Analysis tools in current use in transportation decision-making processes are not well suited for evaluating toll highway alternatives against more traditional "free" highway alternatives. This paper shows how existing analysis tools might be used in evaluating toll options. Using a case study, the paper demonstrates that relatively simple analytical procedures may be used to estimate the impacts of pricing alternatives and to generate information for use by local decisionmakers.
The case study also demonstrates that pricing alternatives can often accomplish the purpose of a major highway project more efficiently and more effectively than conventional alternatives that exclude pricing, while generating revenue to support bonds for project construction or to fund improved transit and paratransit services. With toll revenue to back bonds, project delays due to constrained funding can be avoided, and the public can be provided with superior mobility earlier and at lower public cost.
Major highway expansion projects in metropolitan areas in the U.S. are expensive. Yet they may be needed as solutions to traffic congestion as these areas grow and develop. In some areas, toll options are being proposed as a source of revenue to make them financially feasible. For example, a private company recently sent a proposal to the Virginia Department of Transportation to finance the expansion of the Capital Beltway in Northern Virginia (Washington, DC metropolitan area) by charging tolls to single-occupant vehicles on two lanes proposed to be added in each direction (1). Often such toll options must be evaluated within a short time frame, so that decision-makers can respond to private sector proposals. However, evaluation tools in current use in transportation decision-making processes are not well suited for evaluating such toll alternatives against more traditional "free" highway alternatives. This paper shows how quick-response analysis tools might be used in evaluating toll options. Using the Capital Beltway project (2) as a case study, the paper demonstrates how transportation performance and other impacts of toll options can be evaluated and compared to more traditional alternatives.
Two types of toll options have a reasonable chance of getting public acceptance:
These types of toll options, also known as "value pricing" options, are discussed below.
1.1 HOT Lanes
"HOT" is the acronym for "High Occupancy/Toll." On HOT lanes, low occupancy vehicles are charged a toll, while High-Occupancy Vehicles (HOVs) are allowed to use the lanes free or at a discounted toll rate. Tolls vary by time-of-day and are collected at highway speeds using electronic toll collection technology. There are no toll booths. Tolls may be set "dynamically," i.e., they may be increased or decreased every few minutes, to ensure that the lanes are fully utilized. Motorists are informed of the current toll rates through variable message signs placed in advance of the entrances to the HOT lanes.
HOT lanes can be introduced either by converting existing HOV lanes or by adding new lanes. They have been implemented in California and Texas. By maximizing the use of spare capacity on existing HOV lanes, HOT lanes can reduce congestion on general purpose lanes, and reduce the need for new highway capacity for unrestricted use. Variable tolls on new lanes ensure that new lanes will not get congested, and that spare capacity remaining after limited use by HOVs will be fully utilized.
1.2 FAIR Lanes
The new value pricing concept called FAIR (Fast and Intertwined Regular) lanes was developed to overcome equity concerns that sometimes surround efforts to implement variable tolls on previously untolled highway capacity (3). FAIR lanes involve separating congested freeway lanes into two sections, Fast lanes and Regular lanes. The separation may be done with methods as simple as using plastic pylons and lane striping. The Fast lanes would be electronically tolled, with tolls set dynamically, i.e., in real time, to ensure that traffic moves at the maximum allowable free-flow speed. Users of the Regular lanes would still face congested conditions, but would be eligible to receive credits if their vehicles had electronic toll tags. For example, if the current toll on the Fast lanes were $4.00, vehicles on the Regular lanes could get a credit amounting to $1.00, i.e., 25 percent of the current toll. The credits would be a form of compensation for giving up the right to use the lanes that had been converted to Fast lanes. Accumulated credits could be used as toll payments on days when a traveler chooses to use the Fast lanes, or as payments for transit or paratransit services, which would be subsidized using toll revenue from the Fast lanes.
FAIR lanes could increase vehicle throughput by as much as 50% on Fast lanes. The higher throughput occurs because freeway vehicle throughput under free flow conditions is significantly higher than when it is congested (4). FAIR lanes could increase person throughput even more with the provision of high quality transit and paratransit services. This in turn could lead to more efficient use of existing highway lanes, and thereby reduce the need for new highway capacity.
1.3 Lessons Learned from Implemented Projects
The SR 91 Express Lanes in Orange County, CA opened in December 1995 as a four-lane toll facility in the median of a 10-mile section of one of the most heavily congested highways in the United States. The toll lanes are separated from the general purpose lanes by a painted buffer and plastic pylons. There are eight general purpose lanes, four in each direction.
As of November 1, 2001, tolls on the Express Lanes varied between $1 and $3.60 in the westbound direction and $1 and $4.75 in the eastbound direction, with the tolls changing by time of day to reflect the level of congestion delay avoided in the adjacent free lanes, and to maintain free-flow traffic conditions on the toll lanes. Traffic volumes for discrete time intervals during the day are monitored on a regular basis on the Express Lanes. When volumes approach levels at which free flow of traffic might be at risk, a new toll schedule is developed and published. To discourage additional traffic, tolls are raised for those time periods when volumes are close to the maximum volumes that can support free flow. All vehicles must have an electronic transponder to travel on the Express Lanes. Vehicles with three or more occupants pay a reduced toll. These vehicles go through a special toll collection lane for HOVs so that they can be identified as vehicles eligible for the discount.
During heavy congestion periods, 40 percent of total vehicular traffic is carried on the express lanes even though they comprise only one-third of the capacity (5), because throughput is higher under free flow conditions (4). Due to higher HOV use on the lanes, the percentage of person travel on the lanes is even higher.
Projects such as SR 91 Express Lanes have taught important lessons with regard to such "value pricing" toll options:
2.0 ANALYSIS AND EVALUATION OF PRICING ALTERNATIVES
2.1 Available Analysis Tools
Through the Travel Model Improvement Program (TMIP), FHWA has been developing and disseminating advanced travel demand forecasting techniques to estimate the travel effects of pricing alternatives (7). Four-step travel demand models are being used in many cases to estimate traffic impacts of pricing alternatives, e.g., a recent study in Portland, OR (8). When results from four-step models are available, FHWA's Surface Transportation Efficiency Analysis Model (a.k.a. STEAM) may be used to generate estimates of mobility impacts (9). STEAM can also be used to generate estimates of a variety of environmental impacts from vehicle operation on the highway. These estimates are generated at the system level, to ensure that only net effects are reported. A pricing study in the Twin Cities, MN, has used STEAM to estimate the mobility and environmental impacts of alternative strategies (10).
However, the above procedures tend to be complex and often expensive to implement. In the time frame that government agencies must respond to private sector tolling proposals, such as the proposal for the Capital Beltway, it is not feasible to do extensive four-step modeling of pricing alternatives. Quick-response analysis tools, such as the tool demonstrated in this paper, are a feasible alternative. FHWA has developed quick-response analysis tools which can assist in travel demand and impact estimation. For example, FHWA's IMPACTS model (11) provides for estimation of impacts of toll alternatives. Other tools developed by FHWA, such as its Sketch Planning Analysis Spreadsheet Model (a.k.a. SPASM) (12) and its Spreadsheet Model for Induced Travel Estimation (a.k.a. SMITE) (13) may also be modified for use in evaluation of pricing alternatives. FHWA's SMITE model was modified for the case study analysis presented in this paper. The modified model is called SMITE-Managed Lanes (SMITE-ML).
2.2 Overview of SMITE
SMITE estimates "induced" traffic that might result from faster travel speeds, including new trips generated or attracted to new development, existing trips diverted from other destinations, and existing trips diverted from other modes of travel such as transit.
Goodwin (14) compared travel demand model forecasts, which accounted primarily for re-routed traffic, with observed traffic flows on nine improved urban roads and sixty-one rural roads in the U.K. He found that, for an average road improvement for which traffic growth is forecast by travel models, unpredicted traffic in the first year, over and above the forecast, was +5.7% for urban roads and +13.3% for rural roads. Based on his studies, he equates an additional 10% traffic to a demand elasticity, with respect to travel time, of -0.5.
However, researchers in the past have produced elasticity estimates as low as -0.2 (15). The wide range results from several factors, such as the relative importance of travel time component of the full user cost of the trip, the time frame of analysis, magnitude of new development induced, and length of trips served.
SMITE uses demand elasticity with respect to travel time to estimate induced travel. In order to develop an estimate of induced travel demand using demand elasticities, we must first estimate the reduction in the time "price" from the proposed action. This reduction, converted into a percentage reduction, can then be used with a demand elasticity estimate to provide an estimate of induced travel demand. However, one complication is that as total travel increases due to induced demand, the initial price reduction is slowly eroded. It is therefore necessary to go through a process of equilibration of demand and price. The change in travel time that occurs with increases in demand is easily modeled using speed-volume relationships (16, 17). By iteratively estimating induced travel demand and the resulting time price change, an equilibrium point can be found at which demand and price are in balance.
Figure 1 demonstrates the equilibration process graphically, for an improvement in highway speeds resulting from capacity expansion. Point A is the initial equilibrium point, i.e., the point at which demand and supply are in balance, prior to highway expansion. After highway expansion, travel time "price" is initially reduced to the level represented by point B. However, at that price, travel demand would increase to the level represented by point C on the demand curve. But at this higher demand level, the price would actually be much higher, as represented by point D on the price curve after highway expansion. At this price, demand would in reality be lower, as represented by point E on the demand curve. As the figure demonstrates, one might continue to follow this process and finally end up at point Q. At this point, demand and price are in balance.
An approximation of the procedure illustrated in Figure 1 has been incorporated into SMITE. For the demand curve, SMITE allows the user to provide demand elasticity estimates. Induced traffic associated with increases in highway speeds is calculated using a series of equations (18) approximating the equilibration process demonstrated graphically in Figure 1.
3.0 CASE STUDY ANALYSIS
The case study demonstrates application of quick-response analysis techniques using a speciallymodified version of SMITE called SMITE-Managed Lanes (SMITE-ML). SMITE-ML facilitates estimation of the impacts of toll options on existing or added lanes. Data for the case study demonstration were obtained from the Capital Beltway Study report (2). These data and other data needed as input into SMITE-ML are indicated in bold on the spreadsheet, which is available from www.fhwa.dot.gov/steam. Go to the Related Links page.
The proposed project is 14 miles long. The existing facility has 8 lanes, and the alternatives evaluated involved expansion to 10 lanes or 12 lanes.
3.1 Project Alternatives
Three conventional Build alternatives (i.e., without tolls) were evaluated in the Capital Beltway Study in comparison to the No Build alternative. These alternatives included:
In addition, three pricing alternatives were developed for the case study analysis presented in this paper. These pricing alternatives were variations on the No Build alternative and two of the three Build alternatives. The following pricing alternatives were developed:
Note that a pricing add-on to the 10-lane Alternative 1 was not evaluated. Barrier-separated priced lanes on a 10-lane cross-section are evaluated under Alternative 5.
For the pricing alternatives, tolls would be charged only during peak hours (6-10 am and 3-7 pm) on priced express lanes and would vary dynamically, to ensure that traffic flows freely at all times, including the peak hour of each peak period. This would ensure premium delay-free service for transit and paratransit riders, carpools and toll-paying vehicles. It is important to understand that the primary intent of pricing strategies is not to reduce mobility or freedom of travel, but to increase it by providing funding and uncongested travel conditions for better quality, cost-efficient alternative modes while instituting financial incentives to encourage use of these modes. In the alternatives developed, pricing revenues would fund high quality alternatives to solo-driving, including demand-responsive paratransit services, express bus or Bus Rapid Transit (BRT) services, and non-motorized options. And solo drivers willing to pay would enjoy a higher level of mobility.
3.2 Mode Choice Analysis
A "pivot point" mode choice model (19) has been incorporated into SMITE-ML. It was used to estimate impacts of the alternatives on peak period mode shares, pivoting off of estimated No Build mode shares in the year 2020. Impedance coefficients used in the model were those calibrated for the Washington, DC metropolitan area (19).
Travel corridor person trip demand estimates for the base case No Build alternative were needed as input to the model. These were obtained from the Capital Beltway Study (2), generated by the four-step model maintained by the Metropolitan Planning Organization (MPO). The Study report also provided No Build traffic forecasts for 10 separate sections of the freeway between interchanges. In the base case, with no widening, the MPO model estimated that the southern segment (about 3 miles long) would carry on average about 280,000 vehicles per day on the freeway. The middle segment (about 6 miles in length) would carry the highest volumes, about 310,000 vehicles per day. The lowest-volume segment to the north (about 5 miles in length) would carry about 250,000 daily vehicle trips.
The pivot point logit model estimated the changes in travel demand on alternative modes for each alternative, resulting from generalized user cost changes due to highway capacity improvements, tolls and new and improved transit and paratransit services. The model results reflected the effects of changes in congestion and financial inducements for use of alternative modes. Table 1 presents the results from this analysis for one of the three segments, the southern segment with mid-range traffic levels. For this segment, the model estimated:
The increase in transit trips with pricing amounts to about 10,500 to 12,000 round trips daily. This model estimate is not unreasonable, since it amounts to less than 10 percent of the current employment in the Tyson's Corner area, the major employment and commercial center located in the corridor. (Total employment in the area is 126,000, according to the Fairfax County Economic Development Authority's "Area Business Report", 2002). Moreover, employment is expected to grow significantly in the area by the forecast year 2020.
As shown in Table 1, carpool use estimated by the model ranged from an increase of about 14,000 person trips for the conventional alternatives, to smaller increases of 8,000 to 12,000 person trips under the pricing alternatives. Note that pricing does not generate as much carpool use as conventional alternatives because some would-be carpool commuters are attracted to transit due to the superior transit services under pricing alternatives provided with funding from toll revenue.
3.2 Traffic Forecasts
The pricing alternatives tend to reduce vehicle demand on the freeway and in the corridor, relative to the base case No Build alternative. However, the reduced congestion on the freeway will cause diversions of traffic to the freeway from other routes and destinations, and may even induce additional development and consequent new trips in the corridor. SMITE-ML estimates increases in traffic from these sources. Traffic diverted from arterials to the expanded freeway is first estimated by redistributing traffic such that relative levels of congestion on the freeway and the arterials stay the same. This technique is based on principles in NCHRP Report No. 255 (20).
An elasticity of demand with respect to travel time of -0.2 was used to estimate "induced" travel. This demand elasticity is at the lower end of the range found in the literature (15). It reflects the paucity of vacant land available for increased development in the corridor, the relatively large proportion of short trips, as well as the fact that mode choice changes which contribute to induced travel are already accounted for using pivot point mode choice analysis. The results are presented in Table 2 for the southern segment of the freeway.
It should be noted that, for restricted lanes and priced lanes, the induced demand calculation was suppressed. Induced carpool usage is already estimated by the mode choice analysis. Peak period HOV use on the restricted lanes was estimated assuming that 90% of HOV demand estimated by the pivot point mode choice model would use the lanes. Single-occupant vehicle (SOV) volumes in priced lanes were estimated to be equal to the spare vehicle capacity that would be available on the lanes at a Level of Service C, i.e., about 75% of absolute capacity. Price would be used to regulate the level of travel demand in order to maintain service at Level of Service C.
Note that the restricted lanes and priced lanes would be open to all traffic during off-peak times, free of charge. Therefore, on a daily basis, Alternatives 3 and 6 carry relatively higher volumes of traffic. They have far more capacity available for unrestricted use, i.e., two lanes in each direction. Due to the larger number of new lanes available and used in peak periods by both HOVs and priced SOVs with Alternative 6, diverted and induced traffic is significant, increasing vehicle volumes daily on the freeway by almost 30,000 above the No Build volumes for that segment, and by about 12,000 for the corridor as a whole. On the other hand, with pricing Alternative 4, vehicle volumes are reduced by more than 30,000 vehicles daily on the freeway as well as corridor-wide. This occurs because Alternative 4 shifts many peak period trips to carpools and transit. But it does not encourage significant diversions or new traffic, because no capacity increase is involved.
3.4 Speed, Delay and Toll Revenue Estimates
Table 3 presents estimates of speed, delay and toll revenues per mile for the southern freeway segment. For restricted lanes and priced lanes, vehicle restrictions or variable pricing dampen peak demand and tend to maintain free-flow speeds in the peak periods. The speed estimates are used to estimate delay reductions relative to the base case No Build alternative. SMITE-ML also estimates the value of travel time savings for both "previous" travelers as well as diverted or "induced" travelers. The conversion of time savings to a monetary value was based on the average value of time of $9.00 per person hour, obtained by applying an inflation factor to the value estimated by US DOT (21). Results are presented in Table 4.
Table 3 also shows estimates of tolls paid by SOVs in the priced lanes. The average toll rate per mile was estimated using the estimate of average time saved per mile by vehicles in the priced lanes, and converting it to a monetary value. This value would roughly be equivalent to the toll that those ineligible for free service would be willing to pay.
Note that peak period toll rates would actually be much higher than suggested by this methodology, because peak period speed differentials are much higher than the average daily speed differentials estimated by SMITE-ML. (All pricing alternatives only propose to toll vehicles during the peak periods.) Also, toll rates may actually be higher than that indicated even by peak speed differentials, because toll-payers generally value their time more highly than the average motorist. That is why they choose to pay the toll. For example, commuters in the toll lanes in the median of SR 91 in Orange County, CA, value their time at $13 to $16 per hour (22). SMITE-ML multiplies the conservative estimates of average toll rates by the estimated number of toll-paying vehicles during peak periods (at Level of Service C), to provide a conservative estimate of daily toll revenues, as shown in Table 5. Annual revenues over all three segments are presented in Table 5. They are estimated assuming that tolls would be charged on 250 weekdays a year.
3.5 Estimates of Social Costs and Benefits
Table 4 presents estimates of benefits associated with the alternatives, for the southern segment of the freeway.
Construction delay costs
First, the Table presents excess delay per mile due to construction activities during project implementation. It was assumed that delays would increase by 100% above prior recurring congestion delay, over a period of 250 days. According to a recent study of the impacts of temporary losses of highway capacity by Oak Ridge National Laboratory (23), non-recurring delay on U.S. freeways and principal arterials in 1999 amounted to 2.02 billion vehicle hours, while work zone delays on freeways alone amounted to 0.48 billion vehicle hours. Various studies, such as those done by the Texas Transportation Institute (24), suggest that recurring delay is equal to or less than non-recurring delay. Thus, work zones region-wide add at least 25% to recurring delays on regional networks, and perhaps more if delays on principal arterials are considered. Since we can safely assume that no more than 25% of a region's network is in 11 work zone status at any one time, this suggests that the assumption of 100% increase in delay on a specific facility due to construction is a conservative assumption.
The increase in external costs per mile (including air pollution, noise and crashes) due to increased traffic were estimated using an estimated cost of 6 cents per vehicle mile for traffic in excess of the No Build alternative. These costs per vehicle mile were calculated based on the low-range nationwide estimates of these costs, amounting to $153.7 billion, and nationwide vehicle miles of travel amounting to 2.7 trillion in the year 2000 (25). The low-range estimates were used to ensure conservative estimates of benefits from pricing. Pricing tends to lower external costs, and therefore increase social benefits, since vehicle miles of travel are reduced relative to alternatives providing the same amount of capacity free of charge.
Assuming an average vehicle occupancy of 1.33, travel time delay costs amount to $12.00 per vehicle hour. In addition to the value of travel time saved by reduced delays, motorists save fuel as a result of reduced accelerations and decelerations. FHWA's Highway Economic Requirements System (HERS) model (26) estimates fuel consumption in relation to speeds. Based on the HERS model equations, ECONorthwest (27) calculated excess fuel consumed per minute of delay. On a facility with a free-flow speed of 60 mph, excess fuel consumed ranges from 0.037 gallons per minute of delay for a small car to 0.073 gallons per minute of delay for a sports utility vehicle (SUV). This equates to an added fuel cost of about 7 cents per minute of delay, including about 2 cents in gas taxes, assuming about $1.40 per gallon at the pump.
Since fuel taxes are a transfer, savings to motorists are losses to government agencies, and there is no net change in societal benefit from gas tax savings. Therefore, in computing societal benefits for the alternatives, changes in gas tax receipts were ignored. At 5 cents per minute of delay, fuel costs amount to $3.00 per vehicle hour of delay. Thus, fuel consumption costs from delay amount to about 25% of the travel time delay costs of $12.00 per vehicle hour.
In addition to time and fuel, motorists may incur crash cost savings when delay is reduced, as experience with the toll lanes on SR 91 suggests (22). However, due to lack of definitive data, and to ensure conservative estimates of benefits from pricing, these possible savings from pricing have been ignored for this case study.
3.6 Evaluation of Analysis Results
Table 5 presents key evaluation information for the six alternatives for all three freeway segments combined.
Table 5 first presents the delay reduction estimates (in person hours daily) for each of the six alternatives. The pricing alternatives reduce significantly more delay than alternatives with the same physical configuration but no pricing. Table 5 suggests that pricing Alternative 4, which does not increase capacity, is still able to reduce delay significantly. Delay reduced amounts to more than 75% of the delay reduction with conventional Alternative 1, which does increase capacity by one lane in each direction. Thus, if funding is not available in a timely fashion for capacity improvement alternatives, this alternative could be a very good interim solution. Alternative 5, which involves the same amount of new capacity and the same configuration as conventional Alternative 2, reduces delay by about 50% more than Alternative 2. Similarly, Alternative 6, which involves the same amount of new capacity and the same configuration as conventional Alternative 3, reduces delay by almost 50% more than Alternative 3.
Cost per Hour of Delay Reduced
The bottom line of Table 5 also presents the cost of each alternative per hour of delay reduced. As the table shows, the most cost-effective alternative with regard to delay reduction is Alternative 4. It costs only about $3 per hour of delay reduced. On the other hand, the conventional Alternatives 1, 2 and 3 and pricing Alternative 5 are the least cost-effective. They cost about $9 to 13 per hour of delay reduced.
Gross annual toll revenue is relatively lower for Alternative 6 primarily because toll rates are lower, since the four regular lanes are less congested and travel time saved by taking the priced lanes is correspondingly lower. Revenues shown in the Table are gross revenues. For Alternative 4, approximately 25% of the revenue will be needed to pay for credits to motorists in the regular lanes, assuming a credit payout of 25% of the toll rate.
Estimates of new transit trips from the mode choice model (see Table 1) were used to calculate new transit subsidies that would be needed to support service for the new trips and provide discount fares. Additional public subsidies were estimated at 50 cents per passenger mile, based on nationwide subsidies of $23.5 billion supporting 50 billion passenger miles annually (28). Estimates of new transit trips were used to calculate new transit subsidies that would be needed to support service for the new trips and provide discount fares. Table 5 shows the resulting transit subsidy estimates. Transit use under the pricing alternatives is about two and one-half times its use under the conventional alternatives. This results in a need for new public subsidies for transit which are about eight times the new subsidies needed for the conventional alternatives.
Annualized Public Costs
Table 5 presents estimates of annualized highway facility construction and right-of-way costs based on estimates from the Study report (2). These are then aggregated with annualized transit subsidy and pricing infrastructure and operation costs, to get total annualized public costs. Total annualized public costs, including capital costs for highway facilities, is higher for the pricing Alternatives 5 and 6, relative to the comparable "no pricing" Alternatives 2 and 3, primarily because of the larger transit subsidies needed.
Net Financial Impact
Despite the very conservative assumptions that were used in estimating toll revenue, the pricing alternatives bring in gross toll revenues estimated at about 70-75% of the new public subsidy needs for transit under Alternatives 4 and 5, and about 25% of that needed for Alternative 6. If alternative sources can be found for transit funding, the toll revenue will be available to provide a source of funding to support construction bonds for the highway improvements.
The estimated excess travel delay costs during project construction and external costs from increases in vehicular travel relative to the No Build alternative were subtracted from estimates of user benefits (i.e., time and fuel savings) to get net annual benefits. The present value of benefits over a 20-year period was estimated assuming a 7% discount rate (29).
Net Present Value
The present value of public costs for a 20-year period is then subtracted from the present value of benefits aggregated over all three segments, to get net present value. The pricing alternatives all demonstrate significant positive net present values, ranging from $517 million to $ 1.7 billion. Only one of the three conventional alternatives, Alternative 3, has a positive net present value amounting to $138 million. However, this is much lower than any of the pricing alternatives.
4.0 CONCLUSIONS AND RECOMMENDATIONS
This paper has demonstrated that relatively simple analytical procedures may be used to estimate the impacts of pricing alternatives and generate information in a timely fashion for use by local decision-makers. The case study has also demonstrated that pricing alternatives can often accomplish the purpose of a major highway project more efficiently and more effectively than conventional alternatives that exclude pricing, while generating net revenues to support alternative modes, bonds for timely project construction, and other transportation priorities. Delays due to constrained funding can be avoided, more transportation choices can be provided, and the public can get superior mobility earlier and at lower public cost.
The author would like to acknowledge valuable comments received from Dr. Kiran Bhatt of K.T. Analytics, Dr. Bruce Spear of FHWA, and anonymous reviewers from the TRB Committee on Taxation and Finance. However the author alone is responsible for any errors or omissions, and the views expressed are those of the author and not necessarily those of the U.S. DOT or the FHWA.
LIST OF TABLES AND FIGURES
FIGURE 1. EQUILIBRATING DEMAND AND PRICE
TABLE 1. TRAVEL DEMAND ESTIMATES FOR TRAVEL CORRIDOR - SOUTHERN SEGMENT
TABLE 2. ESTIMATES OF DIVERTED AND INDUCED TRAFFIC - SOUTHERN SEGMENT
TABLE 3. AVERAGE DAILY SPEED, DELAY AND TOLL REVENUE PER MILE - SOUTHERN SEGMENT
TABLE 4. SOCIAL COSTS AND BENEFITS PER MILE - SOUTHERN SEGMENT TABLE 5. SUMMARY OF IMPACTS OF ALTERNATIVES