Tolling and Pricing Program - Links to Tolling and Pricing Program Home

4. Effectiveness of Congestion Pricing Strategies: Literature Review Key Findings

Is there empirical evidence that people will shift travel times and/or travel modes in response to traffic management strategies such as congestion pricing? What is the potential for other modes (e.g., transit, vanpools, carpools, etc.) to attract additional mode share, if free-flow service on all freeways were guaranteed to these modes through aggressive and active traffic management strategies such as congestion pricing?

Many studies have been conducted to answer the many facets and potential solutions to the traffic congestion question. Experience from around the world and in the United States has demonstrated that congestion pricing represents one of the most powerful tools to change driving behavior. Empirical studies of congestion pricing have been conducted in many different venues, and in many types of applications. It should be noted that other direct influences on traffic volumes and occupancy include fuel prices[8] and parking costs (which are not addressed in detail in this review). Comprehensive approaches that include "carrots" of faster travel times, preferred parking and/or lower costs for shared rides, and "sticks" of higher costs and/or greater inconvenience or travel times for single-occupancy vehicles, appear to have the greatest potential for modifying behavior and decreasing congestion on key roadways. For example, TCRP Project B-4, "Cost Effectiveness of TDM Programs", found the following: "The average reduction in vehicle trips among "successful" programs was 15.3%. Programs that provided enhanced alternatives, such as vanpools or shuttle buses, realized an 8.5% reduction in trips. Programs that focused on financial incentives and disincentives realized a 16.4% reduction of trips, and programs that combined enhanced alternatives with incentives/disincentives for their use, realized a 24.5% reduction in vehicle trips." (Reference 7, below, p. 46, citing COMSIS, 1994). Subsequent studies and practical experience have validated the power of even token pricing signals to influence behavior, change traffic patterns, and reduce congestion. The effectiveness is greatly increased when coupled with transit improvements, HOV-3 lanes (including facilitation of informal carpools or "slugging", employer ride share incentives and/or parking disincentives, and all the other tools in the travel demand management (TDM) strategies "toolbox".

The key findings on empirical evidence for the effectiveness of congestion pricing are summarized in the tables below (reference numbers in parentheses), with brief annotated references describing each case and/or reference below the tables.

Table 1: Observed Changes in Traffic due to Congestion Pricing
 Percentage Shift in Person Trips Percentage Shift in Vehicle Trips
Shift from Peak to Off-Peak -14% (1: Seattle) -8% (1: Seattle – varied rates
-7% (3: NY - $1.50 fee)
Change in Peak (or all) Traffic on Tolled Facilities +18% (4: Mn I394) +54% (6: I-15 CA HOV to HOT)-7% (3: NY-$1.50 fee)
-7% (6: FL-$.25/$.50 discount offpk)
-13% (1: Seattle – varied rates)
-15% (7: NJ Tpk)
-20% to 24% (6: Singapore – $2.50)
-21% to 30% (2: London – £5 then £8)
Increase in HOV Use +3% (4: Mn I-394)
+1-% (1: Seattle – work trips)
+25% (6: I-15 CA)
+33% to 40% (6: Singapore)
Increase in Transit Use +3% (1: Seattle – work trips)
+23% (4: Mn I-394)
+37% (2: London – £5 then £8)

The "outlier" in the above table (the 54% increase in vehicle trips on I-15 in California) represents the successful introduction of HOT lanes to a previously under-utilized HOV lane facility. When the enforcement on illegal SOVs increased with the introduction of tolls for SOVs, HOV use increased significantly, as shown. The introduction of tolls results in traffic reductions ranging from 7% to 30%, depending on the pricing and the breadth of the strategy. The highest values, in Singapore and London, represent comprehensive central city core area pricing programs, that were introduced with concurrent increases in transit and shared-ride alternatives. Most examples use electronic tolling. It is significant that decreases in vehicle trips can result in increases in person-trips, both through improved through-put based on reduced congestion, and on improved efficiency based on the balance between SOV and HOV.

Some empirical studies have evaluated changes in traffic with corresponding changes in congestion. Results vary greatly, due to the existing conditions and the methods used to evaluate congestion. However, as demonstrated in the Section 2 Traffic Analysis above, modest changes in volumes can generate significant decreases in delay.

Location Traffic Change Congestion Change
London (2: £5 then £8) -20% -30% (2003-2004)
Stockholm (5: variable fee) -15% -50%
Singapore (6) -20% to 24% -30% (avg. speeds up)

  1. Puget Sound Regional Council, “Traffic Choices Study – Summary Report”, April, 2008. Report on 18-month study of 275 individual drivers who participated in a voluntary GPS-monitored tolling study. As this was a sample, drivers did not have the option to pay a toll to drive on uncongested major roads. Reported results are from page 12 of the study, and extrapolated from Tables 2 and 3 on page 23 of the study.
  2. Transport for London, “Central London Congestion Charging: Impacts Monitoring - Fifth Annual Report, July 2007.” London introduced congestion pricing for its central section in 2003, along with improved bus services, and has been monitoring and adjusting the system since that time. In the first year congestion was reduced by 30% (p.35). In 2006, traffic with four or more wheels was down 21 percent from 2002, and vehicles that could potentially be charged was down 30% (p.19); nevertheless in 2006 congestion increased significantly from earlier years (but is still 8% lower than 2002), largely due to roadway construction, changes in signals and other factors (p. 48-53). This is being studied intensively to work out problems; for example, bus service times and delays increased, even though signal priority schemes for buses were in effect (p.72). (
  3. Breakthrough Technologies Institute and Environmental Defense, “Changing Lanes – Linking Bus Rapid Transit and High Occupancy Toll Networks in Northern Virginia”, September, 2005. The purpose of the study is clear from its title. It included a review of previous experiences. Cited here is the example from the Port Authority of New York-New Jersey that introduced time-of-day tolls on Hudson River bridges and tunnels and Staten Island bridges (7% peak hour traffic reduction- see p. 13).
  4. Turnbull, Katherine, “High-Occupancy Toll (HOT) Lanes and Public Transportation”, TRB 2008 Annual Meeting CD-ROM. Ms. Turnbull examines the introduction of Single-Occupant Vehicles (SOVs) paying tolls onto High-Occupancy Vehicle (HOV) lanes, examining the impacts on public transportation services and ridership levels. The three cases explore bus services operating on the HOV/HOT lanes on I-15 in San Diego, I-394 in Minneapolis, and I-25 in Denver. In San Diego, HOVs account for approximately 75 to 78 percent of total vehicle volumes, with FasTrak™ users accounting for most of the remainder. Bus riders comprise about 10 to 11 percent of daily users, mostly for reverse commutes, and taking advantage of service that was introduced and partially funded by the new tolls. Denver’s I-25 has from 68% to 72% of its lanes occupied by HOV, with 28%to 32% paying tolls. Denver instituted a performance monitoring system, using GPS to track the on-time performance of buses. They have been able to use this information to fine-tune signals near exit ramps and other potential problems on the route for buses as well as other vehicles.
    The Minneapolis HOT/HOV lane deliberately improved transit services when it opened the HOV lanes as HOT lanes. Although carpool and vanpool riders declined by about 10%, bus ridership increased by over 22%, for a 3% increase in total HOV use, and an 18% increase in people moved along the corridor. Figure to right derived from study.

Bar chart depicting the number of trips on Minneapolis I-394 during the first quarter of 2005 and the first quarter of 2007.

  1. Replogle, Michael, "Is Congestion Pricing Ready for Prime Time?", Planning, Vol. 74, No. 5, May 2008, pp. 6-11.
  2. TCRP Report 95, Chapter 14, "Road Value Pricing- Traveler Response to Transportation System Changes" – This comprehensive study completed in 2003 includes detailed analysis of major road, city center, and other US and international experience in congestion pricing strategies, including Singapore's ALS experience beginning in 1975 and the United Kingdom's experience in Durham (beginning in 2000) and London (beginning in 2003). Florida reference page 14-16. I-15 California reference p. 14-56, Table 14-13, Oct. 1996 v. Oct. 1999. Singapore reference p. 14-10 and 14-53, extrapolating between mode shares from Table 14-2 and the traffic volumes shown on page 14-53. Pre-ALS volume is estimated at 308,500 based on the statement on p. 14-10 that inbound traffic during restricted hours has been capped at below 70% of 1975 pre-ALS volumes.
  3. FHWA, "Mitigating Traffic Congestion: The Role of Demand-Side Strategies," 2004, This study provides a few examples of congestion pricing, and is also an accessible and valuable reference with "fast facts" and case studies on a wide range of Transportation Demand Management (TDM) strategies that will enhance the effectiveness of congestion pricing efforts. NJ Reference p. 22 "Fast Facts".

What Does This All Mean?

Traffic congestion and delay have become an endemic component of commuting life in the Washington DC region. To many, the unpredictability of travel time is even more annoying- one day a 10 mile trip may require 20 minutes, and the next day 45 minutes. Because the system is so near capacity, and exceeding capacity in some areas, a minor incident or a rainstorm causes major breakdowns and systemic delays. In this paper we have shown that there is a way to restore reliability and predictability to our highway system, without spending billions on new lanes of traffic.

In Section 2 we established that a 10 to 14 percent decrease in traffic on congested freeways will reduce delay by approximately 75 to 80 percent. In Section 3 we established that from 7 to 9 percent of the longer trips in personal vehicles during peak periods are discretionary. In Section 4 we have established that modest pricing signals for private vehicles can reduce traffic enough to significantly reduce congestion and save time for all drivers, while at the same time increasing the "people-carrying capacity" of the roadway, by increasing the use of carpools, vanpools and transit. It therefore appears feasible to restore and maintain free-flow on the freeways in the Metropolitan Washington area, without adding capacity, by applying congestion pricing to the major facilities, and at the same time increasing transit, carpool and vanpool programs. The combination of diverting most discretionary trips to other times and diverting an additional five to ten percent of personal vehicle work trips to HOV modes should achieve the needed 10 to 14 percent overall decrease in traffic needed to achieve major reductions in delay.[9]

Tolls are typically set to match the value of the reduction in delay. Some of the most effective programs vary the tolls by as little as 15 minute increments based on volumes and demand, in order to keep the facility operating smoothly. Others have a fixed rate for the "peak of the peak" (usually the peak hour), with a discount for the "shoulders" of the peak, and no charge in off-peak hours. Most systems use transponders (similar to the E-Z Pass in use throughout much of the East Coast) with cameras to identify violators, designed to collect information at high speeds.

Going forward with such a system would require political will, as well as phased implementation and experimentation to identify workable technologies and appropriate rates to achieve the desired result. However, based on the cited examples, it could be expected that a charge of approximately $.15 per mile (or $1.50 for the 10 mile freeway segments in the analysis) would be a reasonable starting point, comparable to the value of time saved. It is expected that registered car pools (HOV 3 or more, preferably), vanpools and transit would not be charged. Rates could be adjusted up or down, to ensure the roadway is used to near-maximum capacity, without exceeding capacity to the point of breaking down and failing. Revenues collected could be used to improve HOV alternatives as well as maintain the roadways, address choke points and bottlenecks, and improve alternate routes. Finally, travelers in the region would travel in confidence, knowing that they can reliably predict their travel time on a daily basis.

[8] The recent US surge in fuel prices resulted in significant drops in total driving. In Maryland and Virginia, for example, prices increased by approximately 25 percent over the prior year for the months of March through June, (average prices per and miles driven on urban arterials decreased by a little less than 3 percent over the same period, (per FHWA reports). Washington DC miles actually increased slightly. However, an increase in fuel prices or parking will not necessarily induce a shift from a congested roadway, or to a less-congested time period.

[9] Caution must be used in developing the program and publicizing and enforcing the HOV restrictions. In many cases discretionary trips, such as social and recreational trips, involve multiple people traveling together, who may consider themselves "HOV". Thus it must be made clear that the HOV refers to registered carpools or vanpools for work trips, to avoid the unintended consequence of facilitating discretionary group travel.

Adobe icon. You will need the Adobe Reader to view the PDFs on this page.