Transportation demand management (TDM) refers to a set of strategies aimed at reducing the demand for roadway travel, particularly in single occupancy vehicles. These strategies address a wide range of externalities associated with driving, including congestion, poor air quality, less livable communities, reduced public health, dependence on oil, reduced environmental health, and climate change and GHG emissions. Some TDM strategies are designed to reduce total travel demand, while others are designed to reduce peak period demand, which may disproportionately contribute to these externalities.
This review covers the following eight TDM strategies:
These strategies reduce demand through either mandatory or voluntary mechanisms. The mandatory programs reviewed discourage driving by increasing the cost of driving, as measured in money, time, or other costs. Road pricing programs charge drivers fees according to their use of the roadway, and may charge higher fees during peak periods in particular (called congestion pricing). Parking pricing charges drivers fees for parking their cars, while parking management reduces the availability of parking spaces.
Other strategies convert the fixed costs of driving in a personal vehicle into variable costs, so that the per-trip or per-mile costs are higher. As a result of the variability of trip cost, drivers tend to make fewer trips overall and VMT declines. Car sharing is a model in which participants pay to rent vehicles on a per-trip basis, and may forego owning their own vehicles. Under pay-as-you-drive (PAYD) insurance programs, drivers' premiums vary according to the miles they drive. In both cases, the total costs of driving can be less than they would be under the fixed-cost models. These programs generally benefit those who already drive less because they save money by paying the variable rather than the fixed costs.
TDM strategies may also make alternatives to SOV driving less expensive and more feasible. Ridesharing-meaning that more than one person travels in the vehicle-can be made more attractive by services that match drivers with passengers, provide benefits for ridesharing such as preferred parking, or operate ride sharing vehicles (e.g., corporate vanpools). High occupancy vehicle (HOV) lanes may further incentivize ridesharing by enabling ride sharers to avoid costly congestion or tolls. Transit incentives expressly reduce the cost of transit with fare passes and pre-tax payment programs, while transit improvements can increase the availability, efficiency, convenience, and comfort of transit.
Finally, strategies may reduce the need for mobility. Agencies may encourage or incentivize telework-working from home or a nearby, off-worksite location-to reduce the number or distance of commute trips.
TDM became an important concept in transportation in the 1970s in response to the oil crises of the decade. While a number of strategies were implemented around the country at that time, most prominently ridesharing, other strategies such as car sharing have been adopted more recently. Much research has been devoted to TDM in the decades since. As noted above, TDM strategies address many externalities simultaneously, including key concerns in the 1970s of oil dependence, congestion, and air quality. These are co-benefits of all TDM strategies.
The TDM strategies described above collectively reflect a "carrot-and-stick" approach: road pricing and parking pricing and management can discourage SOV driving (the "sticks") while ridesharing, transit incentives, transit improvements, and telework make alternatives to SOV more attractive (the "carrots"). Road and parking fees can be very effective in reducing demand, and they also generate revenue for transportation agencies. However, they are socially and economically controversial because they add to household and business expenses and the distribution of these expenses may be inequitable. These concerns are particularly important in contexts where drivers have few alternatives to SOV driving (e.g., if walking and biking are impractical or unsafe and transit availability is limited).
Conversely, making alternatives to SOV driving less expensive is typically socially acceptable because the use of those alternatives is voluntary and does not cost those who choose to drive. In part for these reasons, voluntary TDM strategies alone may only have a small effect on GHG emissions. Importantly, these TDM strategies, and car sharing and PAYD, are also vulnerable to induced demand. When some people reduce their SOV trips, the new space on roadways makes driving less expensive (i.e., in terms of travel time), and people who previously did not make those trips or used other modes may be induced to drive. This induced demand could potentially negate some of the reductions that were initially created by these strategies.
This suggests that "carrot" and "stick" TDM strategies may be much more effective in reducing GHGs when implemented together than when either is implemented alone (i.e., they interact positively). The "stick" strategies are more effective at encouraging people to reduce SOV driving and are generally not vulnerable to induced demand because they increase the cost of driving for everyone. Simultaneously, the "carrot" strategies may provide viable alternatives to SOV driving.
Two TDM strategies aimed at encouraging alternatives to SOV driving may involve manufacturing and construction processes that produce significant GHGs. These are (1) transit improvements that involve new vehicles, new infrastructure, or increased levels of service, and (2) the creation of new capacity for HOV lanes. The GHGs produced by these processes must be considered as part of the life-cycle GHG analysis to understand the net effects of these strategies. Substantial increases in ridership or ride sharing may be necessary to produce a net reduction in emissions and these are frequently difficult to achieve, particularly in decentralized areas.
As this suggests, TDM strategies and land use patterns are closely related in that compact land use is associated with lower VMT and higher incidence of non-motorized transport, ridesharing, car sharing, and transit use. In some cases, compact land use makes TDM strategies more effective, while, in other cases, TDM strategies (e.g., transit improvements) may encourage compact land use, like near new transit stops. As with land use, there is uncertainty about cause and effect: it is unclear whether people who participate in TDM programs do so because they prefer to drive less, or whether the TDM program encourages people to drive less. For example, does car sharing cause its members to drive less, or do people who already prefer to drive less choose to participate in these programs? While both phenomena are likely at work, this self-selection may limit the effectiveness of TDM programs among individuals or in areas where driving is strongly preferred.
Lastly, TDM strategies have the most significant effect on GHGs when the emissions from driving are high. For example, reducing the use of a vehicle with low fuel economy (and thus high emissions) has a greater effect than reducing the use of a vehicle with high fuel economy (and thus low emissions). This implies that TDM strategies have a reduced absolute effect when implemented with strategies that seek to reduce emissions from driving, such as fuel economy standards, fuel improvements, and transportation system improvements. All of these strategies are important in combating climate change, but their combined effect will be less than the sum of their individual effects.
Transportation system improvement strategies and vehicle strategies may also reduce the effectiveness of some TDM strategies in another way. Improving the transportation system and improving fuel economy reduces the cost of driving, and this may induce demand and counteract the effects of transit benefits and other TDM strategies that encourage the use of alternative modes. This can be counteracted with road pricing and fuel taxes, which make up for this decrease in cost. Fuel taxes, in particular, interact positively with all TDM strategies because they make driving more expensive and reduce VMT.
Policy: Economic theory suggests that driving is underpriced in that current costs do not cover its significant externalities and it is thus "overconsumed." Road pricing is a market-based strategy that internalizes the costs of these externalities and facilitates reductions in total VMT or driving during peak congestion periods. Road pricing includes existing mechanisms such as toll roads, cordon pricing, and proposed approaches such as VMT charges.
Emissions Benefits and Costs: GHG effects vary depending on the form of road pricing employed and the extent of the charges. Where road pricing has been employed in practice, before-and-after studies have found that VMT was reduced by between 2 and 10%, and, where measured, GHGs declined by 2 to 6%. Some modeling-based studies have found much higher reductions, but only with very high per-mile charges that are well above the range of pricing that is normally considered in planning studies.
Implementation Concerns: While road pricing has been implemented in various forms both in other countries and in several corridors in the U.S., it remains controversial because of equity concerns, resistance to new taxes and fees, sometimes limited transit options, and privacy concerns. Simultaneously, road pricing can serve as a major revenue source, and some states are pursuing road pricing to counteract the effects of declining fuel tax revenues.
According to economic theory, people overconsume goods that are underpriced. In transportation, driving is underpriced in that current costs do not cover the externalities of driving (such as pollution, GHG emissions, crashes, and congestion). As a result, driving is "overconsumed" and these externalities have become significant. Road pricing, like other market-based strategies, seeks to correct this imbalance by internalizing some of the cost of these externalities and reducing total VMT or reducing driving during periods of peak congestion. In road pricing, drivers are charged fees based on their consumption of the roadway (in contrast to fuel taxes, which charge based on the consumption of fuel). The most well-established form of road pricing in the US is toll roads, in which drivers are charged a flat fee for traveling over some section of the road, but other more sophisticated pricing schemes have also been developed.
Road pricing can be implemented in many forms. Because the externalities of driving (like congestion) can vary based on the particular roadway used, the time of day, and other factors, a perfect road pricing system would include a real-time charge on every mile driven based on the cost of externalities produced at that particular moment. For now, this remains technologically out of reach. However, in addition to toll roads, there are three newer types of road pricing:
In many cases, these policies are aimed at reducing congestion by enacting higher fees during peak hours; this is known as congestion pricing.
In the US, only toll lanes and HOT lanes have been implemented to date, and these have been undertaken by local or state agencies such as State DOTs. Studies of HOT lanes and cordon tolls have also been undertaken by local governments and MPOs, and pilot programs of distance-based pricing have been undertaken by several state DOTs. The federal government would play a significant role alongside state agencies in creating a national system of distance-based fees, which several panels have called for (NSTIFC, 2009; NSTPRS, 2007).
Road pricing can target all drivers, but HOT lanes and cordon tolls typically affect only those who live or work in the targeted areas, while distance-based fees apply to most or all drivers. There may be separate rate structures for trucks versus passenger vehicles.
Road pricing has been widely implemented and studied in various forms around the world. The literature on HOT lanes is mostly based on the U.S. experience, while cordon studies draw from experience in London, Stockholm, and Singapore. Several European countries use distance-based pricing for trucks, but it has yet to be implemented anywhere for all vehicles (though several pilot projects are underway). In addition to real-world studies, much research uses modeling to assess the impacts of proposed road pricing systems
Generally, the literature found evidence that people drive less, particularly with cordon systems, but this evidence comes from cities with extensive transit networks; the experience in more auto-oriented regions may be different. Although much of the literature focuses on congestion effects, some studies include GHG effects and this sourcebook focuses on those in particular. The discussion below provides results from three real-world examples of cordon and corridor pricing, as well as results from models. Because HOT lanes have largely been constructed next to free capacity, drivers who do not wish to pay still have an option of driving with no charge in the same corridor. Perhaps for this reason, there is little research on the emissions impacts of HOT lanes.
In 2003, London implemented a £5 congestion charge to enter the central city during weekday business hours (from 7 pm to 6:30 pm). According to annual reports from Transport for London, the number of private cars entering London's cordon fell by almost one-third when the congestion-charging system was launched, and congestion-as measured by travel time delays-was reduced by 30%. It was additionally estimated that CO2 emissions in the first year within the charging zone fell by 19% (TfL, 2004). When the fee was raised from £5 to £8 (from US$8.10 to $13) in 2005, CO2 emissions decreased by an additional 5% (TfL, 2006). Interestingly, this significant decrease in vehicle trips did not correspond to a significant decrease in person trips-suggesting that mobility remained high. As one commentary put it, after the charge was introduced, there were "60,000 fewer car trips coming into the zone, [but] only 4,000 people no longer travelling to central London" (Dix, 2004). The decrease in private cars was to some extent offset by greater use of taxis, buses, and bicycles. When the charging zone was further extended westward in 2007, the reduction in CO2 was estimated at 6.5% (TfL, 2008).
The decreases in congestion and CO2 emissions have not consistently held over time. In 2006 and 2007, measurements of "excess delay"-the amount of additional time it takes to drive a fixed distance when roads are congested versus when traffic is free-flowing-show that congestion in London's charging zone rose slightly during those two years, back to the level in 2003 before the charge went into effect. This occurred despite the fact that the number of vehicles entering the zone remained lower than in 2003. This seems to have happened because conditions on the road network within the charging zone had deteriorated due to a number of construction projects. Based on this evidence, it seems that charging nevertheless reduced congestion from the levels it might have otherwise reached under those conditions. Regarding CO2 emissions, a later report noted, "These attributable reductions have diminished as congestion levels increased from 2006 onwards but have long since been overtaken in magnitude by the beneficial impact on year-on-year improvements to the general emissions performance of the vehicle fleet" (TfL, 2008).
In 2006, Stockholm conducted a seven-month trial of a cordon system that charged between 10 and 20 kroner ($1.46 to $2.95 in 2009 USD), depending on the time of day. Studies of the trial show that the number of vehicles crossing the cordon declined between 22% and 28% from that same time period in the previous year. When the trial ended, the volume of vehicle entries rose immediately to just below 2005 levels. Emissions fell by 41,000 MTCO2 per year, or 2.7%, across greater Stockholm (City of Stockholm, 2006). A later analysis, based on data from when the congestion charging system had been re-implemented beginning in 2007, found that emissions again decreased by 2.7%, or 42,500 MTCO2, per year in Stockholm County (Eliasson, 2009).
Singapore's road pricing system is one of the oldest and most rigorous. It has been in place for more than 35 years and charges currently range from S$0 to S$2.50 (US$1.78 in 2009 USD), depending on the roadway and the time of day. When initially introduced, traffic on roads with charges decreased by 45% during the morning peak period, exceeding the city's 25 to 35% target. Over the course of the system's long history, congestion has not returned on those roads (Goh, 2002). Since 1998, the Land Transport Authority (LTA) has been able to maintain target speeds of 45 km/h (28 mph) on expressways and 20 km/h (12.5 mph) on arterial roads by adjusting charges as needed. This is despite the fact that the city has grown, the number of registered vehicles has increased, and no new road capacity has been added (Menon and Chin, 2004). People who no longer drive during peak hours have either moved their trips outside charging times or switched to other modes of transportation.
For comparison, in 1999, per capita emissions from road transportation in Singapore were 0.89 MTCO2 while they were 5.37 MTCO2 in the US. Researchers have also estimated that, at least in 1990, fuel consumption in Singapore would have been 50% higher without vehicle restraint policies (in addition to road pricing, purchase of vehicles is expensive and requires a permit). It is not clear what proportion of the 50% difference is attributable to road pricing as opposed to other vehicle restraint policies (Ang and Tan, 2001).
Several studies have also used models to assess the effects of pricing schemes in different cities.
The public cost of implementing road pricing systems (borne directly by transportation agencies) varies widely depending on the technology and the extent of the system, and can be several hundred million dollars as discussed below in the section on agency costs. Given the variation in costs and emissions reductions, a unit cost cannot be determined. Importantly, the road pricing systems that have been implemented have resulted in net revenues to agencies. (In practice, it is not likely that an agency would undertake a pricing program with costs that exceed the revenues to be generated unless the program is supported, for instance, by the federal government, through subsidies.)
Many factors affect the estimated emissions impacts of road pricing. As the preceding discussion shows, measured or estimated reductions range from nearly zero to over 25%. One critical factor is the form of road pricing used. Distance-based fares generally produce greater reductions than cordon tolls or HOT lanes because the former are applied widely and at all times, while the latter are limited to certain areas or times. A second factor is the charge: the greater the charge, the greater the impact. Of course, the higher the charge, the higher the likelihood that it will be difficult to address social and economic inequities. As Safirova et al. (2008) note in their abstract, "We also find that full social cost pricing requires very high toll levels and therefore is bound to be controversial."
One study also found that the "optimal toll" varies depending on the model (sometimes by several percentage points), which in turn affects the outcome on emissions (Shepherd, 2008).
FHWA has a Congestion Pricing Primer series available at http://ops.fhwa.dot.gov/tolling_pricing/resources.htm. The seven volumes in the series present issues in congestion pricing to decision-makers, including definitions, benefits, technologies, and case studies.
It is not possible to develop a precise range of cost estimates since they vary so widely depending on the location, technology, and type of implementation. This section provides several real-world cost examples (all provided in 2009 USD). London's cordon charge, implemented with a technology that reads license plates, cost approximately $378 million to implement (ECMT, 2006) and has annual operating costs of $244 million (TfL, 2008). Stockholm's cordon charge, using both automated plate reading and a short-range communications system, was $256 million to set up (ECMT, 2006) and costs $33 million annually to operate (Eliasson, 2007). In Singapore, capital costs for electronic road pricing were $151 million (ECMT, 2006) and operating costs are $11 million (Menon and Chin, 2004). Note, however, that all of these systems raise annual revenue; operating costs as a percentage of gross revenues are 48% (London), 25% (Stockholm), and 7% (Singapore) (ECMT, 2006). This wide variation is a result of the different technologies employed.
Only certain types of road pricing systems have been adopted in the United States, namely toll roads and HOT lanes. A single entity, such as a state DOT or a transportation authority generally implements these. A wider form of road pricing, such as distance-based charges, would require cooperation among a broader range of agencies. There may be challenges with technology compatibility (for example, can all transponders be used with all gantries?), working with vendors, determining which road pricing system best fits policy goals, and ensuring that payment systems function well and are enforceable.
Road pricing is controversial. Several high-profile proposals have been either voted down by the public (see Greco and McQuaid  about Edinburgh) or turned down by elected officials (see Confessore  about New York City). The idea of paying for trips that are now free (as with a cordon toll) often raises equity concerns, but whether regressive effects occur in practice depends upon the policy and context. The ability of drivers to shift modes depends on the availability of transit services and whether land use patterns support non-motorized trips. Places that have implemented cordon tolls have fairly high levels of transit service, compared to many regions in the U.S. While there are other ways to shift driving habits-such as changing the time of travel or carpooling-in regions with road pricing, many trips have shifted to transit.
Proposals to implement distance-based fees would likely involve in-vehicle equipment, which also raises privacy concerns (Sorensen et al., 2009). Privacy can be addressed through technologies that record total fees but delete the actual locations traveled, or through the creation of anonymous accounts. However, given that other types of transponder records can be provided to law enforcement, it may be difficult to convince privacy advocates that these records can be kept confidential.
Outside of conventional tolls (which have been part of the roadway network for decades), the road pricing system that has met with the most popular success in the US is HOT lanes. Public opinion polling in cities with HOT lanes finds fairly widespread support, generally 60 to 70% (Douma, 2005). This acceptance is probably due to the fact that no travel options have been taken away; that is, drivers can still travel the same corridor free of charge (unlike for example cordon systems, in which free options are not available).
A key cost-that borne by drivers-cannot be generalized across program schemes and regional contexts. Setting the toll level has an enormous impact on those costs, both because they determine what drivers pay and because the levels affect driver behavior-that is, the cost may encourage some drivers to forego trips, drive at different times of day, change routes, or change modes. In addition, if drivers avoid certain trips because of the expense, this can impose other types of societal costs (for example, a driver may choose not to visit a friend, or shop in a certain area, or volunteer at a hospital). While many researchers have developed models of these costs, they must be calibrated to specific proposals, charges, and locations.
Some costs may be borne by private businesses; two examples are provided here. First, businesses may be required to assume some of the costs of the equipment that would facilitate road pricing. For example, if gas stations were required to install devices that permit pay-at-the-pump collections, that cost would be borne by the station owners (Sorensen et al., 2009). Second, businesses may experience unintended consequences of road pricing. For example, if shoppers switch from shopping within a cordon-charging area to stores on the outskirts, retail businesses inside the cordon may suffer losses in sales. However, one study examined the several road pricing schemes in Europe and found little evidence of adverse economic effects, even among businesses within cordon rings (May et al., 2010).
Cordon congestion pricing was implemented in London in 2003 and in Stockholm in 2007 (after a trial period in 2006). Road pricing has been in use in Singapore since 1975, and has gone through several changes in form. HOT lanes are in use in Orange County (State Route 91), San Diego (I-15), Minneapolis (I-394), Denver (I-25/US-36), Salt Lake City (I-15), and Houston (Katy Freeway and Northwest Freeway). Distance-based fees for trucks have been adopted in Germany, the Czech Republic, Slovakia, Austria, and Switzerland.
As pricing programs are implemented around the world there is a need to document the results, assess their cost effectiveness, analyze the challenges they face, and develop a database so that future programs can be based upon best practices.
Ang, B. W., and Tan, K. C. (2001). Why Singapore's land transportation energy consumption is relatively low. Natural Resources Forum, 25(2), 135-146.
City of Stockholm. (2006). Facts and results from the Stockholm trial: final version.
Confessore, N. (2008, April 8). $8 traffic fee for Manhattan gets nowhere. New York Times.
Dix, M. (2004). Central London congestion charging. Paper presented at the ECMT Transport for London - London User Charges Conference.
Douma, F., Zmud, J., and Patterson, T. D. (2005). Pricing comes to Minnesota: Attitudinal evaluation of I-394 HOT lane project. Minneapolis: University of Minnesota, Hubert H. Humphrey Institute of Public Affairs.
Eliasson, J. (2009). A cost-benefit analysis of the Stockholm congestion charging system. Transportation Research Part A: Policy and Practice, 43(4), 468-480.
European Conference of Ministers of Transport. (2006). Road charging systems - technology choice and cost effectiveness. Paper presented at the Conference on Road Charging Systems: Technology Choice and Cost Effectiveness, Paris.
Goh, M. (2002). Congestion management and electronic road pricing in Singapore. Journal of Transport Geography, 10, 29-38.
Grieco, M., and McQuaid, R. (2005). Edinburgh and the politics of congestion charging: Negotiating road user charging with affected publics. Transport Policy, 12(5), 475-476.
May, A.D., Koh, A., Blackledge, D., and Fioretto, M. (2010). "Overcoming the barriers to implementing urban road user charging schemes," European Transportation Research Review, 2, 53-68.
Menon, A. P. G., and Chin, K.-K. (2004). ERP in Singapore-what's been learnt from five years of operation? Traffic Engineering and Control 45(2), 62-65.
Metropolitan Transportation Commission (MTC) (2008). Bay Area High-Occupancy Toll (HOT) Network Study: December 2008 Update.
Mitchell, G., Namdeo, A., and Milne, D. (2005). The air quality impact of cordon and distance based road user charging: An empirical study of Leeds, UK. Atmospheric environment, 39(33), 6231-6242.
National Surface Transportation Infrastructure Financing Commission (NSTIFC). (2009). Paying our way: A new framework for transportation finance.
National Surface Transportation Policy and Revenue Study Commission (NSTPRS). (2007). Transportation for tomorrow: Report of the national surface transportation policy and revenue study commission.
Rich, J., Nielsen, O. A., and Nielsen, U. (2006). Socio-economic assessment of road pricing systems: Results from two projects in the Copenhagen region. Paper presented at the Association for European Transport.
Rich, J., and Nielsen, O. A. (2007). A socio-economic assessment of proposed road user charging schemes in Copenhagen. Transport Policy, 14(4), 330-345.
Rich, J., and Nielsen, O. A. (2008). External effects and road charging. In C. Jensen-Butler, B. Sloth, M. M. Larsen, B. Madsen and O. A. Nielsen (Eds.), Road pricing, the economy and the environment. Berlin Heidelberg: Springer.
Safirova, E. A., Houde, S., and Harrington, W. (2008). Marginal social cost pricing on a transportation network: Comparison of second-best policies (Working Paper Entry: 200803).
Santos, G., Rojey, L., and Newbery, D. (2000). The environmental benefits from road pricing (Working Paper Entry: 200404).
Shepherd, S. P. (2008). The effect of complex models of externalities on estimated optimal tolls. Transportation, 35(4), 559-577.
Sorensen, P., Ecola, L., Wachs, M., Donath, M., Munnich, L., and Serian, B. (2009). Implementable strategies for shifting to direct usage-based charges for transportation funding. Washington, DC: Transportation Research Board, NCHRP Web-Only Document 143.
Transport for London. (2004). Impacts monitoring: Second annual report, April 2004.
Transport for London. (2006). Impacts monitoring: Fourth annual report, June 2006.
Transport for London. (2008). Impacts monitoring: Sixth annual report, July 2008.
Policy: How parking is supplied, managed, and paid for varies in the US, but typically localities require developers to provide a minimum number of spaces per development type, in accordance with a formula related to the size of the development. The costs for those spaces are most often "bundled" with other development costs such that parking appears to be supplied free to drivers. This encourages driving. However, a suite of new ideas and technologies has emerged to change this paradigm, such as reducing the amount of parking and making parking more expensive. These strategies could reduce the number of driving trips and/or encourage the use of alternative modes, and many of these strategies are actually more equitable than those currently used most frequently.
Emissions Benefits and Costs: These strategies can reduce SOV trips, but they vary widely in how they are implemented, so a single range of effects cannot be generalized from the literature. In terms of costs, charging drivers for parking raises money for transportation agencies but will also make driving more expensive.
Implementation Concerns: Acceptability would likely vary by region and urban/suburban split, since many areas already have limited parking and paid parking. Both public and private parking managers might be involved, and zoning codes that govern parking requirements may need to be changed.
Plentiful and free parking encourages driving. Indeed, in some cases free parking can be the main factor in the choice to drive: one study found that monthly parking charges explained up to 80% of the difference in the number of employees who drive alone to work (Dowling, Feltham, and Wyco, 1991). Moreover, virtually all vehicle trips in the U.S. have free parking on at least one end (Shoup, 2005). The goal of many parking management and parking pricing strategies is to reduce vehicle trips by making parking less available, more expensive, or both, on the assumption that people will make fewer trips, change modes, or carpool and thereby reduce GHGs.
Importantly, parking management strategies that reduce the number of spaces could create some GHGs if drivers spend significant time and fuel searching for scarce free or underpriced parking. This can be addressed in part by "smart parking" technologies, which provide real-time information about parking availability to reduce the search for parking. Simultaneously, by making parking easier, smart parking reduces some of the cost of driving that parking management and pricing strategies create. The unintended consequences of both parking management and smart parking must be balanced carefully to produce a net reduction in GHGs.
Parking management and parking pricing are closely related strategies. Pricing strategies charge users or owners for parking. Parking management strategies use some combination of approaches to change the amount of available parking or to require multiple users to share parking. Parking is often regulated through zoning codes that specify the minimum number of spaces that must be provided, so parking management efforts might decrease the minimum requirements, set maximum limits on parking spaces, or lower the number of parking spaces required in areas with mixed uses or near transit stations. Shared parking, on the other hand, might require that an office building make its parking spaces available in the evening to restaurant patrons.
Emerging policy ideas include "performance-managed parking" in which the availability of unoccupied spaces is maintained at 15% during peak periods through pricing, and "smart parking," in which technologies provide drivers real-time information on the availability of spaces in a particular location (whether on- or off-street).
Collectively, parking policies are typically implemented locally since cities manage their own on-street parking and set requirements for off-street parking. However, regional policies are possible.
Parking management and pricing can be directed at the business community or individuals. Developers are the target of policies to reduce the amount of parking provided through changes in zoning or parking maximums. Other businesses may be required or encouraged to shift from free to paid parking (for example, through parking cash-out for employees), or to share parking among multiple users. Individuals can also be targeted for paid parking (for example, by charging for public garage or on-street parking). Policies to introduce paid parking for individual drivers are much more common in the U.S. than those requiring businesses to manage employee parking.
There is substantial evidence from empirical studies of U.S. parking scenarios that charging for parking reduces single-occupancy vehicle (SOV) trips. Most such research focuses on commuter (work trip) parking. Studies of areas with newly-introduced paid parking (or comparisons between areas with free and paid parking) generally show that paid parking results in lower SOV mode shares, increased use of other modes, and reductions in vehicle trips. The impact on GHG depends on the number of people who stop driving alone, the emissions from the original trips (calculated based on trip length and fuel economy) and the emissions from the alternatives (whether the trip is foregone or made using another mode). Most research reports on changes in mode share and VMT and does not report on GHG effects.
Research in this area has been limited since free parking is so common and most of the "natural experiments" have been of workplace parking, so the impacts on other trips is less well understood. The elasticity of the demand for parking (that is, the change in behavior that results from a change in price) is not very high: estimates based on multiple studies have found an average of about -0.3, meaning that for every 10% increase in parking costs, the number of cars parked declines about 3% (Vaca et al., 2005). This is considered relatively inelastic, and on par with short-term elasticity for increases in fuel costs.
Empirical studies of workplace parking have found that the difference in SOV mode shares is generally on the order of 7 to 15 percentage points (see various studies cited in Vaca et al. (2005)). In one oft-cited study of cash-out parking, a system in which employers allow employees who previously received free parking to receive a cash payment to stop driving alone, the average SOV mode share fell from 77% to 65% (Shoup, 1997).
It is possible to estimate impacts of paid parking on a single workplace based on the number of employees, on the amount of the parking charge (since higher costs tend to produce greater responses), and on assumptions about their willingness and ability to take fewer trips (which depends, for example, on the availability of transit and the availability of other free parking). It could also be possible to estimate such impacts at other locations, such as shopping centers, entertainment and sports venues, and hotels. However, most of the data collected on responses to parking charges are based on employees (an important demographic given that work trips are made regularly and are, on average, longer than trips for most other purposes) who tend to be more sensitive to prices than drivers with other trip purposes (Vaca et al., 2005).
Few cities have instituted widespread paid parking. Where it has been done in conjunction with existing transit service, it has been fairly successful in reducing trips. In Perth's (Australia) "parking management area," all spaces are charged an annual fee, except in residential areas and areas with fewer than five spaces. Parking charges vary depending on whether the spaces are designated short- or long-stay. These fees were first imposed in 1999. From 1991 to 2001, the percentage of center city commuters driving to work (as drivers or passengers) declined from 66% to 58%, while the percentage commuting by train increased from 5% to 18%. Over the same period, employment in the area increased from 93,000 to 97,000, indicating that the decline in mode share was not due to jobs moving outside of Perth. A major new bus route was introduced during this period, which may have played a role in this shift (Sinclair Knight Merz, 2007).
One can use some simple assumptions about fuel economy and commute distances to determine the GHG reductions from the decline in vehicle mode share that Perth experienced. If there had been no decline in mode share from 1991 to 2001, then Perth would have had 64,000 vehicle commuters in 2001. Instead, the decline to 58% means that Perth had 56,000 vehicle commuters, a difference of approximately 8,000 vehicle commuters. If one assumes vehicle occupancy of 1.15 (roughly the average vehicle occupancy over the decade [Sinclair Knight Merz, 2007]), then this results in approximately 7,000 fewer vehicles traveling into and out of the city for work. If one further assumes 250 working days a year, a 20-mile round-trip commute distance and a fuel economy of 20 mpg, this translates to an annual reduction of 1.7 million gallons of gasoline and 17,000 MTCO2. Although it is not possible to attribute a specific portion of the change in mode share (or subsequent GHG reductions) to parking policies alone, this figure offers an approximate upper bound on the reductions from parking policies (i.e., assuming all reductions are from parking policies, and that the alternative modes produced no additional emissions).
Smart parking is a relatively new technology, so research is still underway. One study examined smart parking at a rail transit station in Oakland, CA at which drivers could reserve spaces in advance. The study found an average VMT reduction of 9.7 miles per month per participating driver. In this pilot program, the first of its kind in the U.S., drivers could reserve spaces either online or via telephone, and changeable message signs along highways leading into downtown alerted drivers that spaces were available at the transit station. The changeable message signs were not found to be a major factor in driver behavior; only 37% of drivers surveyed who used the smart parking spaces had even seen the signs, and only one-third of those said it influenced their decision. Study results were based on a survey of drivers who used the service at least once, so conclusions were self-reported and not observed. The survey did not ask whether any participants had started driving as a result of the pilot (Rodier et al., 2008). As smart parking is still being piloted rather than fully implemented, no other studies that quantified its impacts were identified.
A policy of charging for parking, which is not particularly expensive to implement compared to others, would likely result in net revenues to the implementing jurisdiction and costs to drivers. However, reductions cannot be generalized given that they are specific to policy implementation details.
The greatest unknown is the response to widespread parking charging. Most U.S. studies of the response to parking charges are based on small sample sizes, such as individual worksites or parking garages, and it is unclear how these estimates would "scale up" to an entire district, city, or region.
It is also difficult to estimate the impacts of other parking management strategies, such as changing zoning codes to allow developers to provide less parking or requiring businesses to share parking. These strategies are fairly new and few assessments of their effectiveness exist. In addition, some changes in the approach to parking can take years to produce measurable changes (for example, if the main strategy is changes to zoning to require less parking, but development slows down, then the overall parking stock would not change very quickly).
As noted above, while the public sector would incur some costs for implementing parking management and pricing (such as collection costs, signage, enforcement, and so forth), these policies would likely produce revenue that more than covers these costs. The Perth program, for example, generated AU$9.3 million (US$8.5 million) in revenue in one year (2006-07) (quoted in Sinclair Knight Merz, 2007).
Different parking policies are implemented at different levels of government. Parking ordinances are generally enacted at the municipal level, so changing minimum parking requirements would probably have to be adopted by a city council or similar body. Policies about charging for parking may raise concerns such as how to charge for parking (for example, a previously free lot may require additional infrastructure to allow the physical means of payment) and enforce payment. Smart parking strategies may require both new policies to be adopted as well as new equipment to be procured, installed, tested, and put into service.
There can also be opposition to paid parking from groups that fear the effects of "spill-over" parking, meaning that instead of utilizing paid parking, drivers will seek out free parking. Generally the concern is that drivers will take up spaces in neighborhoods, leaving residents with limited parking options. One way to mitigate this impact is to introduce some type of permit parking, so that only neighborhood residents can park long-term in the area. This would have to be coupled with aggressive parking enforcement to be effective. These concerns may also be alleviated with more widespread parking management and pricing, so that most or all of the spaces in an area are paid or restricted in some way.
Driver response may be very different depending on location, since drivers in urban areas have a wider array of travel options. They are also more accustomed to paying for parking than suburban drivers. For example, Vancouver, Canada implemented a regional parking fee that was repealed after two years in the face of continued opposition (Transport Canada, 2006). In addition, parking charges may be perceived as inequitable to low-income drivers, although this perception may not be borne out in reality.
In some cases, if businesses want to pass parking charges directly to employees or customers, this may require installing equipment or technology to facilitate charging, such as adding a payment booth to a parking facility that does not currently have one. This can be overcome using a parking cash-out scheme, where employees are charged for parking through payroll deduction, unless they choose not to park a vehicle.
Most cities have some paid parking, both private and public, although there does not appear to be any national database that collects this information. While some Australian cities, as well as Amsterdam, use area-wide pricing (as described above in the Perth example), no American cities do so. A relatively small number of cities use parking management techniques such as performance-based pricing, reductions of parking requirements in certain areas or near transit stations, or maximum parking requirements (see, for example, Knepper et al., 2007). Many municipalities are developing smart parking programs in neighborhoods, commercial centers, airports, and other areas. San Francisco's smart parking pilot program may be the most widespread use of these technologies in the United States at this time.
A closer look at examples of regional parking policies in Amsterdam and several other cities in Australia, while beyond the scope of this literature review, is likely to offer more data and information on the effects of these policies.
Bureau of Transportation Statistics (2009). National Transportation Statistics, 2009. U.S. Department of Transportation. Available online at: http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/national_transportation_statistics/index.html.
Cambridge Systematics (2005). Traffic Congestion and Reliability: Trends and Advanced Strategies for Congestion Mitigation, prepared for the Federal Highway Administration. Available online at: http://ops.fhwa.dot.gov/congestion_report/
Dowling, R., Feltham, D., and Wycko, W. (1991). Factors affecting transportation demand management program effectiveness at six San Francisco medical institutions. Transportation Research Record, 1321.
Hu, P. S., and Reuscher, T. R. (2004). Summary of travel trends: 2001 National Household Travel Survey Washington, DC: U. S. Department of Transportation, Federal Highway Administration.
Knepper, V., Wilbur Smith Associates, Michael R. Kodama Planning Consultants, Willson, R., KT Analytics Inc., Rick Williams Consulting, and CHS Consulting Group. (2007). Developing parking policies to support smart growth in local jurisdictions: Best practices. Oakland, CA: Metropolitan Transportation Commission.
Rodier, C. J., S. A. Shaheen, and C, Kemmerer. (2008, June). Smart Parking Management Field Test: A Bay Area Rapid Transit (BART) District Parking Demonstration; Final Report: California Partners for Advanced Transit and Highways Program, Institute of Transportation Studies, University of California at Berkeley, UCB-ITS-PRR-2008-5.
Shoup, D. (1997). Evaluating the effects of parking cash out: Eight case studies, final report. Sacramento, CA: California Air Resources Board Research Division.
Shoup, D. (2005). The high cost of free parking. Chicago: American Planning Association.
Sinclair Knight Merz. (2007). Review of Perth parking policy.
Transport Canada. (2006). Urban transportation showcase program: TransLink parking tax case study. Vancouver: Transport Canada.
Vaca, E., and Kuzmyak, J. R. (2005). Chapter 13 - parking pricing and fees. In TCRP report 95: Traveler response to transportation system changes. Washington, D.C.: Transportation Research Board.
Policy: Most miles driven in the United States are in privately owned vehicles. Because vehicle ownership entails many "sunk costs" (e.g., the purchase price, registration fees, insurance, maintenance, etc.), out-of-pocket costs tend to be low relative to other modes on a per-trip basis, making driving attractive. Car sharing seeks to convert these fixed costs to variable ones by promoting a model in which participants rent vehicles on an as-needed basis, and may forego owning their own vehicles. As a result of the variability of trip cost, drivers tend to make fewer trips overall and VMT declines. Additionally, the vehicles available in car sharing programs are often more fuel-efficient than the average privately owned vehicle, which also reduces GHGs.
Emissions Benefits and Costs: GHG emissions have declined among car sharing members as they both drive fewer miles and more efficient vehicles. Studies in the U.S. and Canada have found that emissions declined on average by between 0.8 and 1.2 MTCO2 annually per member, even after accounting for those members who drive more often because they did not previously own vehicles. No research has linked the cost of these programs to the public sector to adoption rates since most shared cars are managed by for-profit companies. Nevertheless, the public sector can play a role by providing subsidies, publicity, or parking spaces.
Implementation Concerns: While barriers to implementation are low, they may include resistance to converting public parking to parking reserved for car sharing. Importantly, car sharing has thus far been effective primarily in more compact neighborhoods or in areas with already limited parking (such as college campuses).
Car ownership entails many "sunk costs" that are fixed at the same rate regardless of the amount the vehicle is driven. In a car sharing organization, members rent vehicles by the hour or day. This differs from conventional rental cars in several ways: it is marketed to residents and businesses in a city, rather than visitors; it provides hourly rates, while most rental car firms charge by the day or week; it positions vehicles throughout an area so that members can walk to them in their neighborhoods; and it emphasizes quick booking when a vehicle is needed. For some, using a car sharing service may be less expensive overall than privately owning a vehicle. For others, it may offer mobility that they would otherwise not have.
Car sharing can reduce GHGs by reducing the number of trips. Research has shown that drivers make decisions regarding modes for a particular trip based on out-of-pocket costs that vary by trip (gas, tolls, and parking), meaning that many vehicle trips in personally owned vehicles appear quite inexpensive compared with alternatives such as transit (Steininger, Vogl, and Zettl, 1996). In car sharing, these costs are variable and incurred largely per-trip, so drivers are more likely to consider the total costs and make fewer trips overall. Importantly, these programs simultaneously create a way for people who do not own their own car to drive where otherwise they may have walked, used transit, or not taken the trip. Car sharing can also reduce emissions if the vehicles in the service have higher fuel economy than privately owned vehicles, or if members have the flexibility to choose the size of vehicle that meets their needs for each particular trip-meaning that large and less fuel efficient vehicles may be chosen only when needed.
Car sharing began in Europe and has spread to the U.S. in the past decade. Car sharing services are generally operated by commercial or non-profit entities. Members of a car sharing organization generally pay fixed fees to join and an annual membership fee, plus the hourly or daily rental fees. While car sharing organizations can have multiple locations, they tend to be most effective in high-density areas where many other trips can be served by transit or non-motorized transportation. College towns and urban university campuses have also been good markets because campuses are typically compact, students often use cars infrequently, and campus parking may be limited. Business programs in which employers join and provide car sharing as a benefit for their employees have also been growing.
Car sharing tends to be championed by regional or local agencies, such as MPOs, local governments, and/or transit agencies. Since car sharing is largely operated by private entities, the role of the public sector may include subsidies for program start-up costs, provision of parking spaces for the vehicles, tax incentives, encouraging or requiring private developers to include car sharing spaces in multi-family housing, and publicity.
Car sharing has been marketed to both individuals for personal travel and to businesses as a lower-cost alternative to maintaining a vehicle fleet and for employees who need access to vehicles during the work day. Various studies of total ownership costs report "break-even" points (at which the cost of car sharing equals the cost of car ownership) variously at 5,000 (Millard-Ball et al, 2005), 8,000 (Higginbotham, 2001), and 4,000 to 10,000 miles (Litman, 1999, Figure 2). Drivers who drive fewer miles than the break-even point would save money with car sharing and are potential car sharing candidates, while those who drive more are better off economically owning a vehicle and would not be good candidates.
Existing car sharing programs in the U.S., Canada, and Europe have been studied to assess their effectiveness at reducing VMT and emissions among users. Among all car sharing members, both emissions and VMT decline; this reflects both the previous car owners whose emissions and VMT fell substantially, and those who did not previously own cars and now drive more. It is also possible that some emissions reductions are due to changes in the fleet mix; that is, on average shared vehicles may have lower emissions profiles than privately owned vehicles. Millard-Ball et al. (2005) observed that car sharing fleets tend to have more alternatively fueled vehicles, newer vehicles, and smaller vehicles than the overall fleet. While fleets do contain larger vehicles for special purposes (a pick-up truck to haul furniture, for example), members can choose the vehicle most appropriate for their trip. The literature does not tell us how agencies' actions (e.g., provision of subsidies and parking spaces) affect the availability and use of car sharing programs, given that it depends more on density, transit, and land use.
Only a few studies have directly estimated the effects of car sharing on GHG emissions. In a recent study based on survey responses from over 6,200 car sharing members in North America, Martin and Shaheen (2010) found that on average, a household reduces its GHG by 0.84 MTCO2 per year after joining car sharing. This includes two kinds of reductions. First, it includes the "observed impact"-the observed difference between a household's actual VMT before and after joining car sharing. Two changes may be observed. Car sharing offers vehicles to people who previously did not have access, thus increasing their VMT and GHG emissions. This increase was observed in most households participating in car sharing, but the observed increase was small. The minority of households substitute a car-sharing vehicle for a personal vehicle, and significantly reduce their VMT and GHG emissions. The net effect is an overall reduction in VMT and GHG.
Second, this study assessed the avoided emissions, which are not observable. Some households joined a car sharing program instead of purchasing a new vehicle, which they would likely have driven much more than the shared vehicle. Thus, the reduction also includes the VMT avoided by households that chose not to purchase a vehicle. Together, the observed impact and the avoided emissions constitute the "full impact" of car-sharing.
The authors caution that one cannot assume every household will decrease its GHG upon joining a car sharing organization, but that the overall effect is a statistically significant net reduction. When the authors account for the inactive share of car sharing members (between 15 and 40% of all members seldom use shared vehicles), they estimate that the annual aggregate impact of car sharing reduces between 160,000 and 225,000 MTCO2 per year. This seems to include the reductions that result from car sharing fleets being more fuel-efficient; the authors' data show that the vehicles the members shed after joining car sharing organizations had fuel economies of 10 mpg less on average than the shared vehicles (32.8 mpg vs. 23.3 mpg). Emissions reductions attributable to land use and vehicle production were not assessed in this study, and college and business users of car sharing were excluded.
Studies of GHG reductions from car sharing have also been done in other countries. A study found that car sharing in Quebec reduced emissions by 1.2 MTCO2 per member per year. This was based on data that on average, members drove 2,900 fewer kilometers (1,800 miles) per year, and used lower-emissions vehicles (Communauto, Conseil Regional de l'environnement de Montreal, and Equiterre, 2007). Ryden and Morin (2005) claimed that in Europe, car sharing reduces members' CO2 emissions by 40 to 50%. Specifically, two estimates from European programs found decreases in GHG emissions per member of 54% (Bremen, Germany) and 39% (Belgium). Note that VMT reductions in these programs were higher than results from American programs. These estimates account for changes in the vehicle fleet mix, as well as increased emissions from transit use (Rydén and Morin  quoted in Millard-Ball et al. ).
Additional research focuses on VMT effects among car sharing members (without considering the differences in fuel economy between shared and personally-owned vehicles). Although VMT generally rose among those who had not previously owned a vehicle, those increases were generally more than cancelled out by the reductions in VMT among those who previously owned vehicles, leading to net reductions in VMT. One review of four U.S. car sharing programs found that average per-member VMT decreases ranged from 7 to 43%. Of these, two Portland, Oregon studies found average decreases of 18% in those who had previously owned vehicles, and 7.5% total decreases; neither of these findings were statistically significant. One of these Portland studies compared those who had previously owned vehicles to those who had not; VMT among owners dropped from 103 to 84 VMT per week, and VMT among non-owners rose from 0.2 to 25 VMT per week. A San Francisco study found second-year average VMT reductions of 2.8 to 1.5 VMT per weekday, and an Arlington (Virginia) study found average VMT decreases of 43%. The San Francisco and Arlington studies included but did not separately report the effects from members who previously owned vehicles from those who did not. For North American car sharing organizations, the proportion of all members who had previously owned vehicles was on average 40% (Millard-Ball et al., 2005).
In a study of nine European programs, average VMT fell between 26 and 72%. The European programs were more likely than the American ones to report changes in VMT both for members who previously owned vehicles and those who did not. For example, an Austrian study found that members who previously owned cars decreased their VMT from 10,100 to 3,850, while those previously without vehicles increased from 830 to 1,800 (Steininger, Vogl and Zettl, 1996, quoted in Millard-Ball et al, 2005). A British study found that members who previously owned a vehicle reduced their VMT by 1,100 while those who did not own a vehicle increased by 475 (Ledbury, 2004, quoted in Millard-Ball et al, 2005). On average, European car share members were evenly split between those who had previously owned a vehicle and those who did not (Millard-Ball et al, 2005).
This is not currently known. According to Millard-Ball et al. (2005), who surveyed dozens of public entities that partnered with car sharing organizations, about 60% of car sharing organizations have received some public money for start-up costs. However, there are no existing studies that link the public expenditures to promote car sharing to the GHG reductions. It is also impossible to estimate it since there is at best an indirect link between public expenditures and GHG reduction.
Estimates of car sharing effects and effectiveness are based on many assumptions, including:
One challenge in studying GHG reduction is that, because car sharing is entirely voluntary, it is difficult to establish a control group. Those who choose to enroll in car sharing programs may be more concerned about the environmental impact of their actions than the general public. Cervero and Tsai (2003) note that most "early adopters" in San Francisco's program were "environmentalists and avid cyclists who owned no car" (p. 44). Other studies looking at the motivations for joining car sharing found that as car sharing programs matured, the environmental rationale declined and members were more motivated by financial considerations (Harms and Truffer, , quoted in Millard-Ball et al., ).
In addition, because car sharing is quite new in the U.S., it is difficult to estimate the growth rates and potential of car sharing. One study estimated that 12.5% of the over-21 population in major U.S. metropolitan areas are potential candidates for car sharing (Shaheen et al., 2006). In absolute terms, this suggests a potential car sharing membership of 21 million people. However, as most published research has examined car sharing in major metropolitan areas, one cannot say much about the potential impacts in smaller cities.
Finally, there does not appear to be any data source that provides the fleet mix for shared vehicles. This means that it is difficult to accurately estimate the potential reductions in emissions based on the presumed lower emissions profiles of shared vehicles.
Promoting car sharing does not require major infrastructure investments or adoption of new technologies by the public sector, and agencies' costs relative to other strategies would be fairly low. While some public agencies may provide subsidies, such support would generally come at the beginning of the venture, since car sharing organizations can operate successfully based on revenues from members. Revenues to cover program costs can also come from payments for public parking made available to shared cars.
Per MiIlard-Ball et al. (2005), most of these start-up grants have been under $100,000. Cities have also donated vehicle parking spaces, but in the case of on-street spaces it is difficult to estimate a cost. Brookline, Massachusetts values its donated spaces at $750 per year per space, and Philadelphia, Pennsylvania charges the operator a one-time fee of $250 per space to cover the staff and signage costs. Simultaneously, some cities have substituted car sharing vehicles for city fleet cars and saved money.
Because car sharing is fairly new to the U.S., some agencies may not be familiar with it and/or may be skeptical about its viability. There may not be a natural "home" for promoting car sharing within multiple agencies. Zoning regulations may make it difficult to site car sharing vehicles (Millard-Ball et al., 2005).
Because car sharing is generally voluntary, offers more options to travelers, and can be sustained by private companies from revenues, social acceptability of car sharing is generally high. There may be some public objections to using previously public parking for car sharing vehicles, or requiring car sharing parking spaces in new residential development, but this has not been significant.
As car sharing entails costs to members, it may be difficult for low-income groups who could otherwise benefit from occasional access to vehicles to participate.
Current car sharing membership in the U.S. (as of January 2010) is about 390,000, with 7,500 vehicles (IMR, 2010). Car sharing currently operates in dozens of metropolitan areas and college towns in the U.S. A list is available at http://www.carsharing.net/where.html.
Car sharing is not limited to major metropolitan areas-Zipcar, the largest car sharing organization in the U.S., operates in several cities with populations less than 200,000, such as Winona, Minnesota, and Waterville, Maine (Zipcar, 2009).
It would be useful to assess additional studies of individual car sharing programs, as well as analysis of established European programs to determine the long-term impacts of car sharing. Additionally, data on differences among members who previously owned cars versus those who did not, according to the type of city, country, economic situation, and other factors would help in assessing the promise of car sharing programs in other areas. A further area of study includes understanding how diverse fleet mixes offered by car sharing companies may affect GHG emissions.
Bureau of Transportation Statistics (2009). National Transportation Statistics, 2009. U.S. Department of Transportation. Available online at: http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/national_transportation_statistics/index.html.
Cervero, R., and Tsai, Y.-H. (2003). San Francisco city carshare: Travel demand trends and second-year impacts: Institute of Urban and Regional Development, University of California at Berkeley, Working Paper 2003-05.
Communauto, Conseil Regional de l'environnement de Montreal, and Equiterre. (2007). CO2 Emissions Reduced by 168 000 Tons Per Year Thanks to Car-Sharing, press release, Quebec City, February 19.
Harms, S. and Truffer, B. (1998). The Emergence of a Nationwide Carsharing Co-operative in Switzerland. Prepared for EAWAG - Eidg. Anstalt fur Wasserversorgung. Abwasserreinigung und Gewasserschutz. Switzerland.
Innovative Mobility Research. (2010). Carsharing. from http://www.innovativemobility.org/carsharing/index.shtml.
Ledbury, M. (2004). UK Car Clubs: An Effective Way of Cutting Vehicle Usage and Emissions? M.Sc. thesis, Environmental Change Institute, University of Oxford.
Litman, T. (1999). Evaluating carsharing benefits: Victoria Transport Policy Institute.
Martin, E.W. and S.A. Shaheen. (2010) Greenhouse Gas Emission Impacts Of Carsharing in North America: Mineta Transportation Institute.
Millard-Ball, A., Murray, G., Schure, J. t., Fox, C., and Burkhardt, J. (2005). Car-sharing: Where and how it succeeds. Washington, DC: Transportation Research Board, Transit Cooperative Research Program Report 108, Chapter 5.
Rydén, C., and Morin, E. (2005). Moses environmental assessment report.
Shaheen, S. A., Cohen, A. P., and Roberts, J. D. (2006). Carsharing in North America: Market growth, current developments, and future potential: Institute of Transportation Studies, University of California, Davis, UCD-ITS-RR-05-11.
Steininger, K., Vogl, C., and Zettl, R. (1996). Car-sharing organizations: The size of the market segment and revealed change in mobility behavior. Transport Policy, 3(4), 177-185.
Policy: Many auto insurance policies have fixed premiums that are based on driver demographics, driving history, vehicle type, and other factors. Pay-as-you-drive insurance (PAYD, also known as pay-at-the-pump or cents-per-mile insurance) allows drivers to purchase insurance that varies the premium based on the amount driven. This converts some of the presently-fixed costs of driving to variable costs, and drivers can save money by reducing the number of miles driven.
Emissions Benefits and Costs: Pay-as-you-drive insurance is not widespread, so there are no overall figures about how much GHG is reduced by decreases in VMT. Both modeled and empirical studies have found reductions in VMT of approximately 5 to 10% per vehicle/policy, with national modeled results showing greater reductions than small pilot projects.
Implementation Concerns: While costs to public agencies are minimal, in many states pay-as-you drive insurance is not allowed for various reasons. Some states, for example, require insurance costs to be stated at the time insurance is purchased. With PAYD, costs vary based on actual miles driven. It is unclear how quickly PAYD would be offered or used even if more widely available, or how many companies would be interested in offering it. In some cases, the technologies used to determine the number of miles driven may raise privacy or enforcement concerns.
Car ownership entails many "sunk costs" such as the purchase price of the vehicle, registration fees, and insurance, which are not affected by the amount that the vehicle is driven. Some research has shown that drivers make decisions about which mode to choose for a particular trip based on immediate marginal out-of-pocket costs for the trip (e.g., gas, parking, and bus fare). This means that many personal vehicle trips appear to be inexpensive in an absolute sense, and when compared to alternatives such as transit (Steininger, Vogl, and Zettl, 1996).
Pay-as-you-drive insurance (PAYD) allows drivers to purchase insurance that varies the premium based on estimated or actual driving distances within a certain period. The principle behind PAYD is that if costs vary based on vehicle use, drivers will consider the total costs and make fewer personal vehicle trips by making fewer trips or by switching to other modes. In comparison to fixed insurance rates, these options reduce GHGs and can save consumers money (Litman, 2009).
PAYD insurance can be implemented in multiple ways:
Technologies that can be used to account for mileage include odometer readings, global positioning systems (GPS), and units that receive data from the on-board diagnostic equipment. Odometer readings can be self-reported, but verified by mechanics or the company. GPS and the on-board units can transmit data to the insurance company.
Because insurance is regulated at the state level, whether PAYD is legal depends on state policies. Currently 34 states allow some form of PAYD insurance. However, regulations vary, so states that allow some type of mileage-based discount may not allow per-mile premiums. The GHG mitigation strategy is to legalize various forms of PAYD insurance, thereby allowing private insurance companies to develop a broad range of PAYD insurance products. Further action may be to encourage the uptake of these policies through education and publicity campaigns.
PAYD adoption is targeted at two groups: (1) private insurance companies, which would develop and offer PAYD to policyholders; and (2) individual drivers, who would choose them over conventional insurance products.
Studies involving small-scale PAYD pilot programs have been conducted in the U.S., and the literature provides estimates of the effects of PAYD if it were implemented across the U.S. There is consensus that, when PAYD is used, it decreases overall VMT among policyholders by between about 5 and 10%, although empirical studies show reductions on the smaller end of this range, in comparison to studies based on modeling. Note that studies using models assume all drivers have access to PAYD insurance; they do not make assumptions about the rate at which companies begin to offer PAYD policies or drivers switch to them.
A 2004 pilot program in Minnesota tested the response of 130 drivers who volunteered to participate (their actual insurance levels were not affected, but they were able to benefit financially if their mileage declined). Overall VMT declined by 4.4%, with greater decreases seen during weekday peak hours (6.6%) and on weekends (8.1%). Households that were willing or able reduce their VMT did so at fairly low payment thresholds (5 cents per mile), but other households did not change their mileage even at much higher levels (25 cents per mile) (Cambridge Systematics et al., 2006). In a study of about 3,000 households in Texas who participated in a Progressive Insurance study, the average reduction in VMT was 5%. Peak hour miles were reduced by only 3.2%, which was the opposite of the Minnesota findings (Progressive Insurance and NCTCOG, 2007).
If one assumes that PAYD insurance policyholders reduce their annual VMT by 5%, one can compute the annual CO2 reductions for cars and light trucks:
For passenger cars: Using 2008 values, a 5% reduction from the average 11,800 VMT annually (National Transportation Statistics, 2009) to 11,200 VMT annually, and assuming 22.6 MPG (the average fuel economy for passenger vehicles in 2008 (Bureau of Transportation Statistics, 2009)), results in approximately 26 gallons of gas saved and 510 lbs of CO2 reduced annually.
For light trucks: Using 2008 values, a 5% reduction from the average 11,000 VMT annually (National Transportation Statistics, 2009) to 10,450 VMT annually, and assuming 18.1 MPG (the average fuel economy for light trucks in 2006 (Bureau of Transportation Statistics, 2009)), results in approximately 30 gallons of gas saved and 595 lbs of CO2 reduced annually.
Three nationwide studies using models are widely cited in the discussion of the impacts of PAYD. All of them assume that all drivers would be using PAYD insurance. One found an overall reduction in VMT from 9.2 to10%, depending on the type of PAYD model used and the state, since states differ in their average insurance costs per mile (Edlin, 2003). A second study claims that "fully implementing" PAYD (which is not precisely defined in the paper, but presumably means that all U.S. drivers are PAYD subscribers) would save 11.4 billion gallons of gasoline per year (a savings of 9.1%) (Parry, 2005). This translates to approximately 100 million MTCO2. The most recent study found that PAYD would annually reduce VMT by 8%, oil consumption by 4%, and CO2 emissions by 2% from 2006 levels (Bordoff and Noel, 2007). This estimate is lower than the others because of fuel price increases, meaning that consumer response would be smaller since the proportion of driving costs attributable to insurance has declined.
There are no existing studies that specifically address this question, nor even any that estimate the cost of introducing PAYD insurance. Bordoff and Noel (2008) declined to estimate costs, stating that the main cost would be to install equipment and prices for nationwide implementation are very uncertain (p. 36). They also claim that any cost would be outweighed by the societal benefits of reduced congestion, crashes, and savings to drivers, although these assertions were not included in the model.
One main uncertainty in estimating the effects of PAYD programs is how quickly PAYD insurance would spread in the marketplace, once legal. Only a handful of companies currently offer policies based on miles driven, and other insurance companies do not seem to be coming to market with similar products. It is also unknown how many drivers would change to PAYD insurance. While there is a clear incentive for those who drive fewer miles than average to switch policy types, those who drive the average or above would pay the same or more (depending how the policy is structured) with PAYD, so it is not clear whether they would switch voluntarily.
Another key uncertainty is the degree to which drivers might reduce their VMT with PAYD insurance. Studies that use models to predict changes must make assumptions about all of these factors. The empirical studies reviewed found lower reductions, between 4 and 5%, than the studies that used models, between 8 and 10%. When annual VMT is measured in the trillions, even a few percentage points make an enormous difference in determining PAYD effectiveness in reducing GHG emissions.
There are no particular agency costs associated with PAYD insurance, because no additional infrastructure is required and most observers assume that, once low-mileage drivers realize they can save money, the programs will grow on their own in the marketplace.
State insurance regulators are the primary government players in legalizing PAYD insurance. Some current state policies prohibit PAYD insurance; for example, in a state requires that insurance premiums be quoted to the customer before purchasing insurance, a PAYD policy that bills the customer afterward based on miles driven would be in violation of state law. Lifting some of these restrictions may conflict with other policy goals. State DOTs and MPOs have not generally played large roles in advocating for PAYD to be legalized, but they could certainly do so.
PAYD has had high consumer satisfaction where implemented (87% in one survey of a pilot program [Progressive Insurance and NCTCOG, 2007]), with most people signing up because of the opportunity to save money. While PAYD would have a positive impact on people who drive less than average-one estimate states that two-thirds of drivers would save money with PAYD insurance (Bordoff and Noel, 2008)-it is not clear what the adoption or satisfaction rate would be among drivers who drive more than average, since presumably their premiums would increase. This has so far been avoided by pricing strategies that implement PAYD as a series of discounts, under which no driver pays more, and because PAYD is of course voluntary. It seems possible that PAYD could result in a system, at least in the short-term, in which low-mileage drivers pay their fair share, while high-mileage drivers continue to be subsidized. Bordoff and Noel (2008) assume that as adoption becomes more widespread, insurance companies would be forced to raise their rates on high-mileage drivers, resulting in a "virtuous circle" in which drivers would be compelled to drive less to keep their insurance rates low.
Depending on the technology used, some drivers may be reluctant to switch to PAYD insurance for privacy reasons (e.g., if they perceive that the insurance company is tracking where they drive). Insurance companies in other countries have implemented PAYD with technologies such as global positioning systems (GPS) and other types of on-board units that plug into the vehicle's on-board diagnostics port to record data related to speed, which can then be used to determine mileage. While all types of on-board units can be configured to provide only the number of miles driven, and not the location, drivers may not be convinced that such protections are in place.
Bordoff and Noel (2008) pointed to three key barriers to adoption: the difficulty of monitoring mileage driven, state insurance regulations, and patented technology. First, various technologies exist today to meter mileage or otherwise tie insurance coverage to miles driven, although privacy concerns and expense make them difficult to adopt. One experiment with PAYD by British insurance company Norwich Union purportedly ended when the equipment cost was found to be too high relative to the program's benefits (Norwich Union axes "Pay as you Drive" Scheme, 2008). Second, many state insurance regulations (e.g., which require stating the premium cost up-front) prohibit or conflict with PAYD characteristics such as basing payment on miles driven. Third, Progressive Insurance, the only U.S. company to offer PAYD insurance with an after-market technology for metering mileage, has obtained patents that seem to make it difficult for other companies to bring similar technology to market (Bordoff and Noel, 2008).
It is possible that drivers may fraudulently try to lower their premiums by reporting lower mileage than they actually drove. A federal report on odometer fraud found that the possibility of an odometer being rolled back (tampered with to show a lower number of miles driven) is about 3.5% over the first 11 years of the vehicle's life. Annually there are about 450,000 cases of odometer fraud (NHTSA, 2002). Carfax, a private company that supplies vehicle reports, says, "Digital odometers, thought to be the answer to odometer tampering and fraud, are as easy, if not easier, to alter as their mechanical predecessors." (Carfax, 2010).
Insurance companies may experience lower premium revenues. However, if PAYD programs encourage less or safer driving, accidents could be avoided and thus companies could save on insurance payouts.
Currently 34 states allow some form of PAYD insurance, as do a number of countries in Europe and Asia (EDF, 2009). However, only one company in the U.S., Progressive Insurance, currently offers PAYD on a mileage basis in nearly 20 states. GM vehicle owners can apply for mileage discounts if their vehicles are equipped with OnStar, a GPS system, and Milemeter (which operates only in Texas) offers PAYD insurance based on odometer readings. There are no published figures on the number of American drivers who currently have PAYD insurance. PAYD has been more widely adopted in Europe, Asia, and South Africa.
Studies of market penetration in other countries where PAYD has been adopted may exist. Such studies should be assessed for lessons learned and to give an indication of how PAYD might fare in the U.S.
Bordoff, J. E., and Noel, P. J. (2008). Pay-as-you-drive auto insurance: A simple way to reduce driving-related harms and increase equity. Washington, DC: The Brookings Institution, Discussion Paper 2008-09.
Bureau of Transportation Statistics (2009). National Transportation Statistics, 2009. U.S. Department of Transportation. Available online at: http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/national_transportation_statistics/index.html.
Cambridge Systematics, GeoStats, and MarketLine Research. (2006). Mileage-based user fee demonstration project: Pay-as-you-drive experimental findings, final report. Minneapolis, MN: Minnesota Department of Transportation, MN/RC - 2006-39A.
Carfax. (2010). Uncovering Odometer Fraud. http://www.carfax.com/car_buying/odometer.cfx, accessed on March 1, 2010.
Edlin, A. S. (2003). "Per-mile premiums for auto insurance." In Economics for an imperfect world: Essays in honor of Joseph Stiglitz. Cambridge, MA: MIT Press.
Environmental Defense Fund. (2009). Drive less, pay less for insurance. Web page at www.edf.org/page.cfm?tagID=31651, accessed December 16, 2009.
Litman, T. (2009). "Pay-As-You-Drive Pricing For Insurance Affordability," Victoria Transport Policy Institute.
National Highway Traffic Safety Administration. (2002, April). Preliminary Report: The Incidence Rate of Odometer Fraud. DOT HS 809 441, NHTSA Technical Report.
Norwich Union axes "Pay as you drive" Scheme. (2008, June 18). Motortrader.com.
Parry, I. W. H. (2005). Is pay-as-you-drive insurance a better way to reduce gasoline than gasoline taxes? AEA Papers and Proceedings, 95(2), 288-293.
Progressive County Mutual Insurance Company, and The North Central Texas Council of Governments. (2007). Pay as you drive (PAYD) insurance pilot program, phase 2 final project report.
Steininger, K., Vogl, C., and Zettl, R. (1996). Car-sharing organizations: The size of the market segment and revealed change in mobility behavior. Transport Policy, 3(4), 177-185.
Policy: Most vehicle trips, especially commuter trips, are taken in single-occupant vehicles (SOV). The capacity of the existing roadway network could be increased if vehicle occupancy increased. Ridesharing strategies include conducting outreach programs and providing services to increase carpooling and vanpooling, and thereby reducing VMT and GHGs). Agencies can also create high-occupancy vehicle (HOV) lanes, which enable ride sharers to avoid congestion or tolls, serving as a further inducement.
Emissions Benefits and Costs: The emissions effects of ridesharing vary greatly depending on the types of policies used to encourage it and the context in which it is encouraged. Additionally, many studies report vehicle occupancy rates and mode share rather than GHGs. For these reasons, generalizations about GHG effects cannot be made. Additionally, costs to implement ridesharing programs are often bundled with other programs, so it is difficult to develop cost estimates. There is even less evidence about the effects of HOV lanes; some studies have found increases in ridesharing along HOV corridors, while others have not, and no studies were identified that assessed GHG emissions specifically. Importantly, creating new HOV lanes produces GHGs, and these life-cycle emissions must be considered in an assessment of HOV effectiveness as a GHG mitigation strategy.
Implementation Concerns: For ridesharing programs, concerns are few: they are widely implemented and well accepted given the benefits they provide to travelers. For HOV lanes, there is concern that they take away capacity from SOV driving and create more congestion. Recently there has been a trend toward high-occupancy toll (HOT) lanes, which are thought to be a more effective means of managing demand. HOV lanes may also involve construction costs if new lanes are created, which are much higher than other TDM strategy costs.
In 2001, just over 85% of all trips in the U.S. were made by car, and 65% of those car trips were single-occupant vehicles (SOVs) (NHTS, 2001). One aphorism in transportation planning is that the most underutilized capacity is the three or more empty seats in every SOV. If the same number of person trips were made in fewer vehicles, the transportation system would operate more efficiently. Moving a larger number of people with the same capacity, fuel consumption, and GHGs is an obvious way to increase efficiency.
Ridesharing, of course, occurs without any policy intervention, since many people are willing to share rides for convenience, cost savings, or company. This strategy seeks to increase the amount of ridesharing, particularly for commuter trips, which are more likely to be made in SOVs than other trip types. According to various surveys reviewed in Commuting in America, the SOV mode share for commute trips was about 75 to 77% in the early 2000s, an increase from 65% in 1980. Most of this change resulted from a decline in ridesharing, from 19 to 12% (Pisarski, 2006).
Ridesharing is generally divided into carpooling, in which ride sharers use their personal vehicles, and vanpooling, in which employers provide group transportation in larger vans and buses. Most efforts to increase carpooling and vanpooling are made at the regional level by commuter assistance organizations. In some regions, high-occupancy vehicle (HOV) lanes exist to encourage ridesharing.
While no special accommodations are needed for people to carpool, several strategies have been used by commuter assistance organizations to increase carpooling. One strategy is to provide rideshare matching services, which allow prospective ride sharers to find others who work and live near them. Second, many firms provide "dynamic ridesharing," which makes quick matches online for one-time rides (as opposed to conventional matching systems in which both ride sharers are interested in ridesharing for an extended period of time). Third, employers can encourage carpooling through preferred parking, cheaper parking rates for carpoolers and others, and commute assistance organizations often encourage employers to adopt such policies.
In a few areas, "casual carpooling" has become possible. That is, in order to gain access to HOV lanes and/or avoid tolls within these regions, solo drivers pick up passengers who wait at designated pick-up sites, often at park-and-rides or along transit routes, and bring them to designated drop-off points, generally in central business districts or other high-employment areas. Drivers and passengers who participate in casual carpooling generally agree to a few rules, which tend to be self-enforced, and safety has not proven to be a major issue with these informal programs. Local governments may assist such programs by installing signage, though they can operate independently.
A key difference between vanpools and carpools is that vanpools generally charge riders a fee to cover operating expenses, and federal law also provides a tax credit for vanpoolers (but not carpoolers). Commuter assistance organizations also promote vanpooling, for example by providing technical assistance (for example, working with an employer to set up a vanpooling program), by operating vanpools, or by providing direct subsidies. Some organizations, such as transportation management associations, also promote or operate vanpools, and there are several commercial vanpool providers.
HOV lanes enable ride sharers to avoid congestion-and in some cases, tolls-by designating specified lanes off-limits to SOVs. The number of occupants required in order to use HOV lanes varies by region; sometimes two people are required, sometimes three. The time of day that the lanes are restricted can vary as well; some operate during peak hours only, others 24 hours a day. HOV lanes can be converted from traditional lanes, or built as new lanes.
Like other TDM policies, ridesharing strategies target both employers and employees. HOV lanes can be used by all travelers, although many HOV lanes are in effect only during commute hours.
While ridesharing trends in the U.S. are well documented, there is little formal evaluation of the effectiveness of ridesharing promotion. One reason it is difficult to measure the impact of either ridesharing programs or the presence of HOV lanes on carpooling is that a number of other factors affect drivers' willingness to form carpools. A review of the literature on carpool formation in Parkany (1998) noted that factors such as cost of SOV driving, distance to work, education levels, whether the driver is a professional, the number of employers in an area, gender, and household size all play a role in decisions to carpool.
Importantly, people are more likely to rideshare for trips other than commuting, since much natural ridesharing occurs between family members. In 2001, the average vehicle occupancy for work trips was 1.13, while for social trips it was 2.03; the average for all trips was 1.63 (Hu and Reuscher, 2001).
The studies cited below largely rely on commuter surveys, not on observed behavior. The few reports that have been published on casual carpooling and dynamic ridesharing do not examine their effectiveness in reducing SOV driving. There have been some evaluations of HOV lane effectiveness in the U.S.
The Metropolitan Washington (D.C.) Council of Governments (MWCOG), which does much ridesharing assessment, relies on commuter surveys to assess the programs' impacts. For FY 2003 to 2005, their integrated ridesharing program (consisting of online ridematching as well as stand-alone interactive kiosks located throughout the region) reduced vehicle trips by 5,600 and reduced 146,000 VMT per day (LDA Consulting et al., 2005). Assuming an average fleet fuel economy of 20.7 mpg, this means a reduction of 62 MTCO2 per day. In the following evaluation period (FY 2006 to 2008), upgrades to rideshare software that supports the commuter operations center were evaluated separately and were found to reduce daily trips by 4,500 and VMT by 84,000, and annual MTCO2 by 15,100 (LDA Consulting et al., 2009). Percentage decreases from the baseline were not provided. The FY 2008 overall evaluation for all D.C.-area commuter programs combined found an aggregate reduction of 264,500 MTCO2 per year (NCRTPB, 2009).
In Atlanta, an evaluation of carpooling and ridesharing found a total daily trip reduction of 8,170 (5,500 attributed to carpooling and 2,670 to vanpooling) and net daily VMT reductions of 218,000 (127,000 to carpooling and 91,000 to vanpooling) (CTE, 2002). A later evaluation of four TDM measures, three of which were related to ridesharing (rideshare placement, vanpooling, and cash incentives to switch from SOV driving) found their combined impacts to be a daily reduction of 41,000 vehicle trips and 885,000 VMT. For both studies, percentage decreases from the baseline were not provided, and the study did not assess GHG reductions (CTE et al., 2004). For illustration purposes, if one (reasonably) assumes that the displaced trips took place equally in cars or light-duty trucks, which together have an average fuel economy of 20.7 mpg, the daily reductions from the 885,000 VMT reduced is about 380 MTCO2.
The state of Washington has estimated that its commute trip reduction program, which included a variety of TDM initiatives, reduces VMT by 170 million per year, or 680,800 per day, and emissions by 85,700 MTCO2 per year, or 342 MTCO2 per day (CTR Interim Report to the State Legislature, 2007).
In the 1980s, the Los Angeles region adopted several mandatory developer- and employer-based travel demand management programs. Developers in certain areas had to ensure that new developments reduce the number of SOV trips, and all employers with more than 100 employees had to reduce the number of SOV trips to their worksites to a specified amount. A study of one developer-based program found that carpooling was twice as high at buildings covered by the ordinance than at buildings that were not (7.4 vs. 3.5%) (Blankson and Wachs, 1990). Evaluations of the employer-based programs found that after one year, the average vehicle ridership (AVR) during the morning peak increased from 1.213 to 1.246, and for employers who participated for two years, it increased from 1.258 to 1.304 (Giuliano et al., 1993). Neither study reported trip distances, so it is not possible to estimate emission reductions.
In a study of casual carpooling, the two existing casual carpool systems in the San Francisco Bay Area and northern Virginia-which account for an estimated 3,000 and 3,500 carpools per weekday-were found to save about 3 million gallons of gasoline per year. The report estimated that a group of 150 commuters who switched from SOV commuting to casual carpooling would save about 52,000 gallons of gasoline per year, roughly the same as an express bus service. This was based on assumptions of 12-mile commutes and HOV lanes with higher traffic speeds than the general-purpose lanes (Dorinson et al., 2009). This amounts to approximately 460 MTCO2.
While HOV lanes may be a factor in individuals' decisions to rideshare, the extent of this is unknown and depends on many factors. Our review found that HOVs have a mixed record of promoting rideshare formation and that relatively little information on emissions impacts is available. In terms of emissions, older studies (from the 1970s) estimated reductions in fuel consumption ranging from 7-10% to up to 26% (Turnbull et al., 2006). A California study also found that HOV lane emissions rates (for criteria pollutants) were about half of the adjacent free lanes, but the study did not consider the extent to which HOV lanes may have contributed to congestion and emissions in the free lanes. The study did not assess GHG emissions (Parsons Brinckerhoff Quade and Douglas et al., 2002). A modeled study of returning HOV lanes to general purpose lanes in Minneapolis found a savings in fuel consumption of 4,000 gallons per day because of increased speeds throughout the region (Cambridge Systematics and URS, 2002). A recent overview of the literature on HOV lanes and emissions concluded that there is a "lack of in-depth information on the air quality, energy, and other related environmental impacts of HOV facilities" (Turnbull et al., 2006).
One report on California, which has more HOV lanes than any other state, found that a number of carpoolers in the San Francisco Bay Area cited the HOV lane as a factor in their decision to carpool. Survey data from the rest of the state was not available. In southern California, HOVs saw increases of 25 to 35% in peak period carpools compared to highways without carpools (Long, 2000). An evaluation of the HOV lanes in southern California found that about half of all carpools using the HOV lanes were formed in response to the HOV lane, and that average vehicle occupancies have increased on the facilities with HOV lanes compared to two control routes (Parsons Brinckerhoff Quade and Douglas et al., 2002).
A study in Dallas found that AM peak hour carpools at least doubled on all four HOV segments, and that average vehicle occupancy increased by 8-12%, while a control route without HOV experienced a 2% decrease in vehicle occupancy over the same time period. The HOV lanes also carried more persons per lane than the free lanes, a key measure of HOV lane efficiency (Skowronek et al., 1999).
Vancouver, Washington added a new HOV lane to an existing highway, in part to improve travel time reliability for carpools, vanpools, and bus transit. A study of this lane found increases in bus transit use of 18% in the first two years after HOV lanes were implemented. While baseline figures were not available to examine the growth in carpooling, the number of people using the HOV lane was nearly double the number using the free lanes before the HOV opened, suggesting that some carpools and vanpools must have been newly formed. Emissions reductions were not evaluated (Parsons Brinckerhoff Quade and Douglas, 2004).
In Oregon, an evaluation found that vehicle occupancy increased from 1.37 to 1.39, suggesting that carpool rates increased from 37 to 39% (quoted in Martin et al., 2005). A study in Utah found that during the AM peak, the HOV lanes carried fewer people than the average free lane (1,267 vs. 1,549), while in the PM peak they carried about 8% more people (1,700 vs. 1,568). Pre-HOV baseline figures were not reported, so the study could not measure whether total vehicle occupancy increased (Martin et al., 2005).
Finally, if new lanes are being built specifically for HOV use, the emissions from construction may be significant and must be taken into account in order to know the true effect of HOV lanes as a GHG mitigation strategy.
Without more detailed data of the impacts of ridesharing programs on behavior changes, this is very difficult to estimate. The Washington, D.C. region estimated a cost per CO2 ton reduced of $15, but this included all commuter assistance and was not specific to ridesharing. The report noted that, "The Commuter Connections Program is generally regarded as among the most effective commuter assistance programs in the nation in terms of reductions effected in vehicle trips and vehicle miles of travel" (NCRTPB, 2009). If this is correct, and the authors have found no evidence to the contrary, other regions would have higher costs per ton. With so little information available on the GHG impacts of HOV lanes, it is impossible to make a reliable calculation for these strategies.
The largest uncertainty in estimating effects is the degree to which SOV drivers respond to incentives to rideshare and to the availability of HOV lanes. As noted above, many factors can influence these decisions, so it is very difficult to assign impacts to specific TDM measures. Although the regional evaluations cited in the preceding section assigned such impacts, they are based on commuter surveys and not observed behavior, and make assumptions about the extent to which self-reported behaviors reflect actual changes in behavior.
Studies may also make assumptions about unintended effects of ridesharing. For example, although each ridesharing trip may remove one or more vehicles from the road, the vehicle that is being used is likely to travel farther to pick up or drop off each passenger. Side trips may increase for ride sharers if they are no longer able to combine activities like picking up groceries on a commute trip. These effects seem likely to be small in comparison to the VMT reductions, but many assessments disregard the effects entirely.
Where new HOV lanes are created, the GHG emissions from HOV lane construction, operations, and maintenance may be unknown but could reduce or even negate the benefits of roundabouts. Where HOV lanes are converted from traditional lanes, they may increase congestion and emissions in the free lanes. As cited earlier, there is a "lack of in-depth information on the air quality, energy, and other related environmental impacts of HOV facilities" (Turnbull et al., 2006).
The evaluation methodology of ridesharing and other TDM programs developed by the Washington Metropolitan Council of Governments is one of the most sophisticated in use in the U.S. The techniques are described in LDA Consulting et al. (2007).
As noted elsewhere, ridesharing is not often treated separately from other commuter assistance programs, and the same is true for ridesharing budgets. In Washington, D.C., the annual budget for the Commuter Connections program is $5.2 million, including staff time, operating the ridematching database, and marketing (NCRTPB, 2009). In Washington State, costs for the two-year period from 2007 to 2009 were $7.3 million, or an average of $3.65 million per year; again, this includes all components of commuter assistance, not just ridesharing (WSDOT, 2009). These are large programs; many regions presumably operate with far smaller budgets. As most regions currently operate commuter assistance programs, one cannot estimate start-up costs here.
Costs for HOV lanes vary since some have been converted from existing capacity, and others have been built as new construction. Adding one lane mile to an urban highway is estimated to cost roughly $10 to $15 million (FHWA, 2008).
Ridesharing efforts need to be sustained over time. As carpools dissolve, people and worksites move, and new employees and employers enter a region, rideshare matching efforts and more general education about travel demand management must be ongoing.
Ridesharing on a voluntary basis is already a widely accepted strategy. While several areas have passed mandatory TDM ordinances, these tend to be more controversial. Los Angeles had fairly stringent requirements in place that were eventually softened due to pressure from the business community.
HOV lanes have met with controversy as well, sometimes because they are perceived as taking capacity away from SOV drivers in congested free lanes, and sometimes on the environmental grounds that in freeing capacity they induce more travel demand for driving (Turnbull et al., 2006).
Barriers to increased ridesharing include difficulties in finding rideshare partners, lack of schedule flexibility, and low commute costs. Some of the difficulty in finding partners can be solved with rideshare matching services, while others are linked to decentralized workplaces (since the odds of finding a good rideshare partner, or a vanpool, presumably rise with a higher residential density and higher density of jobs, living and working in low-density locations can make it more difficult).
Vanpools can also have problems since vanpools are generally paid services and must have a certain number of riders to remain viable. This is less of a structural problem and stems from the need to do some continuous marketing and outreach to identify new riders when previous riders drop out for whatever reason.
Most metropolitan regions have a commuter assistance program whose function is to decrease SOV commuting in a region. These programs generally work with employers not only to encourage employees not to drive alone to work, but also to provide services to employees such as rideshare matching. Many also do general outreach through media campaigns and special promotions ("try transit" or "bike to work week") to raise the public's awareness of commuting options. In three regions-Seattle, Southern California, and Tucson-these employer programs are mandatory; in other areas they are voluntary. Casual carpooling takes place in the San Francisco Bay Area; Washington, D.C.; Houston; and Pittsburgh (Kelley, 2007).
HOV lanes of various types (full-day vs. only certain hours, reversible vs. permanent, etc.) have been built in 25 states as of 2007 (FHWA, 2010).
As noted above, relatively little is known about the systemic impact of promoting ridesharing, and even less about the differences in particular means (general outreach and ridematching vs. employer-based incentives, static vs. dynamic ridematching, and carpooling vs. vanpooling). It is also important to study further the long-term ridesharing retention rates, since job, schedule, and residential changes mean that people frequently return to SOV driving if ridesharing is no longer convenient. It is additionally difficult to draw conclusions about the effectiveness of such programs across different regions, especially when most of the research on these programs is from large regions.
With regard to HOV lanes, more before-and-after data could be collected and analyzed to determine their effectiveness in carpool formation and whether they move more vehicle occupants per lane per hour than conventional lanes. It might also be useful to conduct research on how effective they have been in effecting a market shift to lower-emissions vehicles; anecdotally, those HOV lanes that exempt hybrid or electric vehicles have become more crowded as those vehicles gain in popularity (Ginsberg, 2005).
Blankson, C. and M. Wachs. (1990, January). Preliminary Evaluation of the Coastal Transportation Corridor Ordinance in Los Angeles. Transportation Research Record.
Cambridge Systematics and URS (2002, February). Twin Cities HOV Study, Volume I, Final Report, Minnesota Department of Transportation.
Center For Transportation and the Environment (CTE). (FY2002). Evaluation of the effectiveness of programs contained in the "framework for cooperation to reduce traffic congestion and improve air quality", phase three, FY2002 Atlanta TDM framework final report: Georgia Department of Transportation.
Center For Transportation and the Environment, Georgia Department of Transportation, Federal Highway Administration, CIC Research Inc., Earthmatters Inc., ESTC, and LDA Consulting. (2004). Voluntary mobile emission source program (VMEP) state implementation plan (SIP) assessment, 2004 VMEP assessment.
Dorinson, D., D. Gay, P. Minett, and S. Shaheen. (2009). Flexible Carpooling: Exploratory Study. Davis, Institute of Transportation Studies, University of California at Davis.
Federal Highway Administration. (2010). HOV Clearinghouse, available at http://hovpfs.ops.fhwa.dot.gov/clearing.aspx.
Federal Highway Administration. (2008). Congestion Pricing A Primer: Overview. FHWAHOP-08-039. Available at http://ops.fhwa.dot.gov/publications/fhwahop08039/cp_prim1_00.htm. Accessed on September 6, 2010.
Ginsberg, Steven. (2005). Hybrids Could Lose HOV Perk Early; Va. Offers Options Aimed at Restricting Usage of I-95/395 Lanes. Washington Post, March 2.
Giuliano, G., K. Hwang, and M. Wachs. (1993). "Employee Trip Reduction in Southern California: First Year Results. Transportation Research vol. 27a, no. 2.
Hu, P. S., and Reuscher, T. R. (2004). Summary of travel trends: 2001 National Household Travel Survey. Washington, DC: U. S. Department of Transportation, Federal Highway Administration.
Kelley, K. L. (2007). Casual carpooling-enhanced. Journal of Public Transportation, 10(4), 119-130.
LDA Consulting, CIC Research Inc., ESTC, and Center for Urban Transportation Research. (2009).
Transportation emission reduction measure (TERM) analysis report, FY 2006-2008. Washington, DC: National Capital Region Transportation Planning Board Commuter Connections Program.
LDA Consulting, CIC Research Inc., ESTC, Shirazi, E., and Collier, C. (2005). Transportation emission reduction measure (term) analysis report, FY 2003-2005. Washington, DC: Metropolitan Washington Council of Governments Commuter Connections Program.
Long, R. (2000). HOV lanes in California: Are they achieving their goals? Sacramento: Legislative Analyst's Office.
Martin, P. T., Lahon, D., and Stevanovic, A. (2005). Review of the effectiveness of the high occupancy vehicle (HOV) lanes extension. Salt Lake City: Department of Civil and Environmental Engineering, University of Utah.
National Capital Region Transportation Planning Board (NCRTPB). (2009). Work program for the commuter connections program for the greater Washington metropolitan region, fiscal year 2010. Washington, DC: Metropolitan Washington Council Of Governments.
National Household Transportation Survey. (2001). Data from online analysis tool, accessed December 16, 2009.
Parkany, E. (1998). Can HOT Lanes Encourage Carpooling? A Case Study of Carpooling Behavior on the 91 Express Lanes, Institute of Transportation Studies, University of California at Irvine.
Parsons Brinckerhoff Quade and Douglas Inc., Kaku Associates Inc., Texas Transportation Institute, Strategic Consulting and Research, and HS Public Affairs. (2002). HOV performance program: Evaluation report, Los Angeles County Metropolitan Transportation Authority.
Parsons Brinckerhoff Quade and Douglas Inc. (2004). Vancouver HOV lane pilot project, evaluation report #6.
Pisarski, A. E. (2006). Commuting in America III: The third national report on commuting patterns and trends. Washington, DC: Transportation Research Board, NCHRP Report 550/TCRP Report 110.
Skowronek, D. A., Ranft, S. E., and Slack, J. D. (1999). Project monitors HOV lane operations Collage Station: Texas Transportation Institute, Project Summary Report 3942-S, Project 7-3942: Investigation of HOV Lane Implementation and Operational Issues.
Turnbull, K. F., Levinson, H. S., Pratt, R. H., John E. (Jay) Evans, I., and Bhatt, K. U. (2006). HOV facilities: Traveler response to transportation system changes. Washington, DC: Transportation Research Board, TCRP Report 95, Chapter 2.
Washington State Commute Trip Reduction Board. (2007). CTR interim report to the Washington state legislature.
Washington State Department of Transportation (WSDOT). (2009). Funding and structure (web site), www.wsdot.wa.gov/TDM/CTR/funding.htm. (inactive link 10/2012).
Policy: While most cities, even small ones, have some type of transit service, the out-of-pocket fares and inconvenience of riding transit may result in low ridership. One way to encourage people to switch from driving to transit is to make transit cheaper for riders. Federal law now contains tax incentives that allow employers to reduce employees' transit fares. Transit agencies have also adopted a variety of special programs to decrease riders' costs. Together, these can reduce GHG emissions if new riders switch from driving alone.
Emissions Benefits and Costs: The effect of employers' decisions to offer transit benefits is unknown, and where benefits are offered, the effects on ridership and transit mode share can vary greatly, from 0 to 17 percentage points. The ultimate effect on GHG emissions is unknown, but could be large in some cases. Transit agency promotions to reduce fares have not been systematically studied. The administrative costs for transit agencies to facilitate employer provision of transit incentives can vary significantly depending on the type of program. Studies report costs ranging from $100,000 to $500,000 annually, which principally include marketing and staff costs. Transit incentive programs also affect transit revenues: employer-based programs tend to increase revenues while fare reduction programs can decrease revenues.
Implementation Concerns: Barriers to implementing employer-based transit incentives are generally low because the programs are voluntary for consumers and may be voluntary for employers. While they include costs to employers, these benefits typically become part of an employer's benefits package.
Transit is available in many regions, albeit with different types and amounts of service and different ridership levels. To the extent that new riders can be accommodated with existing capacity, increased transit ridership can reduce GHG emissions, provided that the new transit trips replace vehicle trips, particularly SOV trips. Many regions try to increase the use of transit by generating more demand, generally by using incentives that reduce riders' costs. These incentives are most commonly provided through employer-based "transit benefit" programs, but they can also be provided with fare discounts or free ride programs for all transit users. (Increases in transit services to boost ridership are discussed separately in the discussion on Transit Improvements.)
Employer-based transit benefits are possible because of a provision of the U.S. tax code that allows employers to provide direct or indirect assistance toward employees' transit fares. Until the early 1990s, employers were allowed to provide free parking as an untaxed benefit to their employees, but any assistance above $15 to ride transit would be taxed. To correct this imbalance, the federal government re-defined "qualified transportation fringe benefit" in 1992 to allow employers to provide transit and vanpool benefits to employees tax-free up to certain levels. The provision also requires that employers provide transit passes and vouchers in regions where they are available, instead of paying the employees directly.
Transit benefits can be provided by employers in two ways: they can give an employee a specific amount in transit costs as a direct subsidy, or they can allow an employee to purchase transit passes with pre-tax income (similar to a health flexible spending account). They can also combine these two options so that an employee can receive a subsidy and set aside additional pre-tax income.
Two types of policies as transit incentives are identified below:
The first policy is transit benefits. As with other employer-based tax provisions, such as 401(k) plans, the employer must offer the benefit before an employee can utilize it. Therefore, one policy is for a public agency to market transit benefits to employers, encouraging them to provide their employees with transit benefits. The agencies that typically promote transit benefits to employers include commuter assistance organizations, transit agencies, and transportation management associations. In some areas, employers having a certain number of employees are subject to mandatory commuter trip reduction programs, in which employers must take actions to try to reduce the number of SOV commute trips. Transit benefits can be one way of fulfilling this mandate.
The second policy is for transit agencies to incentivize transit use by reducing the cost to riders. For example, transit agencies can implement universal pass programs that offer deeply discounted fares to employers on behalf of their employees, provided employers purchase passes for all or a certain portion of their workforce. Other incentives such as discounted or free fares can be directly offered to all riders, either on a permanent or promotional basis.
It is possible to use these policies together or separately. For example, Washington, D.C. employers make heavy use of employer-based transit benefits programs, but their employees do not receive any discounts from the regular rail fare. Universal pass programs, such as Eco Pass in Denver, are offered to employers at deep discounts, who then provide them as a transit benefit to employees. Portland's "Fareless Square" provides free rail transit to all riders within an area of downtown, regardless of their employment situation.
Like other TDM strategies, employer-based transit benefit strategies target both employers and employees because employees cannot take advantage of the tax benefits unless employers implement transit benefit programs. More general fare incentives are widely aimed at existing and potential transit riders.
There are no known studies that measure how effective public sector efforts are in persuading employers to offer transit benefits. Further, no studies develop national models or estimates. However, there is some literature on the effects of employer-based transit benefit programs on transit ridership at individual workplaces and in regions. There is also some research on the effectiveness of fare policies on transit ridership, and it suggests that external factors are more influential than fares (See the review of Transit Improvements).
Research suggests that the effect of transit benefit availability on ridership and mode share can vary significantly. A review based on survey data from 13 cities (ICF and CUTR, 2005) found that, where employers implemented transit benefits voluntarily (that is, they were not required to offer benefits), transit ridership-the number of employees riding transit on a given day-at those worksites increased by at least 10% (results were reported for cities as a whole, not for individual worksites). Gains were made both from new riders and from increased transit use from existing transit riders. While the review included various pass types, there was not enough data to determine whether certain pass types were more effective than others.
Specifically, increases in ridership at worksites ranged from 10% to over 150%, with about half of all surveys finding increases between 10% and 40%. Perhaps a more important indicator than gains in ridership is the increase in transit mode share before and after benefits were offered. In seven of the thirteen cities for which this information was available, the increases in transit mode share ranged from about 2 to 17 percentage points. The starting mode shares ranged from under 10 to over 40%.
In the ICF and CUTR review, transit benefits that were implemented in response to a mandatory commuter trip reduction (CTR) program (which requires employers to offer transit benefits or other incentives not to drive alone) had little and sometimes no effect on transit ridership. Interviews with CTR staff revealed that many employers implemented transit benefits to meet regional requirements, since they would receive credit for an employer program, even if few or no employees used the benefit (ICF and CUTR, 2005). The mandatory programs that this report addressed were in southern California, Washington State, and Tucson, Arizona. These have been less effective than the mandatory program enacted under Regulation XV in Los Angeles, which imposed much more stringent requirements on worksites to lower their average vehicle occupancy. However, pressure from the business community brought about its repeal (Sorensen et al., 2008).
ICF and CUTR (2005) did not estimate CO2 reductions from the increases in transit ridership, and this would vary significantly depending on the emissions from a trip on the original mode (which depend on the vehicle type, distance, system efficiency, etc.) and the emissions from a trip via transit. Nevertheless, the report does note that in seven of the twelve cities, 90-100% of new transit commuters were previous single occupancy vehicle (SOV) commuters. In seven of the 12 cities that had such data, SOV ridership at the surveyed employers declined between 1 and 15 percentage points.
The ICF and CUTR report also looked at the regional impact of transit benefits, but data was available in only three regions. The impact was largest in Washington, D.C., where a change that required federal employers to provide transit benefits was estimated to have resulted in a 29% increase in overall ridership, or 60,000 new riders per day. In Denver, the Eco Pass program (a universal pass program available to employers) was estimated to have brought in 6,000 new riders. In San Jose, the report estimated that the Eco Pass program might have brought in 16,000 new riders; however, the estimate was based on a small number of worksites, so the accuracy of this estimate is unclear. Unfortunately, because of the paucity of data at the regional level, and the many determinants of transit ridership, this sourcebook cannot provide even a range of potential regional impacts.
For those regions where data was available, one can approximate the emissions reductions achieved by programs by considering the extent to which the new trips replace SOV trips. Denver and San Jose were among the cities in which over 90% of new transit riders were previously SOV drivers; in Washington, 60% were SOV drivers. Assuming an average fleet fuel economy of 20.7 mpg and a 12.1-mile average one-way commute (Hu and Reuscher, 2001), this yields reductions of roughly 55 MTCO2 per day in Denver, 150 in San Jose, and 375 in Washington.
Transit incentives in the form of universal pass programs have generally shown positive results in increasing transit ridership in one particular case: universities. A number of universities have implemented such programs for their students, faculty, and/or staff. At the University of Washington, one of the first campuses to institute such a program, student transit mode share rose from 21 to 35%, and faculty/staff from 21 to 28%; SOV shares fell in both groups. At the University of California at Berkeley, student transit mode share rose from 5.6 to 14.1% (Nuworsoo, 2005). At UCLA, faculty/staff transit mode share for people living within the bus service area grew from 9 to 20%, and for students, from 17 to 24%. While both groups reduced their SOV use, there were declines in other modes as well, indicating that some people switched from carpooling, vanpooling, and bicycling (Brown et al., 2003).
BART, a heavy rail system in the San Francisco Bay Area, estimated GHG reductions for a variety of programs that it could implement with regard to transit fares and discounts. A "kids ride free" program, allowing free Saturday travel when accompanied by an adult, would reduce an estimated 15,000 MTCO2 annually. Unlimited ride passes, which would allow unlimited rides for a period of time within certain zones, were estimated to reduce 85,000 MTCO2 annually. Universal passes were estimated to reduce 148,000 MTCO2 per year, provided they were given to 10% of all adults in BART's service area (Nelson\Nygaard Consulting Associates, 2008).
It is not possible to estimate this at the national level with available data for several reasons:
In sum, neither the GHG reductions, nor the overall costs, can be effectively generalized.
BART estimated costs per metric tons of GHG reduced. The "kids ride free" program would cost from -$10 to $185 per MTCO2 reduced (-$10 because under one scenario the program would actually generate revenue from new adult riders). The unlimited ride passes would cost about $120 per MTCO2 reduced, while the universal passes would be $150 per MTCO2.
A number of factors affect whether providing transit benefits increases the number of riders at a workplace. Workplaces in auto-oriented, suburban locations, with little transit service, low benefits levels, other competing TDM programs, and lots of free parking will probably see a relatively small absolute increase in transit use, even when benefits are provided. Unfortunately, the data are not robust enough to determine the impacts of specific factors. In addition, it can be difficult to try to make comparisons across cities based on the level of transit supply. Transit agencies define their service areas differently, and multiple operators often serve one region, making it difficult to construct an objective measure of the level of transit availability in a region. It is even more difficult to make these comparisons across neighborhoods.
The effect of increased ridership on CO2 reductions, in turn, depends largely on the previous mode of transportation and, specifically, the percent of new riders who switched from SOV to transit. This figure can also vary widely (ICF and CUTR, 2005).
Agency costs to administer transit benefits programs (that is, the cost of the marketing, outreach, and fulfillment) are generally in the range of $100,000 to $500,000 annually, a fraction of the revenues that these programs produce (for most agencies, the administrative cost is just a few percent of the revenues) (ICF and CUTR, 2005).
A potential concern with discounted fare programs is the potential loss of revenue to the transit agency. While most transit agencies obtain over half of their operating revenues from sources other than the fare box, an agency facing a deficit may be considering raising fares, rather than reducing them. While no analyses of this issue appear to exist, anecdotally it seems some transit agencies are looking to tighten their discount programs for financial reasons (Sun Media, 2009; Grynbaum, 2009). This is probably less of an issue for those transit agencies that do not offer discounts to employers, in which case employees use their transit benefits to pay the full fare.
As transit benefits can be implemented by multiple agencies (generally, transit agencies, MPOs, and TMAs), some effort may be required to ensure that the most effective institutional structure is in place for each particular region. Agencies should not either leave gaps with regard to their target markets, or spend undue effort on overlapping initiatives. Transit benefits can also be implemented in many ways, so developing programs appropriate to the transit service and the audience may be challenging. Finally, setting appropriate prices and, if needed, determining how revenues will be divided among multiple agencies are important issues.
Transit benefits are largely acceptable and are already fairly widespread, although the level of use varies from region to region. Use of transit benefits, even if the employer is required to purchase a pass, is voluntary. Transit fare programs, since they decrease riders' costs, are generally well-accepted.
When employers provide direct subsidies to employees, they can incur fairly large total costs, depending on the number of participating employees and the transit fare structure in the region. Employers generally regard these non-taxable costs as part of a benefits package. Where employees set aside pre-tax money, there are small tax savings (perhaps 5% of the amount) to the employer, since those monies are exempt from payroll taxes. Since programs are voluntary, employers generally weigh these costs against other employee benefits when determining whether to implement them.
Most U.S. transit agencies that serve mid- to large-size regions have an employer-based transit pass program, which is marketed to employers either by the transit agency or other agencies. Almost all transit agencies provide some type of discounted fare, whether for students, the elderly, or disabled persons. The use of programs such as Portland's Fareless Square is much less common. As of the late 1990s, about 35 universities had universal pass programs (Brown et al., 2001)
While transit agencies and commuter assistance organizations may have data on the effects of transit benefits on workplace commute modes, much of it is unpublished. An effort to collect and analyze this data would offer a better understanding of the impacts of transit benefits programs.
Brown, J., Hess, D. B., and Shoup, D. (2001). Unlimited Access. Transportation, 28, 233-267.
Brown, J., Hess, D. B., and Shoup, D. (2003). BruinGO: An Evaluation University of California Transportation Center, University of California at Berkeley.
Grynbaum, M. (2009). "Seeing Political Pressure in Proposal to Cut Student Transit Fare Subsidy." New York Times, December 16.
Hu, P. S., and Reuscher, T. R. (2004). Summary of travel trends: 2001 National Household Travel Survey. Washington, DC: U. S. Department of Transportation, Federal Highway Administration.
ICF Consulting, and Center for Urban Transportation Research (CUTR). (2005). Analyzing the effectiveness of commuter benefits programs. Washington, DC: Transportation Research Board, TCRP Report 107.
Nelson\Nygaard Consulting Associates. (2008). BART Actions to Reduce Greenhouse Gas Emissions: A Cost-Effectiveness Analysis: San Francisco Bay Area Rapid Transit District.
Nuworsoo, C. (2005). Discounting Transit Passes. Access(26), 22-27.
Sorensen, P., M. Wachs, E. Y. Min, A. Kofner, L. Ecola, M. Hanson, A. Yoh, T. Light, and J. Griffin (2008). Moving Los Angeles: Short-Term Policy Options for Improving Transportation. Santa Monica, CA, RAND Corporation.
Sun Media. (2009). "Transit fare hike is in the cards; TTC deficit blamed on Metropass." Toronto Sun, September 24.
Policy: Many TDM strategies seek to reduce single-occupant vehicle trips. One mode that may absorb many of these trips is transit, especially for longer trips where walking or bicycling are not feasible. However, in many regions, transit is not a viable alternative to driving because the areas are not served by transit, the service frequencies are too low, or transit is not viewed as desirable. Transit improvements are aimed at increasing the potential for transit to absorb higher shares of trips by either creating new routes, increasing service frequencies, or increasing the comfort of transit to make it more attractive to potential riders.
Emissions Benefits and Costs: Emissions benefits and unit costs depend greatly on the size, nature, and the context of the investment made, and generalizations are not appropriate. Some research has found that transit improvements do encourage ridership and reduce GHG emissions, but may not be enough to stem the declines in ridership that have resulted from decentralized land use, relatively inexpensive fuel until recently, and other trends of recent decades. Additionally, certain kinds of improvements-such as adding new rail lines-may produce significant GHGs that must be included in emissions accounting.
Implementation Concerns: Transit improvements can be costly (especially new heavy rail service), controversial, and may not produce the anticipated ridership gains. For cash-strapped transit agencies, building and operating new service, or even increasing the frequency of existing service, may not be feasible.
Moderately- or heavily-utilized transit systems (which include bus, light rail, heavy rail, commuter rail, and paratransit) can generally transport people more efficiently, with fewer GHG emissions, than cars, particularly in comparison to single-occupancy vehicle trips. However, at the national level, transit use constitutes only 1.6% of all trips (even fewer than walking) and 1.2% of all miles traveled. Transit tends to be most popular for commute trips; about 3.4% of all trips to and from work were on transit (Hu and Reuscher, 2001). The use of transit varies significantly by region and density; about 40% of all transit trips in the US were in the New York region alone (APTA, 2009).
Transit improvement projects seek to increase transit use by increasing transit availability, convenience, and comfort. Transit improvements can include:
Transit agencies-which are operated at the local, regional, and even state level-are integral to improving transit services. State DOTs also build and operate some rail lines. While many projects are undertaken with local, county, regional, and state funds, major transit improvements, such as new rail lines and extensions, are often funded in part by the Federal Transit Administration, which allocates New Starts funds to transit agencies applying for funding for capital projects. Other FTA grant programs help fund purchases of buses and upgrades to rail systems. Major capital improvements generally involve collaboration between a transit agency and other local units of government (for example, to select new rail alignments), and states may help plan and fund major transit projects.
Transit improvements can encourage transit use among people who usually drive and further increase its use among existing riders. While rural transit systems exist, transit improvements have larger impacts on ridership and VMT in urbanized areas.
Transit use is affected by many factors, which Taylor and Fink (2003) suggested are the following:
These do not contribute equally to ridership: one study found that internal factors (that is, service provision and fare levels) explain about one-quarter of the variation in transit ridership levels; the other three-quarters depend on external factors (Taylor et al., 2009).
Many studies have been conducted about the impact of transit availability on travel behavior, often in conjunction with land use issues. Importantly, as with land use, the issue of self-selection makes it difficult to draw conclusions about the effects of transit improvements: that is, do people change their behavior because they move to a neighborhood served by transit, or do they select a neighborhood with transit because they wish to ride transit? Most research draws comparisons between neighborhoods with similar demographic characteristics, but this does not entirely eliminate the self-selection problem.
For these reasons, the link between making improvements to transit services and achieving reductions in GHG emissions is too tenuous to be able to make inferences about how much reduction might be achieved with specific investments. Instead, this section looks at the evidence about the intermediate step-whether and by how much ridership increases with such improvements.
One study looked at the effects of light and heavy rail expansion on the use of transit for commuting between 1970 and 2000, a period that saw a large degree of suburbanization and decentralization across the U.S. (While the focus was on the effects of rail expansion, transit figures include use of both rail and buses.) Across the seven regions that had pre-existing rail lines in 1970, the average transit mode share for commuting fell from 30 to 23%. (The actual decline in mode share varied by city, but all seven cities showed decreases.) In 14 regions that added rail lines by 2000 where none existed in 1970, transit use decreased in seven, remained the same in two, and increased in five. With only one exception, the commute mode share in 2000 was 7% or less. (In comparison, in cities with only bus transit, the share of commuting by transit bus also fell, from 5 to 2%.) Essentially, the study found that the addition of rail helped stem a decline in transit use, particularly among suburban commuters. While the use of existing rail and bus lines by urban commuters fell, the new rail lines increased the number of suburban commuters-but not by enough to overcome decreases among other passengers (Baum-Snow and Kahn, 2005).
Interestingly, the study also calculates the transit use that would have been expected if decentralization had not occurred between 1970 and 2000. In all but two regions, the percentage of transit use is higher in that hypothetical, centralized case, supporting the idea that transit investments are more effective in centralized than decentralized regions (Baum-Snow and Kahn, 2005).
Another study looked at the 226 transit agencies that saw increased ridership in the second half of the 1990s. Of these, 188 agencies increased their service levels, while only 38 decreased them (service levels were measured in revenue vehicle hours, and included all transit modes). The study found that ridership increased with service increases, but at declining rates of return. That is, agencies that increased their service hours the least, an average of a 4.3% increase, saw ridership gains (in unlinked trips) of 8.5%, while those that increased service the most, an average of a 79% increase, saw a 64.1% increase in ridership-large but proportionally less. While service provision contributed more to ridership gains than did fare decreases, the study noted, "because the level of transit service provided is, to a large degree, a function of the demand for transit service, there is no guarantee that simply increasing service will result in corresponding ridership growth" (Taylor et al., 2002, p. 46).
One report estimated that the total fuel savings due to transit availability across the U.S. is approximately 5.2 billion gallons per year. This includes primary effects, meaning the use of transit as a substitute for private car travel, as well as secondary impacts, meaning that the more dense neighborhoods made possible by transit that are widely shown to reduce per capita VMT. This translates to CO2 emissions "savings" of about 46 million MTCO2 attributable to transit (Bailey et al., 2008).
The recent Moving Cooler report estimated that transit capital investments across the U.S. could reduce CO2 emissions by 144 to 575 million MTCO2 cumulatively by 2050. The range reflects three scenarios at different levels of aggressiveness, assuming that investments are made to increase transit ridership on all modes by 3%, 3.5%, and 4.67% starting in 2010 (Cambridge Systematics, 2009). Other types of transit improvements, defined as increasing the number of revenue service miles and increasing travel speeds, would, if implemented alone, achieve less than one-third of the GHG reductions that the capital improvements would. When combined with projected decreases in emissions from investments in intercity and high-speed passenger rail, the range of possible reductions is from 0.4 to 1.1% of total GHGs from on-road transportation in the US.
BART, a heavy rail system in the San Francisco Bay Area, estimated GHG reductions for a variety of programs that it could implement with regard to transit improvements. Increasing off-peak train frequency was estimated to reduce 1,000 MTCO2 per year. For rail extensions (the analysis was for a single 5.4-mile planned extension), the annual reduction could range from 30,000 to 79,000 MTCO2 (the high end of this range includes land use impacts from more compact development patterns). This does not include emissions from the project's construction, which could be significant. Another 10-mile extension, using a different technology, would reduce GHG by 38,000 to 111,000 MTCO2, depending on land use impacts (Nelson\Nygaard Consulting Associates, 2008).
Costs and ridership changes vary so widely from region to region that deriving a consolidated estimate of unit costs is not appropriate. Moreover, the only analysis of this relationship that appears to exist is at the national level in the Moving Cooler report. For the three capital investment scenarios, cumulative costs for capital expansion are assumed to be $256 billion, $505 billion, and $1,203 billion (2009 USD) over the period from 2010 to 2050. This would imply costs of $1,770 to $2,080 per metric ton reduced. For the scenarios about other transit improvements (increasing the number of revenue service miles and increasing travel speeds), costs per ton are $1,166 to $1,451 (Cambridge Systematics, 2009).
BART's estimates included costs per metric ton reduced. For increasing off-peak service, the cost was $2,000 per metric ton. The 5.4-mile extension would cost $2,000 per ton, or $720 per ton if emission reductions were realized from land use changes. The 10-mile extension would cost $940 per ton, or $280 per ton with land use impacts. Although the report does not state it directly, the difference in cost between the two extension projects seems to be a function of the different proposed rail types.
As the studies suggest, a key uncertainty is how much transit ridership will result from improvements. Taylor et al. (2002) found that in some regions, transit agencies that increased their service still saw decreases in ridership, and in the cities analyzed by Baum-Snow and Kahn (2005), the proportion of transit commuting declined in most cities with new rail investments. Clearly, increasing transit service in and of itself does not guarantee increased use. Therefore, new transit investments or services should be analyzed with respect to the specific characteristics of a region, particularly those factors that seem to affect transit use the most: the degree of centralization and land use factors such as parking and density. In addition, as with transit incentives, a number of external factors, such as gas prices, also influence mode choice.
In addition, many forms of transit improvements will produce significant GHG emissions-for example, creating new vehicles and rail lines and increasing service levels. These emissions must be considered in the life-cycle emissions of transit improvements in order to assess their effectiveness as GHG mitigation strategies. The studies cited in this report do not include these life-cycle emissions, and no estimates of the emissions associated with creating new vehicles or rail lines appear to be available.
Because costs vary so widely from project to project depending on the technology, costs of land, and assumptions about the useful life of investments, it is not possible to provide a meaningful range of costs. Rail is always more expensive than buses: costs per mile of rail construction in the study of 16 cities ranged from $7.7 to $70 million per mile for light rail and $17.7 to $365 million per mile for heavy rail (Baum-Snow et al., 2005). The years of useful life for rail guideway (the right-of-way on which it operates) range from 20 years (for at-grade guideway in mixed traffic) to 125 years (for underground tunnel work). The track itself lasts from 20 to 35 years (FTA, 2010).
Increases in service by adding transit vehicles are much less costly; a new bus costs on average about $425,000, while rail vehicles cost between $1 and $3 million (APTA, 2009). For new projects, FTA assumes that rail cars have a lifespan of 25 years, and buses from 12 to 18 years (FTA, 2010). Average operating costs as measured per vehicle revenue hour are $115 for buses, $189 for heavy rail, $434 for commuter rail, and $219 for light rail (FTA, 2008).
These costs must be weighed against the relative capacity of rail and bus to move people, operations and maintenance costs, and other factors.
In addition to large costs, many factors can make service improvements difficult: the need to focus on current operations rather than new projects, the difficulty of working with other units of government or interest groups, the difficulty of trying to coordinate station or corridor development with transit planning (transit agencies in the U.S. generally have little ability to influence how land around their stations will be used), and the long time frames for certain types of investments.
Adding new transit service can be controversial, and attitudes depend on the location, cost, proposed fares, and proposed land uses. Arguments used by opponents against transit improvements include high costs of improvements relative to ridership or other perceived benefits, fears that better transit will bring "undesirables" to a neighborhood and reduce property values, concerns about additional traffic and limited parking in the station area, and general anti-urban sentiment. Additionally, in several particularly contentious cases, racial prejudice influenced opinions toward transit improvements, with whites opposing transit out of concerns that non-whites would more easily access their neighborhoods. A major public opinion poll about overall attitudes toward transit found about one-third in favor, one-third opposed, and the remainder neutral (Weitz, 2008).
Most cities have some type of transit service, although the modes and service provision vary widely. The U.S. currently has 2,400 transit agencies operating over 5 billion vehicle miles annually (APTA 2009).
The life-cycle effects of transit improvements are important but cannot be assessed without an understanding of the GHGs that are produced from transit improvements. Research should be undertaken to assess these emissions from different kinds of transit improvements.
American Public Transportation Association. (2009). 2009 Public Transportation Fact Book, 60th Edition.
American Public Transportation Association. (2009). Public Transportation Vehicle Database.
Bailey, L., Mokhtarian, P. L., and Little, A. (2008). The broader connection between public transportation, energy conservation and greenhouse gas reduction. Washington, DC: American Public Transportation Association.
Baum-Snow, N., and Kahn, M. E. (2005). Effects of urban rail transit expansions: Evidence from sixteen cities, 1970-2000 [with comment]. In Brookings-Wharton papers on urban affairs (pp. 147-206). Washington, DC: The Brookings Institution.
Cambridge Systematics. (2009). Moving cooler: An analysis of transportation strategies for reducing greenhouse gas emissions. Washington, DC: Urban Land Institute.
Federal Transit Administration. (2008). 2008 National Transit Profile.
Federal Transit Administration. (2010, June 1). Standard Cost Categories workbook, revision 13. Available at http://www.fta.dot.gov/12305_15612.html.
Hu, P. S., and Reuscher, T. R. (2004). Summary of travel trends: 2001 national household travel survey Washington, DC: U. S. Department of Transportation, Federal Highway Administration.
Motavalli, Jim. (2001). Second-Class Transit? Los Angeles' Bus Riders Union Fights for Equal Treatment. E - The Environmental Magazine, September/October, XII(5), online at http://www.thefreelibrary.com/Second-Class+Transit%3F-a077749302
Nelson\Nygaard Consulting Associates. (2008). BART Actions to Reduce Greenhouse Gas Emissions: A Cost-Effectiveness Analysis: San Francisco Bay Area Rapid Transit District.
Taylor, B. D., and Fink, C. N. Y. (2003). The factors influencing transit ridership: A review and analysis of the ridership literature. UCLA Department of Urban Planning Working Paper.
Taylor, B. D., Haas, P., Boyd, B., Hess, D. B., Iseki, H., and Yoh, A. (2002). Increasing transit ridership: Lessons from the most successful transit systems in the 1990s: Norman Y. Mineta International Institute for Surface Transportation Policy Studies, MTI Report 01-22.
Taylor, B. D., Miller, D., Iseki, H., and Fink, C. N. Y. (2009). Nature and/or nurture? Analyzing the determinants of transit ridership across us urbanized areas. Transportation Research Part A: Policy and Practice, 43(1), 60-77.
Transport for London (TfL). (2004). Impacts monitoring - Second Annual Report.
Weitz, R. (2008). Who's afraid of the big bad bus? Nimbyism and popular images of public transit Journal of Urbanism: International Research on Placemaking and Urban Sustainability, 1(2), 157-172.
Policy: Commuting accounts for about one-third of all miles driven in the U.S. As information technology continues to improve, telework-working from home or an off-site location-has become increasingly feasible and attractive. Governments at all levels may encourage or provide incentives for employers to offer their employees the option of teleworking, thereby reducing commuter VMT.
Emissions Benefits and Costs: GHG reductions have been estimated for employees who choose to telework. Based on one national model, each teleworker reduces emissions by about 0.5 MTCO2 per year. However, it is unclear how telework encouragement programs affect decisions to telework. The main cost to public agencies for telework promotion is staff time, suggesting costs in the range of a few hundred thousand dollars annually. One study suggests costs may be as little as $3-4 per MTCO2.
Implementation Concerns: Such programs are generally acceptable to the public but may be resisted by employers due to concerns about management and productivity.
Telework-a term generally interchangeable with telecommuting-means working from home or an alternative location closer to home. Almost one-third of the vehicle miles driven in the U.S. are to and from work, making commuting the single largest element of total vehicle travel (Hu and Reuscher, 2004). Moreover, congestion tends to peak during the hours when drivers are likely to be going to work. Therefore, some reductions in both congestion and emissions could be achieved if some employees did not drive to work and instead worked remotely (i.e., telework).
The public sector role in telework is to encourage employers to adopt policies to allow employees to telework. Efforts to encourage greater use of telework have been undertaken at all levels of government. Localities and MPOs tend toward policies that provide outreach and technical assistance to employers, or in a few cases directly provide telework centers for employees to use. State and federal policies tend toward providing tax incentives to employers; a federal tax credit has been introduced in Congress several times but never adopted. Public sector employers have in many cases adopted telework policies for their own employees, and the Telework Enhancement Act of 2010 provides a framework for encouraging flexible work hours and telework opportunities for Federal workers.
Telework policies, like many TDM policies, are aimed at both employers and employees. Employers generally view telework as an employee benefit, rather than as a transportation program, and often tie it to issues unrelated to commuting, such as job description or length of service. Telework programs also target the employees who are eligible to telework (generally employees whose work can be performed away from conventional worksites).
No organization regularly collects information on telework, so it is difficult to assess national trends. Most empirical studies of telework are based on small-scale programs, and only one study was found that models the nationwide impact of telework. However, there is almost no data on the effectiveness of public sector programs to promote telework. Where telework has been implemented and studied in the U.S., researchers have found that it leads to fewer and shorter trips among teleworkers, but that overall results have been modest, at best. The main reason is that the proportion of employees who telework is low, and these employees telework occasionally rather than full-time.
The transportation goal of telework is to reduce the number of vehicle trips or the trip length. One review of multiple studies found that on average, the VMT on a telework day decreased anywhere from 53 to 77%, and that non-commute trips did not increase (Walls and Safirova, 2004). However, since most teleworkers do not switch from driving to telework on a full-time basis, overall impacts must be gauged based on the VMT reduction of an individual driver over a longer period of time. One study of a pilot program in Los Angeles found that after two years, teleworkers worked from home on average eight days per month (Nilles, 1993). Another California study of telework centers found that the average VMT declined among teleworkers on teleworking days by 65% (38 miles). When averaged over all days, teleworking and non-teleworking, total VMT declined by 17% (Belapur et al., 1998). Both of these studies looked at the behavior of teleworkers, not at how many employees began teleworking.
A Washington, D.C. study of a short-term intensive telework promotion program designed to encourage the adoption of telework throughout the region resulted in an average reduction in VMT of 7.6 miles per day per teleworker (Ramfos and Albiero, 2006). Assuming an average car and light truck fuel economy of 20.7 MPG, this reduces about 7 lbs of CO2 per day per teleworker. The D.C. study did not provide the percentage of target employees who began teleworking, but the 4,200 new teleworkers that resulted from the program fell far short of the 113,000 goal (Ramfos and Albiero, 2006).
Perhaps the best assessment of ongoing telework assistance is at the Metropolitan Washington Council of Governments, which releases an assessment every three years of its many outreach efforts, including telework. Overall, the existing telework program in the region was estimated to account for a reduction of about 47,100 MTCO2 annually over the period FY 2006-2008 (LDA Consulting et al., 2009).
One widely cited study provides an opportunity for estimating overall effects at a national level. Choo et al. (2005) used data about past telework trends and developed a model to estimate the national reduction in VMT for 1998 (that is, given what we know about how other factors affect VMT, how much less VMT is driven because of telework?). They assumed a 27-mile round-trip commute, with 76% of those miles driven alone; telework frequency of 1.2 days per week; and 15.7 million teleworkers (12% of the national workforce). The model found that telework causes total US annual passenger vehicle VMT to be 19.3 billion miles less than it would be without telework. This represents a VMT savings of 0.8%. Again, assuming an average fleet fuel economy of 20.7 mpg, this reduces GHGs by 8.2 million MTCO2 annually, or about .5 MTCO2 per teleworker.
There is little literature about the effectiveness of public sector programs to promote telework, and there appears to be no literature that links the costs of public investment in telework programs to resulting declines in VMT. Most programs are implemented at the regional level through a commuter assistance program, and these programs typically do not do a detailed analysis of their outcomes. Nevertheless, the Metropolitan Washington Council of Governments telework program in the region offers one data point. It was estimated to account for a reduction of about 47,100 MTCO2 annually over the period FY 2006-2008 (LDA Consulting et al., 2009). The annual cost for the region's telework program is therefore about $166,000 (National Capital Region Transportation Planning Board, 2006). This corresponds roughly to a reduction of about $3.50 per MTCO2. The main cost component of these programs is generally staff time.
Key assumptions include the number of U.S. workers whose positions are amenable to telework, the proportion who actually will take up teleworking, the average commute VMT driven by teleworkers (i.e., whether it is the same or higher than the overall commuting population), and the number of days per week teleworked. In addition, if unemployment remains high (at the time of this writing it is 8.3%) this may have long-term impacts on telework in two ways. First, the number of employees who commute may decrease as unemployment increases, and, second, if managers perceive that workers who telework are expendable or less productive, this may dampen the acceptance of telework. Choo et al. (2005) also posit that telework may have a natural plateau, or a point at which new teleworkers are balanced against those who return to commuting, for whatever reason, including changes in jobs or preferences.
No single reliable source exists for determining the number of teleworkers. Various data are available from a number of sources, such as the Census, American Housing Survey, Current Population Survey, the National Household Travel Survey, and several private market research firms, and their figures do not always agree. Long-term study of telework is hampered by this lack of consistency (Mokhtarian et al., 2005). In addition, most commuter-assistance organizations do not track the impacts of their efforts to increase telework.
Operating costs for outreach programs are generally on the order of hundreds of thousands of dollars, since they largely comprise staff time. Operating costs would increase if a public sector program were to provide incentives to defray the costs incurred by employers, such as purchasing computer equipment for employees. The Telework!VA program, operated by the Virginia Department of Rail and Public Transportation, offers to reimburse qualified employers up to $35,000, provided their resulting telecommuting program meets certain benchmarks (VDRPT, 2009).
Telework centers are more expensive, but less common. Setting up telecommuting centers represents a moderate capital investment. One report recommended at least three years of public funding for a center to establish itself. Estimated total cost for a 12,000 square foot facility with 60 workstations is $1.4 million in 2009 USD($625,000 for start-up costs, plus operating costs of $22,600 per month over three years) (Bacharach et al., 2005).
While there are no specific concerns associated with encouraging telework, it must be noted that agencies' abilities to influence telework habits may be limited.
There are few barriers at the public or individual level to telework implementation. It is more common for employers to resist telework programs out of concerns about the difficulty of managing employees remotely. For those individuals who telework, it may be that the face-to-face interaction is too important to forgo on a daily basis, which helps explain why it is more common to telework on occasion rather than daily (Rosenberg, 2007).
One study estimated the employer cost to establish a telecommuting program for their employees is roughly $3,000 in one-time costs and $1,100 in recurring costs. These costs include computer equipment and the associated telecommunications upkeep (JALA, 2009). Some evidence suggests that employers with teleworking employees experience some productivity gains, which may offset these costs (Butler et al., 2007), but this continues to be a topic of debate in the literature (Bailey and Kurland, 2002).
Telework is fairly widespread in both the public and private sectors, although, as noted above, no comprehensive estimates or databases exist.
It is possible that other regions in addition to Washington, DC might have assessed the cost and effectiveness of telework promotion programs, and a broader review of project reports would be valuable.
Bacharach, Jacki and Siembab, Walter. (2005, June 30). Implementation steps for two strategies promoting jobs/housing balance: Local preference housing and share work/communications centers. Final Report. Prepared for Ventura Council of Governments.
Bailey, D. E., and Kurland, N. B. (2002). A review of telework research: Findings, new directions, and lessons for the study of modern work Journal of Organizational Behavior, 23(4) Special Issue: Brave New Workplace: Organizational Behavior in the Electronic Age, 383-400.
Belapur, Prashent, Varma, Krishna, and Mokhtarian, Patricia. (1998). "The transportation impacts of center-based telecommuting: interim findings from the neighborhood telecenters project." Transportation, vol. 25, no. 3.
Butler, E. S., Aasheim, C., and Williams, S. (2007). Does telecommuting improve productivity? Seeking solid evidence of demonstrable productivity gains. Communications of the ACM, 50(4), 101-105.
Choo, S., Mokhtarian, P. L., and Salomon, I. (2004). Does telecommuting reduce vehicle-miles traveled? An aggregate time series analysis for the U.S. Transportation, 32(1), 37-64.
JALA International. (2009). Home-based Telecommuting Cost-Benefit Analysis, web page available at www.jala.com/homecba.php, accessed December 15, 2009.
LDA Consulting, CIC Research Inc., ESTC, and Center for Urban Transportation Research. (2009).
Transportation emission reduction measure (TERM) analysis report, FY 2006-2008. Washington, DC: National Capital Region Transportation Planning Board Commuter Connections Program.
Mokhtarian, P., I. Salomon, and S. Choo. (2005). "Measuring the measurable: Why can't we agree on the number of telecommuters in the U.S." Quality and Quantity 39(4): 423-452.
National Capital Region Transportation Planning Board Metropolitan Washington Council Of Governments. (2006). Work Program For The Commuter Connections Program For The Greater Washington Metropolitan Region, Fiscal Year 2007.
Nilles, Jack (1993, March). City of Los Angeles telecommuting project: Final report.
Ramfos, Nicholas and Albiero, Rick. (2006). Expanded telecommuting transportation emissions.
reduction measure (TERM) final report. Prepared for Metropolitan Washington Council of Governments. January 7, 2006. Available online at www.mwcog.org/store/item.asp?PUBLICATION_ID=258.
Rosenberg, M. (2007). Slow but steady "Telework revolution" eyed. Cascadia Prospectus.
Virginia Department of Rail and Public Transportation. (2009). Telework! VA web page, available at www.teleworkva.org/, accessed December 15, 2009.
Walls, M. and E. Safirova (2004). A Review of the Literature on Telecommuting and Its Implications for Vehicle Travel and Emissions. Washington, D.C., Resources for the Future.