular exogenous factors (changes in household demographics) and represents the most critical examination of these considerations that was found in the literature. In the case of this study, the findings related to exogenous factors are more significant than the travel impacts.
It was reported that more than half of the mileage-based value pricing households reduced their travel and that, overall, vehicle miles driven was reduced by 3 percent relative to the baseline (no incentive) period. However, researchers concluded that the VMT reduction was not significant given the significant variability in before-after household travel changes among households.11
The researchers concluded that changes in household demographics over the course of their study had a significant impact on observed VMT changes and more of an impact than other exogenous factors. They found that of the 95 households in the case study, only 28 households remained stable with respect to all six major demographic characteristics: home location, work status, household structure, income, school(s) attended, and vehicle ownership. The most common change was vehicle ownership (40 percent), followed by work status (34 percent). The researchers concluded that among the demographic changes, the impact of work status change was most evident, but that home location and household structure changes were also important influences on VMT. Given the small sample size, they were not able to form conclusions about how the other demographic parameters affect travel.
Based on these findings, the researchers identified a number of methodology enhancements that are necessary in order to form valid conclusions regarding the household travel impacts of congestion pricing projects. These recommendations focus on larger sample sizes and collection of more detailed information from participants to better understand the reasons behind the reported travel behavior. The researchers recommend using “case study” approaches such as theirs in which the impact of changes in household demographics are thoroughly explored at the individual household level. They also suggest that home interviews and focus groups would be helpful in understanding travel behavior changes.
Finally, the researchers cast some degree of doubt on the results of other congestion pricing studies that may not have adequately controlled for household demographic changes (including two studies discussed in this report: Oregon and Puget Sound). Specifically, they state that: “…the findings of similar studies that have been conducted should be eyed with caution and researchers need to be careful in drawing any conclusions on the impact of pricing incentives from these studies.”12
The Congestion Pricing Strategy and Objectives
In May 2005, the Minnesota Department of Transportation (Mn/DOT) started operation of the State’s first high occupancy toll facility on a segment of the I-394 corridor in the Minneapolis/ St. Paul region. The system, known locally as MnPASS, was the first deployment of HOT lane strategies in Minnesota and the second in the United States that dynamically adjusts pricing levels in response to varying traffic conditions. The travel behavior/traffic objective of the congestion pricing was to adjust tolls so as to meter use of the HOT lane to levels that would provide consistent free-flow speeds.
Mn/DOT conducted a comprehensive monitoring and evaluation effort to assess the I-394 MnPASS system, including investigation of traffic impacts (using both pre- and post-deployment data); traveler behavior changes, including mode choice; and environmental impacts.
Travel related performance measures utilized by Mn/DOT in their evaluation included the following:
Reported travel behavior was collected through three waves of panel surveys of project corridor residents and, as a control group, users of HOV lanes in a different corridor, I-35W. Between 800 and 900 residents participated in each wave. The first wave of surveying was conducted prior to HOT lane deployment and the second and third waves were conducted post-deployment. The survey effort included telephone surveys, mailed questionnaires, and travel logs. The travel logs included a travel diary for a single day and general travel behavior information for one week.
Traffic volumes, speeds and travel times were collected directly, or derived, from Mn/DOT Regional Transportation Management Center (RTMC) detectors. The detectors provided a nearly continuous source of data on vehicle volumes for a period between January 2003 (pre-deployment) and July 2006 (post-deployment). Data were collected from all detectors located within the roadway project section. Data from selected stations upstream and downstream of
Environmental performance measures consisted of:
Before and after data on CO levels were collected with emissions sensing stations deployed at several strategic locations near the roadway. Pre-implementation data were supplemented by historical emissions data from existing sensor stations located in the corridor. One-hour CO averages were recorded for the a.m. and p.m. peak hours for the pre- and post-implementation lane operation. Each hour of data collected, plus the previous seven hours collected, were averaged to calculate the eight-hour CO average.14 The one-hour CO averages were compared for a.m. and p.m. peak hours, for pre- and post-MnPASS lane operation for the same dates that noise monitoring was conducted and traffic counts were taken.
Before and after data on roadway noise levels were collected from field sensors temporarily deployed at several strategic locations adjacent to the roadway. The noise data were collected in close coordination with several detailed vehicle counts, documenting the number and type of vehicles using the roadway. These data were collected prior to the opening of the MnPASS lanes to provide an assessment of pre-implementation noise levels. Field noise data was again collected after implementation.15
To evaluate and explain specific noise level changes in the a.m. and p.m. peak hours, the MINNOISE model was used. The MINNOISE model uses geography, traffic, and vehicle speeds as input parameters for the program and is based on the FHWA Traffic Noise Model. Sound level differences between measured after and before, and modeled sound level differences between after and before were compared and tested for statistically significant changes in noise level.16
Environmental justice was examined through the same survey effort that collected data on traveler behavior and attitudes. Demographic data including income, education, employment status, gender, age and ethnicity were collected. These data allowed researchers to compare survey responses pertaining to social equity, traveling experiences, use of the HOT lanes and attitudes about MnPASS tolling operations across various demographics.
Findings related to traffic measures consisted of the following:17
Findings related to travel behavior performance measures and travelers’ attitudes and opinions consisted of the following:18
In regard to air quality, a 0.20 parts per million (ppm) CO one-hour average increase was found in the a.m. peak hours at the north monitoring site, with a 0.03 ppm CO decrease in the p.m. peak hours. At the south monitoring site a one-hour average increase of 0.29 ppm CO was found during a.m. peak hours and a 0.01 ppm increase during the p.m. peak hours. The increases in the one-hour average CO levels were considered minimal, and the CO concentrations remained well below the 30 ppm one-hour CO air quality standard for the State of Minnesota. It was concluded that the operation of the MnPASS lane did not result in a substantial impact on the air quality due to any changes in traffic patterns in the project area.19
In regard to noise levels, taking all sites into account and averaging the a.m. and p.m. noise level measurements, it was found that there was not a statistically significant change in the average neighborhood sound pressure level, during the peak hours. The analysis showed instances where noise level changes were confidently attributed to changes in traffic patterns due to the MnPASS lane; nevertheless, there was not a statistically significant change in average neighborhood sound pressure level.20
In regard to environmental justice, the evaluation did not identify any significant correlation between demographics and project benefits. It was noted that beneficiaries of the HOT lane include a diverse population across all income, age, race/ethnicity, employment, and mode usage groups.
The Congestion Pricing Strategy and Objectives
Eight miles of HOV lanes on Interstate-15 (I-15) were converted to HOT lanes (opened to paying single occupant vehicles) in December 1996. The I-15 HOT lane facility—“FasTrak Lanes”—uses a dynamic, real-time tolling structure in which tolls vary with the level of congestion in order to maintain free-flow traffic conditions. Fees can vary in 25-cent increments as often as every six minutes. All transactions are electronic; overhead antennas read a transponder affixed to the inside of a vehicle’s windshield and deduct the toll electronically from the driver’s prepaid account.21 Pricing is based on maintaining a Level of Service “C” in the HOT lanes.
San Diego State University (SDSU) researchers conducted an independent, multi-element, three-year (1997-2000) evaluation to assess HOT lane impacts on the I-15 corridor and the San Diego region. The research team studied changes in I-15 corridor traffic, travel behavior, and attitudes toward the project throughout its duration. A control corridor (a portion of Interstate 8) was used for traffic-related analyses in order to help differentiate project-related changes from exogenous factors.22
The evaluation was conducted in a series of periodic waves, which generally occurred in the spring and fall of each year to avoid interference from the typical seasonal changes in traffic patterns. For most of the studies, SDSU conducted five waves of data collection between fall 1997 and fall 1999.
Travel related performance measures consisted of the following:
Before-and-after traffic volume, speed, travel time, toll violations, and vehicle classification data were collected from various roadway detectors, including those associated with the HOT lane electronic toll collection system. The data on vehicle occupancy were objective—as opposed to estimated or modeled—but the specific data collection method was not identified in the literature that was reviewed. The source of incident data is unknown.
Post-deployment travel behavior data were collected through a 5-wave panel survey of three groups: HOT lane users, I-15 general purpose lane users (both solo drivers and carpoolers), and I-8 (control corridor) travelers. Surveys were conducted between 1997 and 1999.
Environmental performance measures were restricted to total emissions of volatile organic compounds (VOCs), nitrogen oxides (NOx), PM10, and CO.
Total emissions of each pollutant were calculated as the product of emission factors—derived using the California Air Resources Board (CARB) EMission FACtors model (EMFAC)—multiplied by the number of vehicles and by the length of the corridor segment. Total emissions for each peak period along the corridor were determined by aggregating emissions over all I-15 segments in the corridor for all time periods and all vehicle types.
Overall (all lanes) traffic volume increases along the I-15 corridor were attributed to the substantial volume increases in the I-15 express lanes (48 percent) during the 3-year study period. The I-15 pricing project alleviated congestion on the I-15 main lanes by redirecting an increasing share of volume onto the I-15 express lanes. The study team concluded the increases in total I-15 corridor volume reflected more the pressures of population and employment growth in the travel corridor.23
In the monitoring period, the I-15 corridor experienced a substantial increase in SOV volume and a corresponding decrease in HOV volume during the a.m. peak period. The increase in SOV volume along the I-15 express lanes was attributed to scheduled program expansion, but could also have been a result of strong demographic and socioeconomic pressures in the corridor because of relatively high rates of commercial and residential development.24
A decline in HOV main lane volume along I-15 contrasted sharply with an observed rise in HOV volume along the I-8 control corridor from 1997 to 1999. The results strongly suggested corridor-specific factors, including the I-15 pricing project, were responsible for these differences. Unrewarded carpooling on the express lanes may have played a role plus limited access to the lanes in the study period, with only one entrance and one exit.25
LOS C, required by law to be maintained at all times on the express lanes, was sustained at virtually all times. There was a decrease in variance of volume distribution from fall 1996 to fall 1997 in the a.m. peak period and a subsequent general increasing trend through fall 1999 in the variance of peak-period volume distributions in both a.m. and p.m. peak periods. This result strongly suggested that the dynamic pricing structure was able to create desirable redistribution of a portion of express-lane traffic from the middle of the peak to the shoulders. Researchers were unable to find a sufficient explanation for the HOV portion of the shift to shoulder periods.26
Researchers noted a significant increase in express lane use from spring 1998 to spring 1999. They concluded that it may have reflected the effectiveness of the shoulder pricing policy (further decreased toll prices in the off-peak hours) introduced in August 1998 to distribute traffic more evenly throughout the peak period away from the peak hour.27
The dynamic pricing influenced the times at which people traveled. FasTrak customers exhibited later departure times than did other I-15 users in all survey waves except for one. Researchers also found that 76 percent of FasTrak customers would leave at a different time in the morning if there were no FasTrak. The majority of those would leave earlier for work to account for the longer and highly unreliable travel times without FasTrak.28
The SDSU evaluation team reported that data from the fall study waves from 1997 to 1999 demonstrated that the FasTrak program moderated emission levels along the I-15 corridor during a period in which emission levels increased substantially along the I-8 corridor. The average relative increases along I-8 were three times larger than the average relative increases along the I-15 corridor in the a.m. peak period. In the p.m. peak period, this difference was even greater (five times larger). The changes in average emission levels along the I-15 main lanes and express lanes over the same period reflected the influence of the FasTrak program in displacing traffic from the main lanes to the express lanes. Average emission levels of all four pollutants on the express lanes increased substantially from fall 1997 to fall 1999 in both peak periods.29
Since the study was observational in nature, and other potentially influential factors could not be controlled or measured precisely, the team could not definitively attribute all observed differences in I-15 and I-8 emission profiles to the HOT lane program. However, the effects of the corridor-specific factors were more pronounced along I-15 than I-8 and could be expected to have increased emission levels along the I-15 corridor. No factors, other than the FasTrak program, were identified that could have reduced or mitigated increases in I-15 emission levels.30
The Congestion Pricing Strategy and Objectives
In January of 2006, the City of Stockholm implemented a cordon/area congestion pricing project spanning seven months, known as The Stockholm Trial. The stated goals of the trial were:31
The congestion pricing component of the trial was to charge motorists a tax whenever they entered Inner City Stockholm. Inner City Stockholm borders or boundaries were defined, and were equipped with control points around the charging zone to monitor vehicles entering and exiting the zone; the vehicles were identified through photographing the license plates or via onboard units. Reducing traffic to improve traffic flow and manage congestion were clear expectations of this project.
An extensive evaluation of the Stockholm Trial was performed using a wide variety of before-after data.
The evaluation of the Stockholm Trial included the following travel performance measures:
Traffic volumes were collected in the field using various data collection methods. Vehicle kilometers traveled were estimated using traffic models. Traffic queue lengths were measured with instrumented test vehicles. Journey times data were collected using two methods: 1) Instrumented vehicles driven by volunteers (50 commuters), and 2) License plate reader vehicle matching.
Individual traveler behavior data were collected through a travel diary (one day) panel study. The panel study included three waves of data collection—two before the pricing project and one after pricing began. More than 30,000 individuals participated in the panel study.
Transit ridership and travel times were collected using automated on-board detectors, with the exception of ridership on underground rail which was collected manually. Transit rider perceptions of the pricing project were gathered through on-board surveys.
The evaluation of the Stockholm Trial included the following environmental performance measures:
Vehicle emissions of PM10, NOx, NO2, CO2, CO, and VOCs were calculated using observed traffic data (vehicle kilometers traveled) and model-derived emission factors. Ambient pollutant levels were measured using roadside monitors. Estimates of exposure to vehicle-generated emissions were made using air quality dispersion models. Noise levels were measured using roadside monitors.
The environmental justice impacts of the Stockholm Trial were evaluated using traveler origin-destination data collected through a panel travel diary study conducted in two waves which included approximately 24,000 participants, coupled with a regional travel demand model. The travel-diary derived information on trip making was fed into a regional travel demand model and the model was used to estimate the travel times, congestion charges paid, and adaptation costs (e.g., switching to transit) associated with the observed (travel diary) trips under both “with pricing” and “without pricing” scenarios. The travel diary-derived trip data were also used to calculate the impacts of three hypothetical pricing revenue redistribution scenarios: 1) Revenues distributed evenly to all county residents, 2) Reduction of transit fares, and 3) Reductions to income tax.
Evaluators concluded that the Stockholm Trial was able to manage congestion and increase flow and accessibility. The traffic volume in Inner City Stockholm decreased 16 percent in the morning and 24 percent in the afternoon and early evening. The sum of distances traveled by all motor vehicles (vehicle kilometers traveled) declined 14 percent within the charging zone from 2005 to 2006. Journey times decreased by 3 percent. Data on average queue lengths did not illuminate clear project impacts. Public transport utilization increased by 6 percent overall and by as much as 10 percent during peak hours. Transit customer satisfaction varied by route. Satisfaction among passengers on existing routes dropped slightly, from 66 percent to 61 percent, while 87 percent of passengers on new routes were satisfied32.
Travel behavior findings included the following:
Evaluators concluded that decreases in traffic volume and increases in traffic flow from the Stockholm Trial impacted air quality and noise. Reductions in VOC, CO, NOx, CO2, and PM10 ranged from 8.5 percent to 14 percent. Monitored noise levels showed minor declines.
Major findings of the environmental justice evaluation consisted of the following:
Researchers identified a number of exogenous factors that probably significantly impacted evaluation findings, including increases in fuel prices, seasonal variation, weather conditions which impacted monitored air quality, and other transportation projects. These factors could not be controlled; however, when possible the evaluators attempted to isolate the effect of exogenous factors by comparing different monitoring occasions and different areas.33
The Congestion Pricing Strategy and Objectives
London launched a cordon road pricing project focusing on Central London in February 2003 and the charging zone has since been expanded—the “Western Extension.” The objective of this project was to reduce traffic, improve the speed of buses, create revenue, and improve quality of life.34 Motorists pay a standard, flat rate to drive cars within the congestion charging zone. The rates did not vary per location and after paying the vehicles can exit and enter as many times as desired. Charges are applied on weekdays from 7:00 a.m. to 6:30 p.m. Congestion charges are paid in advance or on the day of travel by telephone, regular mail, Internet or at retail outlets. There are no toll booths or other roadside payment infrastructure. To enforce and monitor the payments, the system made use of networked video cameras (automatic license plate recognition) to record license plate numbers, then matched the license plate numbers to a paid list.35 Over the years, many aspects of the charging program have been modified, including increases in the daily charge.
The London congestion charging projects have been rather extensively studied. These studies have included before-after evaluations of a variety of travel and environmental impacts.
The organization Transport for London has conducted a long-term and on-going evaluation of the London congestion charging initiative. Six annual reports on the monitoring efforts have been published to date, with the first report in 2003 describing baseline (pre-pricing) conditions.36 Although many evaluation methods have remained the same over time and as the pricing was extended westward, some methods have evolved over time or have been applied only to the central London or Western Extension studies. Of course, different findings have emerged from the various studies. This summary endeavors to present a composite view of all of the travel and environmental impact evaluation methodologies that have been applied.
The London congestion pricing evaluations have included before and after measurement of the following travel-related performance measures:
Over the years, a wide variety of data collection methods have been used to collect these various performance measures, with multiple data collection methods often used to collect the same performance measures. The vast majority of data for travel-related performance measures has been collected directly, either in the field using various detectors and manual methods or with surveys and interviews with travelers and other stakeholders. Although travel demand modeling has been used to complement observed traffic data, very few of the travel-related performance measures have been derived solely through modeling or simulation.
Data collection methods utilized for travel-related performance measures have included the following:
In addition, a wide variety of surveys, interviews and focus groups have been conducted to collect attitudinal, general travel behavior, and detailed (travel diary) travel behavior. Evaluation of the London congestion pricing scheme has featured the most extensive and sophisticated use of surveys, interviews and focus groups of all of the projects that have been reviewed as part of this study. These efforts are distinguished in the following respects:
Specific survey, interview and focus group activities used to collect travel behavior and attitudinal travel-related performance measures have included:
Evaluations of Central London Congestion Charging have included before and after measurement of the following environmental impacts:
Many data collection and analysis methods have changed over time, but cumulatively, the following methods have been used:
The methodology used to calculate emissions was compatible with the approach used in the Mayor of London’s Air Quality Strategy.37 The emissions estimation approach included breaking down vehicle emissions by specific traffic variable, e.g., traffic volume, speed changes and emission rates (reflecting changes in vehicle fleet mix and emissions technology changes). This approach was not evidenced in any of the other study projects. From the beginning, monitored pollutant levels were not expected to reveal impacts that could be traced specifically to the congestion charging scheme, but were performed regardless.
Original plans included both modeling and monitoring noise levels. Noise modeling was planned to utilize travel flow, composition, and speed to produce noise mapping and generate noise predictions. However, no results of any modeling efforts were located in the literature, which may indicate that only monitoring was performed.
The consideration of environmental justice issues was conducted within the broader assessment of “social impacts”—the comprehensive assessment of the impact of the pricing on people’s attitudes, perceptions, and abilities. This analysis did consider differences among different income groups but income and race were not as central of a focus as is sometimes found in U.S. studies. Rather, the London analysis has focused on user groups defined by other criteria, such as transit users, people living outside the charging zone, the disabled, people lacking automobiles, and shift workers. Most of the data pertaining to these issues were collected through surveys, interviews and focus groups. Although general traveler/household surveys included demographic data allowing results to be sorted by various characteristics relevant to environmental justice, a number of special surveys and focus groups were conducted focusing specifically on user groups of interest.
A tremendous volume of evaluation results has been published over the course of six annual monitoring reports and numerous additional studies. Overall, the latest annual report (published in July 2008) concludes that congestion charging continues to meet its fundamental traffic and transport objectives and that the scheme continues to deliver congestion reduction generally equal to the 30 percent reduction achieved in the first year.38 The same report cites results from the Fifth Annual Report stating that, given the influence of many non-project related factors, it can be misleading to compare recent congestion levels (2006 data) to pre-charging levels but that nevertheless, such a comparison shows 2006 congestion levels to be 8 percent lower than pre-charging levels. The Fifth Annual Report (published in 2007) elaborates that charging has accentuated positive trends such as reduced accidents and emissions while mitigating negative trends like increasing congestion.39
Additional specific travel findings as well as environmental findings are summarized below. The specific findings that follow are adapted primarily from the two most recent annual reports—the fifth and sixth.
After the first year of the project in the Central London charging zone, VKT decreased by 15 percent for vehicles with four wheels or more and the number of vehicles entering the zone declined by 18 percent. A 14 percent decrease occurred in journey times. The average network speeds in 2003 were 14 kilometers per hour (9 miles per hour), which increased to 17 kilometers per hour (11 miles per hour) in 2006.40 Research found that in the beginning of the congestion pricing project the traffic volume reductions were significant and as the congestion pricing continued the reductions in traffic volume also continued, but at a slower rate. Overall, the charging zones that began in 2003 and continued through 2006 led to 21 percent less traffic entering the zones.
In the Western Extension zone traffic volume decreased by 14 percent compared with pre-charging conditions in 2005-2006. Also an 11 percent decrease in VKT for vehicles with four wheels or more was reported. Average network speeds and journey time results for the Western Extension were not clearly represented. The volume of commuter trips increased 33 to 38 percent.41
Examinations of mode split—the percentage of trips made by the various modes, including driving, public transportation and elimination of the trip entirely—were conducted in support of the second annual monitoring report which reported the first post-deployment year findings related to the central charging zone, and the sixth annual monitoring report, which reported results pertaining to the Western Extension of the pricing zone. Overall, the results suggest that the charging scheme has prompted significant shifting from driving to alternate modes, primarily bus transit.
The early results for the central zone showed that 65,000 to 70,000 car trips no longer cross into the charging zone and estimated the displacement of those trips as follows:42
The results for the Western Extension show that 32 percent of sampled drivers changed their behavior in response to the charge.43 Among that group that changed their behavior, the breakdown of specific changes was as follows:
Examples of findings related to public transportation usage include the following:
As anticipated, it has not been possible to distinguish project impacts from the impacts of exogenous factors in the monitored air quality results. However, results of the emissions calculations indicate that the project has contributed to emissions reductions. In Central London the overall traffic emissions change between 2002 and 2003 included a decrease in NOx of 13.4 percent, a decrease in PM10 of 15.5 percent, and a decrease in CO2 of 16.4 percent. The Western Extension results comparing 2006 to 2007 also showed a decrease of 2.5 percent in NOx, a decrease of 4.2 percent in PM10, and a decrease of 6.5 percent in CO2 based on combined traffic volumes and composition change.44
Early noise results—changes from the baseline (2002) observed in the first year of post-deployment (2003) were small and considered imperceptible in typical urban conditions. No project-related impacts on noise were identified. The fourth annual report noted that “Limited sample surveys of ambient noise in and around the charging zone continue to suggest the absence of a detectable congestion charging impact.” Later reports do not contain any additional new noise findings.
The various results related to environmental justice issues—the impact of the charging scheme on various types of people—do not identify broad, significant adverse impacts but do note some concerns. Explicit, prominent identification of “winners” and “losers” are not at all prominent in the Transport for London annual reports. This passage from the sixth annual report which focuses on the Western Extension is typical of the sorts of summations found in the Transport for London annual reports:
Examples of specific reported findings include the following:
The Congestion Pricing Strategy and Objectives
Singapore first implemented cordon congestion pricing in 1975 in order to better manage traffic and to reduce vehicle emissions. The initial pricing project was an Area Licensing Scheme in which vehicles were charged a fee for entering a 2.0-mile square central business area during the a.m. commute period. Vehicles entering the priced zone along any of 28 entry points were required to display a pre-purchased daily or monthly windshield license. Transit buses, motorcycles and vehicles carrying more than 4 people (high occupancy vehicle [HOV] 4+) were excluded from the charge. That initial scheme evolved through the late 1990’s, including extension of the charging periods, increases in the charges, and, by virtue of expansion of price points to roadways outside the central business area, evolution to an area-wide pricing scheme. In 1998, the system was converted to a fully automated, electronic system whereby charges are collected using vehicle transponders with smart cards, dedicated short-range communication (DSRC), and readers mounted on overhead gantries.
No detailed information on evaluation methods was found in the published literature. Although general in nature, the bulk of the best information is found in the FHWA report “Lessons Learned from International Experience in Congestion Pricing.”46 That report summarizes various evaluations of the Singapore congestion pricing activities and indicates that the following travel performance measures have been investigated:
There is no specific information available on data collection, but based on the discussion of impacts, it appears that most of the performance measures were analyzed based on objective, observed data, e.g., traffic volumes and speeds derived from various detector data and traffic crashes based on police reports. It appears that at least some of the data, especially mode share information, were collected via traveler surveys.
Environmental impacts considered in the Singapore evaluations include:
Analysis of vehicle emissions has included both roadside monitoring as well as calculation of emissions based on travel impacts. Perceived pedestrian safety was evidently assessed through the use of surveys. Equity implications were assessed via a modeling analysis and surveys of travelers. It appears that the modeling analysis utilized the geographic distribution of various travel impacts coupled with geographic socioeconomic data to infer how travel impacts distributed across various impact groups. In addition, equity implications—specifically, various types of travelers’ perceptions of and responses to congestion pricing—were assessed through surveys.
Although published literature does not explicitly identify how exogenous factors were considered in the various evaluations of the Singapore congestion pricing projects, various summaries of project impacts include references to several types of exogenous factors. These factors include auto ownership and employment.
Overall, the Singapore congestion pricing projects have been effective in reducing congestion and vehicle emissions and are generally not believed to have significantly and disproportionately negatively impacted lower income populations. The initial system introduced in 1975 reduced traffic volumes entering the priced zone by 44 percent; the share of HOV 4+ trips increased from 8 to 19 percent and bus share increased from 33 to 46 percent; a.m. peak speeds inside the priced zone increased by 20 percent or more; and speeds increased 10 percent on inbound roadways leading to the priced zone.47 Other travel responses to the initial system included motorists shifting their trips to just before or after the priced time periods and diversion of trips to alternate, non-priced routes.48
Immediately following the introduction of the initial pricing project, measured CO levels in the morning peak period within the priced zone declined to below pre-project levels and monthly average NOx levels also decreased. These reductions were attributed to the large reduction in automobile travel. Measurements of smoke and haze also showed declines but those declines could not be fully attributed to the pricing project.49
Reported equity impacts include the following:50
5 Whitty, James M., “Oregon’s Mileage Fee Concept and Road User Fee Pilot Program: Final Report,” Oregon Department of Transportation, November 2007.
6 Puget Sound Regional Council, “Traffic Choices Study–Summary Report,” April 2008.
8 Xu, Zuyeva, Kall, Elango, Guensler, “Mileage Based Value Pricing: Phase II Case Study Implications of the Commute Atlanta Project.” Transportation Research Board 2009 Annual Meeting.
9 Xu, Zuyeva, Kall, Elango, Guensler. 2009
10 Ross, Guensler, et al. 2008. “Final Report – Congestion Pricing Response: Study for Potential Implementation in the Metropolitan Atlanta Area.” Prepared for the Georgia Department of Transportation by the Center for Quality Growth and Regional Development & School of Civil and Environmental Engineering at the Georgia Institute of Technology.
11 Xu, Zuyeva, Kall, Elango, Guensler. 2009.
12 Xu, Zuyeva, Kall, Elango, Guensler. 2009.
13 Cambridge Systematics, Inc., Short-Elliott-Hendrickson, Inc., and LJR, Inc. “I-394 MnPASS Technical Evaluation-Final Report.” Minnesota Department of Transportation and the Metropolitan Council, Minneapolis, Minnesota. November 2006.
14 Jordahl-Larson, Marilyn, et. al. “MnPASS Air & Noise Analysis—Final Draft.” Environmental Modeling and Testing Unit, Office of Environmental Services, Minnesota Department of Transportation, Minneapolis, Minnesota. October 2005.
15 Jordahl-Larson, Marilyn, et. al., 2005.
16 Jordahl-Larson, Marilyn, et. al., 2005.
17 Cambridge Systematics, Inc., Short-Elliott-Hendrickson, Inc., and LJR, Inc. 2006.
18 Zmud, Johanna. “MnPASS Evaluation Attitudinal Panel Survey Wave 3—Final Report.” Humphrey Institute of Public Affairs, University of Minnesota; NuStats, Austin, Texas. August 2006.
19 Jordahl-Larson, Marilyn, et. al., 2005.
20 Jordahl-Larson, Marilyn, et. al. 2005.
21 K.T. Analytics and Cambridge Systematics, Inc., 2008.
22 Supernak, Janusz, et. al. “San Diego’s I-15 Congestion Pricing Project: Traffic-Related Issues.” Transportation Research Record, No. 1812, Paper No. 02-4169. Transportation Research Board, National Academies, Washington, D.C., 43-52. 2002.
23 Supernak, Janusz, et. al. 2002.
24 Supernak, Janusz, et. al. 2002.
25Supermak, Janusz, et. al. 2002.
26Supermak, Janusz, et. al. 2002.
27Supermak, Janusz, et. al. 2002.
28Supernak, Janusz, et. al. 2002.
29 Supernak, Janusz, et. al. 2002.
30 Supernak, Janusz, et. al. 2002.
31 Stockholmsforsoket. “Facts and Results from the Stockholm Trials.” December 2006.
32 Stockholmsforsoket. December 2006.
33 Stockholmforsoket. “Evaluation of the Effects of the Stockholm Trial on Road Traffic.” June 2006.
34 United States Department of Transportation, Federal Highway Administration. “Lessons Learned from International Experience in Congestion Pricing.” Prepared by K.T. Analytics. August 2008.
35 Litman, Todd. “London Congestion Pricing: Implications for Other Cities.” Victoria Transport Policy Institute. January 2006.
36 Transport for London, “Central London Congestion Charging – Impacts Monitoring, First Annual Report,” Mayor of London, 2003.
37 Transport for London. 2003.
38 Transport for London, “Central London Congestion Charging – Impacts Monitoring, Sixth Annual Report,” Mayor of London, July 2008.
39 Transport for London, “Central London Congestion Charging – Impacts Monitoring, Fifth Annual Report,” Mayor of London, July 2007.
40 Transport for London, 2007.
41 Transport for London, 2008.
42 Transport for London, 2003.
43 Transport for London, 2008.
44 Transport for London, 2007.
45 Transport for London, 2008. Pg. 7.
46 K.T. Analytics, 2008.
47 K.T. Analytics, 2008.
48 Toh, Rex S. (2003), “Road Congestion Pricing in Singapore: 1975 to 2003.” Transportation Journal, March 22, 2004.
49 K.T. Analytics, 2008.
50 K.T. Analytics, 2008.
United States Department of Transportation - Federal Highway Administration