Estimating the Benefits from Mileage-based Vehicle Insurance, Taxes and Fees

Transcription

1 Paper No Estimating the Benefits from Mileage-based Vehicle Insurance, Taxes and Fees Prepared for presentation at the TRB 2002 Annual Meeting in January 2002 Paper Word Count: 5984 Abstract Word Count: 193 Patrick DeCorla-Souza, AICP Team Leader FEDERAL HIGHWAY ADMINISTRATION Office of Transportation Policy Studies, HPTS 400 Seventh St. SW, Room 3324 Washington, DC Tel: (202) Fax: (202) November 15, 2001 DISCLAIMER: The views expressed in this paper or those of the author, and not necessarily those of the US Department of Transportation (U.S. DOT) or the Federal Highway Administration (FHWA)

2 DeCorla -Souza 2 Abstract Estimating the Benefits from Mileage-based Vehicle Insurance, Taxes and Fees Patrick DeCorla-Souza, AICP Mileage-based pricing involves variabilization strategies, i.e., strategies such as pay-asyou-drive insurance or car-sharing which are aimed at converting the fixed costs of driving to variable costs. They do not directly involve pricing of highway facilities. They nevertheless can reduce congestion by inducing some drivers to change their mode of travel or drive shorter distances. Variabilization strategies are particularly promising since about 80% of the cost of driving is fixed. Once a vehicle has been purchased, and taxes, fees and insurance paid, there is little financial incentive not to use it heavily. How effective are such variabilization strategies relative to highway facility pricing strategies, such as pricing of added highway capacity? This paper estimates the congestion reduction and other economic benefits of policies to variabilize the fixed costs of vehicle use and compares them to conventional highway facility expansion and a pricing strategy involving adding a lane and pricing it for free flow. The analysis suggests that a nationwide variabilization policy which adds 10 cents per mile to the perceived variable cost of vehicle use (without increasing total vehicle use costs) could produce a 20-year stream of benefits conservatively estimated at over $44 billion.

3 DeCorla -Souza 3 Estimating the Benefits from Mileage-based Vehicle Insurance, Taxes and Fees Patrick DeCorla-Souza, AICP INTRODUCTION There is increasing support for the use of variable pricing of roads (also known as value pricing or congestion pricing) to help address problems such as traffic congestion and pollution. Variable pricing of roads includes a variety of strategies that charge motorists directly for road use including variable tolls on toll facilities, i.e., tolls which vary based on the level of demand; variable pricing on existing or added highway lanes; and mileage-based charges, i.e., charges based on how much a vehicle is driven, regardless of location and time. In this paper we discuss a type of road pricing involving the variabilization of existing fixed vehicle charges, that is, converting fixed vehicle charges such as taxes, insurance, registration and lease fees into distance-based charges. For example, assuming that the average motor vehicle is driven 12,000 miles annually, an annual vehicle insurance premium of $683, i.e., the average premium in 1999 (1) would convert to 5.7 cents per mile; a $240 annual vehicle license fee or property tax becomes a 2 cents per mile charge; and vehicle purchase taxes of $1,200 would convert to 3.3 cents per mile if paid over the first three years of a vehicle s operating life. Such strategies price vehicle use without directly pricing highway facilities. They nevertheless could reduce congestion (and thereby reduce pollution and energy consumption) by inducing some drivers to change their mode of travel or drive shorter distances. Existing fixed prices violate market principles, since costs such as costs associated with crash risks depend on usage and increase with vehicle mileage. About 80% of the cost of driving is fixed (2). Once a vehicle has been purchased, and taxes, fees and insurance paid, there is little financial incentive not to use it heavily. Distance-based pricing (variabilization) helps correct existing market distortions. It is justified on basic economic principles as a way to support increased economic efficiency. It can also be justified on equity grounds, because under existing fixed charging systems low mileage drivers overpay insurance and road costs while high-mileage drivers underpay these costs. Since lower-income motorists drive their vehicles significantly less than higher-income motorists, fixed charging systems can be shown to be regressive. Distance-based pricing can also make insurance more affordable, and reduce the cost of new vehicle purchases inducing drivers to buy newer, less polluting and safer vehicles. The Online TDM Encyclopedia maintained by the Victoria Transportation Policy Institute at provides more information on these benefits. Distance-based pricing requires collection of vehicle travel data. This can be done in one of three ways: (a) using odometer audits performed periodically (e.g., once a year); (b) using an electronic on-board data collection unit that records vehicle mileage and transmits the results electronically or manually; or (c) using a Global Positioning System (GPS) to track the travel of each vehicle.

4 DeCorla -Souza 4 Under the Value Pricing Pilot Program of the Intermodal Transportation Efficiency Act for the 21 st Century (TEA-21), three variabilization projects are just getting off the ground in Atlanta, Georgia; in Minneapolis-St. Paul, Minnesota; and in Boston, Massachusetts. The Atlanta and Boston projects involve pay-as-you-drive (PAYD) insurance (3, 4), and the Minnesota project involves conversion of fixed vehicle-lease costs and fixed vehicle registration taxes and fees to mileage-based charges (5). Results from these experiments will begin to become available in the near future. Variabilization strategies essentially achieve the same types of benefits as highway facility pricing strategies. However, an important difference is that variabilization is not a new charge. It simply changes how consumers pay for an existing charge in order to give motorists a new opportunity to save money. Motorists who reduce their mileage can save money in a way that is not currently possible. In other words, variabilization returns to individual motorists the economic savings that result when they drive less. Mileage reductions that result from variabilization represent low-value travel that consumers would rather forego in exchange for financial savings, with consequent net consumer surplus gains. Travel that consumers consider of higher value would continue, and users will be better off due to reduced congestion. In the U.S., one highway facility pricing strategy that has proved to be politically feasible is pricing of new highway lanes. This involves making peak period motorists pay more of the true costs of capacity to accommodate them. It reduces inequities to off-peak motorists and others, who must otherwise subsidize costs of new capacity for peak period motorists. With dynamic pricing, pricing of new lanes also keeps added lanes congestion-free, protecting the public investment, and reducing pollution and fuel consumption. However, the biggest benefit of priced lanes, from the standpoint of the public, is that priced lanes provide an option for premium service which a motorist can use when he or she needs to avoid delays in order to be on time for a business or social appointment. Priced new lanes also reduce travel induced by highway expansion, a major concern of environmental groups. On privately operated new priced lanes on SR91 in Orange County, CA, variable tolls have been successfully used to manage demand, while at the same time ensuring premium service and recovery of the private sector s investment in new capacity (6). The Federal Highway Administration s (FHWA) Value Pricing Pilot Program is also supporting pre-implementation studies of new priced lanes in San Diego (I-15 HOT lanes extension) ; in Santa Cruz County, California on State Route 1; and in Denver, Colorado on Route C-470 (7, 8). This paper uses an illustrative example to compare estimated benefits from a variabilization approach with those from a conventional highway expansion and a new priced lane approach. It then discusses implications for metropolitan transportation decision-making. ESTIMATING BENFITS FROM VARIABILIZATION STRATEGIES Travel demand models calibrated to estimate the effects of transportation investments and policies on metropolitan travel have clearly shown a lack of sensitivity of travel demand to fixed vehicle use costs. The models show that while travel time and out-of-pocket costs (e.g., parking charges and tolls) are fully considered by motorists in making marginal travel decisions, other costs such as the fixed costs of vehicle-ownership (including monthly payments for car loans, annual or bi-annual payments for insurance or property taxes and annual vehicle registration fees) are ignored. This is because the fixed amounts that the motorist pays are generally

5 DeCorla -Souza 5 unaffected by his or her marginal travel decision. If policies can be instituted to convert a portion of fixed private costs to variable costs, motorist behavior may change and the quantity of travel demanded may be reduced. This will in turn affect the magnitude of congestion delay and other external costs of travel, and aggregate social benefits. Private fixed costs of vehicle use for a vehicle driven 10,000 miles a year amount to about 40 to 60 cents per mile driven on average, or about 80% of the full cost of about 50 to 73 cents per mile (9). The variable portion of vehicle use costs averages 9.5 to 12.4 cents per mile, comprising fuel, oil, maintenance and tires, based on AAA data (9) updated to 2000 dollars. As discussed above, although fixed costs are borne directly by motorists, they are not perceived by them at the margin and are therefore not taken into consideration in their marginal decisions to travel from day to day. Therefore, when estimating vehicle travel demand and user cost functions, only variable vehicle use costs should be used, and fixed costs should be ignored. As shown in Figure 1, the demand function or demand curve relates travel demand to user cost (i.e., price) and the user cost function or price curve relates user cost to the quantity of travel. Variabilization of fixed costs results in a shift in the price curve, affecting volume of travel demanded, as shown in Figure 2. Essentially, variabilization of the fixed costs of insurance, registration and taxes increases variable vehicle expenses without increasing total vehicle use costs. Conventional user benefit analysis can be used to estimate consumer welfare changes that result (10). To complete a social benefit analysis, external benefits and costs must be accounted for, in addition to private benefits and costs. A cost is considered to be "external" when it is borne by someone other than the traveler, and the traveler does not have to compensate those who bear the cost. For example, air pollution costs are considered external when they are borne by someone other than the driver who pollutes, and the driver who pollutes provides no compensation or inadequate compensation to those who suffer losses due to the pollution. To the extent a driver pays for losses to others, those costs are considered to be internal or private costs. As in the case of fixed internal costs, external costs are generally not taken into consideration by the traveler in making a short-run decision to travel. Therefore, for a full accounting of social benefits from a transportation improvement or policy change, external cost changes must be accounted for separately, using procedures outside of conventional private userbenefit analysis, and then combined with estimated private user-perceived benefits to get aggregate social benefits (10). Delucchi (11) produced original lower bound and upper bound estimates of U.S. nationwide social costs of motor vehicle use. These include monetary externalities, such as uncompensated lost wages due to incapacitation from accidents; and non-monetary externalities, such as air pollution. Delucchi also estimated the magnitude of bundled costs, i.e., costs paid for indirectly by travelers. Bundled costs include private costs such as parking costs at shopping centers or work sites. They also include public costs such as highway patrol; municipal off-street parking; judicial, police and fire protection costs related to motor vehicle use; costs for the Strategic Petroleum Reserve; military expenditures related to use of Persian Gulf oil; costs for regulation of air pollution, water pollution, and solid waste related to motor vehicle use; costs for energy and technology research and development related to motor vehicle use; and a portion of roadway costs. One can estimate average external and bundled costs per vehicle mile based on Delucchi s national estimates. Delucchi s estimates of annualized nationwide social costs of

6 DeCorla -Souza 6 vehicle use are presented in Table 1. Total costs range from $1,658 billion to $3,272 billion (1991 $). They amount to $0.76 to $1.50 per vehicle mile of travel (VMT) in 1991 dollars. Delucchi s lower bound estimate for bundled costs (excluding roadway costs) amounts to 5.1 cents per VMT in 1991 dollars. This includes 3.5 cents in private costs and 1.6 cents in public costs. His upper bound estimate for bundled costs per VMT amounts to 15.8 cents, including 12.9 cents private and 2.9 cents public costs, in 1991 dollars. His lower and upper bound estimates for external costs amount to 4.5 cents and 40 cents respectively, in 1991 dollars. In the initial analysis presented in this paper, Delucchi s lower bound estimates of external and bundled costs are used. An evaluation using his upper bound cost estimates is presented in the summary at the end. Since bundled costs are not paid for directly by motorists on a per trip basis, and therefore not taken into consideration in their driving decisions, they are treated in the same manner as external costs. The estimate of total lower bound bundled and external costs amounts to 9.6 cents in 1991 dollars, or about 12 cents in 2000 dollars. The estimate of total upper bound bundled and external costs amounts to 55.8 cents in 1991 dollars, or about 70 cents in 2000 dollars. Bundled and external cost changes are estimated and combined with estimates of private benefits to get total social benefits. PROTOTYPICAL EXAMPLE: BASE CASE This section presents an analysis of travel demand and user costs for a typical scenario in a metropolitan area, involving a severely congested 4-lane freeway facility with 2 lanes in each direction. To simplify the analysis, only passenger vehicles are considered. Commercial truck traffic, which normally comprises about 5% of peak period traffic, is ignored. In the following sections, some alternatives to reduce congestion on the freeway are presented and their relative benefits are estimated. Currently, the variable costs for use of a highway facility are comprised of : (a) a travel time cost, which tends to increase significantly with higher traffic volumes under congested conditions; and (b) a vehicle operating cost comprised mainly of fuel costs, which also varies with traffic volume, but less significantly. Figure 1 graphically presents the peak period user cost curve X1 for the peak direction of the 4-lane freeway facility. The value of time used in developing the price curve is based on the U.S. Department of Transportation recommended value of time of $8.90 per person hour for local travel and $12.20 per person hour for inter-city travel, in 1995 dollars (12). These values equate to $9.70 and $13.30 respectively in year 2000 dollars. For this analysis, all peak travel on the freeway is assumed to be local, to ensure conservative estimates of benefits. Based on 1995 national data (13), the occupancy rate for all travel is 1.59 person miles per vehicle mile, ranging from a low of 1.14 for work to 2.17 for other social and recreational purposes. For convenience in the analysis, and since typically a predominant share of morning and afternoon peak period trips is commute trips with an average vehicle occupancy of 1.14, an average occupancy in the peak periods of 1.24 has been assumed for this analysis, rounding off the resulting value of time to $12 per hour (i.e., $9.70 times 1.24). At a value of $12.00 per vehicle hour, 20 cents equates to one minute of vehicular travel time. Under free-flow (up to 3,200 vehicles per hour for both lanes), the freeway operates at an average speed of 60 mph or 1 minute per mile of travel. Thus travel time cost under free flow equates to 20 cents per mile. As discussed earlier, the variable portion of vehicle operating cost

7 DeCorla -Souza 7 averages 9.5 to 12.4 cents per mile. For convenience, a vehicular operating cost of 10 cents per mile is used in this analysis. Thus, under free flow conditions, total variable cost per mile is about 30 cents per mile, comprising 20 cents for travel time and 10 cents for vehicle operating cost. Based on the widely used Bureau of Public Roads (BPR) traffic volume-to-delay relationship (14), as travel demand volume on the freeway increases beyond 3,200 vehicles per hour for both lanes, two changes occur. First, delay increases rapidly, reaching a magnitude of 1 minute per mile (equating to 20 cents) at a volume of 4,000 vehicles per hour when traffic flows at 30 mph, and an additional 2 minutes (equating to 40 cents) at a volume of 4,500 vehicles per hour when vehicles crawl at a speed of 15 mph. It should be noted that these travel times and the corresponding time price curve reflect average costs over all vehicles. Marginal costs (per added vehicle) increase at a much more rapid rate than average costs and are generally more than twice average costs under congested conditions. Fuel consumption also increases with delay. Based on the Highway Economic Requirements System (HERS) model equations (15), excess fuel consumed per minute of delay on a facility with a free-flow speed of 60 mph ranges from gallons for a small car to gallons for a sports utility vehicle (SUV). This equates to an added user cost of about 5 cents per minute of delay. Thus total vehicle operating costs increase to 15 cents per mile at a volume of 4,000 vehicles (30 mph), reaching about 25 cents per mile at a volume of 4,500 vehicles per hour. As indicated in Figure 1, the freeway carries 4,500 vehicles during the peak hour in the peak direction. At a travel time of 4 minutes per mile (15 mph), travel time cost equates to 80 cents per mile. Due to excess fuel consumption, vehicle operating cost is 25 cents per mile, 15 cents above the 10 cents per mile average for operation under free-flow conditions. Figure 1 also shows a demand curve representing one-way peak hour travel demand on the severely congested 4-lane freeway. The demand curve reflects a travel demand elasticity of This is the short-term demand elasticity recommended by the Standing Advisory Committee on Trunk Road Assessment, commonly known as SACTRA (16). This means that a 1% increase in travel cost would result in a 0.5% reduction in travel demand. It should be noted that this is an average elasticity. Elasticity will be higher when travel alternatives are available. For example availability of transit service to employment centers during rush hours can increase highway travel demand elasticity. On the other hand, when reasonable alternatives are not available, such as for suburb-to-suburb social-recreation trips during the off-peak hours, demand elasticity may be lower. Since the analysis in this paper focuses on congested rush-hour periods, estimates of demand changes are conservative. ALTERNATIVE 1: VARIABLIZE FIXED VEHICLE USE COSTS Now let us take an alternative involving variabilization of the fixed costs of vehicle use. As Figure 2-A shows, when a portion of the fixed cost of vehicular travel is converted to a variable cost perceived at the margin, the private cost (i.e., price) curve shifts to the left. In Figure 2-A, the private cost (i.e., price) curve X2 reflects the case where the perceived internal cost per vehicle mile is increased by 10 cents per mile, by converting some of the fixed costs of vehicle use to variable costs. This reflects institution of a metropolitan region-wide policy converting the following fixed costs of vehicle use to a mileage basis: (1) annual vehicle property taxes and fees; (2) the portion of annual real-estate taxes used for transportation

8 DeCorla -Souza 8 purposes; (3) vehicle sales taxes and (4) annual or semi-annual vehicle insurance premiums. The conversion of annual insurance premiums to pay-as-you-drive would be encouraged through tax incentives to insurers and/or motorists. As Figure 2-A shows, the result of the variabilization policy is that the quantity of travel demanded is reduced to 4,450 vehicles per hour in the peak hours, and total variable travel cost increases from 105 to107 cents per mile. The change in travel demand is estimated using a midpoint elasticity formulation (17). The cost includes the variabilized cost of 10 cents per mile which is not really an additional cost to motorists (as a group), over and above what they would have paid in the base case through fixed insurance premiums, taxes or fees. With variabilization, they will see an equivalent (or larger) reduction in their fixed charges. The potential for a larger reduction stems from the reduction in accident risk which would result from reductions in quantity of travel. The 10 cent cost can be considered to be merely a transfer from fixed to variable cost, with a net financial impact of zero on motorists as a group. (Note, however, that the policy would cause transfers among individual owners/operators of motor vehicles.) Thus, the real cost to motorists as a group averages only 97 cents per mile. The social benefits per mile (for two lanes) under this alternative are computed as follows: Consumer surplus benefits = ( )/2 * ( )c = $ Fixed cost savings = 4,450 * 10 cents = External benefits = (4,500*12c) - (4,450*12c) = 6.0 Total benefits = Note that external benefits are calculated using the lower bound estimates of external and bundled costs amounting to 12 cents per mile, as explained earlier. The varibilization policy will also affect other highways not as heavily congested as the example 4-lane freeway. Calculation of benefits in the peak hour for a single direction of a typical less-congested 4-lane freeway is shown in Figure 2-B. As Figure 2-B shows, the result of the variabilization policy is that the quantity of travel demanded is reduced from 4,000 to 3,880 vehicles per hour in the peak hours. The demand reduction is achieved at a total variable travel cost of 58 cents per mile, a 3 cent increase above the variable cost of 55 cents per mile at the previous volume of travel. However, since the cost includes the variabilized cost of 10 cents per mile which is not really an additional cost to the motorists as a group (as explained earlier), the real cost to the motorist on average is only 48 cents. The additional social benefits per mile are computed as follows: Consumer surplus benefits = (3, )/2 * (55c-58c) = $ Fixed cost savings = 3,880 * 10 cents = External benefits = (4,000*12c) - (3,880*12c) = 14.4 Total benefits = Assuming that for every severely congested freeway mile there will be one less-congested freeway mile similar to the example above, total peak hour benefits from Alternative 1 per mile of severely congested facility in a metropolitan area will be: Benefits per mile of severely congested facilities = $ 361.5

9 DeCorla -Souza 9 Benefits per mile of less-congested facilities = Total areawide benefits per severely congested mile = ALTERNATIVE 2: ADD FREE GENERAL PURPOSE LANE To compare the predicted results of Alternative 1 with a conventional congestionreduction strategy, a second alternative is considered, involving an upgrade of the freeway to a 6- lane facility. Figure 3 shows the corresponding user cost curve X3. Based on the travel demand function, travel volume is anticipated to increase to 6,150 vehicles per hour. The increase is due to diverted travel and travel previously suppressed due to the heavily congested peak hour conditions. Travel time is reduced to 2 minutes per mile (30 mph), a reduction of 2 minutes relative to the base case. This equates to a cost of 40 cents per mile. Vehicle operating cost is 15 cents per mile, a reduction of 10 cents from the base case. Thus, total variable cost per mile is 55 cents, a reduction of 50 cents relative to the base case. The social benefits per mile under this alternative are computed as follows: Consumer surplus benefits = (4, ,150)/2 * (105-55)c = $ External benefits = (4,500*12c) - (6,150*12c) = -198 Total benefits = ALTERNATIVE 3: ADD PRICED LANE For comparison of the variabilization strategy with a facility-based pricing strategy, a third alternative is evaluated. Outright tolling of an existing free lane or the entire existing facility, while it might produce substantial social benefits, is not generally considered to be politically acceptable in the U.S. More politically acceptable would be an alternative which involves adding a priced lane in each direction, providing free service for transit buses, but charging other vehicles a toll per mile bearing some relation to the variable user costs saved. Such a lane would be similar to the priced Express lanes on SR 91 in Orange County, CA described earlier. For ridesharing vehicles, the toll charge would be shared by the travelers in the vehicle. Figure 4 shows the corresponding private cost curves for the two sections of the freeway facility (priced and unpriced) during peak hours. It is assumed that 1,600 vehicles would use the priced lane, the maximum that can be reliably accommodated (in addition to transit buses) at a travel speed of 60 mph (i.e., travel time of 1 minute per mile). The toll would be set high enough to keep volumes at 1,600 vehicles per hour to eliminate congestion. It is estimated that a total of 2,230 travelers, previously accommodated in 1,800 vehicles, would shift to the priced lane. However, because of the incentive to share rides, it is estimated that some of these travelers would form additional carpools or ride transit. The resulting average vehicle occupancy is estimated to be persons per vehicle, so that the 2,230 travelers would be accommodated in 1,600 private vehicles or transit vehicles. The increase in vehicle occupancy reflects approximately a 50% increase in ridesharing on the added lane. By comparison, the average travel impact of parking cash-out policies at eight California case study sites was an increase in the carpool mode share from 12.9% to 20%, i.e., a greater than 50% increase in ridesharing. Transit use increased from 5.8% to 8.3%, i.e., a 43% increase (17).

10 DeCorla -Souza 10 Solo-drivers who shift to carpooling in order to receive cash in lieu of free parking save roughly the same amount as they would by carpooling to save tolls. For this analysis it is assumed that the toll on the priced lane for non-transit vehicles would be about 25 cents per mile, an amount slightly higher than the total of vehicle operating cost savings of 4.5 cents and the value of time saved (0.9 minute, valued at 18 cents), relative to the unpriced lanes. The toll would be higher than savings estimated on the basis of the average value of time, because those who choose to use the priced lane would have a higher value of time than the average. (That is why they choose to use the priced lane). As Figure 4-A shows, the shift of some travel to the priced lane causes the demand curve for the unpriced lanes to shift to the left, so that demand would be reduced to 2,700 vehicles per hour, if user cost stayed at 105 cents per mile. However, based on the travel demand function, the quantity of travel demanded would actually be 3,780 vehicles per hour, at a user cost of 52.5 cents, comprising a travel time cost of 1.9 minutes per mile (valued at 38 cents) and a vehicle operating cost of 14.5 cents. The social benefits for the regular lanes are estimated as follows: Consumer surplus benefits = ( )/2 * ( ) c = $ External benefits = (4500*12c) - (3780*12c) = Total benefits = $ To be conservative, the social benefits for the priced lane are computed ignoring the benefits to previous bus passengers. The benefits on the priced lane are computed as follows: Travel time savings = 2230 * (60/1.24) c per person = $1079 Vehicle operating cost savings = 1,600 * 15c = $240 Tolls paid = 1,600 * 25 cents = -$400 Toll revenues = 1,600 * 25 cents = +$400 External benefits = -(1,600) * 12c = -192 Total benefits = $1127 Total peak hour benefits from Alternative 3 will be: Benefits from regular (unpriced) lanes = $ Benefits from priced lane = $1127 Total benefits = $ SUMMARY OF BENEFIT ESTIMATES Table 2 summarizes the estimates of benefits from the three alternatives. The table first presents the estimates in previous sections which were calculated using Delucchi s lower bound estimates of external and bundled costs. Next estimates are presented reflecting Delucchi s upper bound estimates of external and bundled costs. Peak hour benefits are converted to average annual weekday benefits using a factor of 5.1. This factor is based on the relationship between benefits in the peak hour of the peak month and average annual weekday benefits, calculated using the draft AASHTO Redbook update (18), for a peak hour volume equal to 9% of daily

11 DeCorla -Souza 11 traffic volume, and a ratio of peak month average daily traffic volume (ADT) to average annual ADT of 1.1. Annual benefits are estimated as 250 times average annual weekday benefits, based on number of working days in the year. Weekend benefits are ignored to ensure conservative estimates. The analysis results presented in Table 2 are based on a short-run travel demand curve. However, there is strong evidence that, in the long-run, the demand curve may actually shift. For example, many things can change in the long-run, as a result of variabilization strategies -- vehicle ownership rates could be affected, location decisions and land use patterns could change, and even the availability of alternative transportation modes can be affected, affecting travel demand elasticities. Therefore, the long-run (20-year) extrapolation of short-term benefits may be inappropriate. However, most of the factors which affect long-run shifts in demand are likely to increase net benefits to society. We currently do not have any data on the basis of which to develop valid models to forecast long-run shifts in the demand curve. However, FHWA s Value Pricing Pilot Program has three variabilization simulation projects (3, 4, 5) which could shed light for improved modeling of these longer term effects. IMPLICATIONS FOR METROPOLITAN AND NATIONAL POLICY As Table 2 shows, variabilization strategies could achieve approximately $0.82 to $0.95 million in benefits per year per direction for each mile of severely congested 4-lane freeways, or $1.6 to $1.9 million per mile for both directions. This amounts to a present value of about $16 to $19 million per mile of severely congested 4-lane freeway at the federally recommended discount rate of 7% (19), assuming a 20-year stream of benefits. Thus, in an urban area with just 10 miles of severely congested freeway, an up-front expenditure of up to $160 million in public funds for implementation of variabilization strategies could be economically justified. Nationally, there are 2,780 miles of severely congested urban freeways and expressways, i.e., those with a volume to service flow ratio above 0.95 (20). Even if we conservatively assume that none of these facilities have more than 4 lanes, the present value of a 20-year stream of benefits for these 2,780 miles would exceed $44 billion if variabilization strategies of the type evaluated in this paper could be implemented nationwide. This suggests that public expenditures of up to $44 billion could be justified to implement a policy which adds 10 cents per mile to the perceived variable cost of vehicle use. It should be noted that the estimated $44 billion in benefits on major urban highways is a conservative estimate of total social benefits. Conservative assumptions were used in the analysis. For example, benefits to transit vehicles and trucks and benefits on weekends were ignored, conservative estimates were used for value of time savings and travel demand elasticty, and all severely congested freeways were assumed to have only four lanes. Also, congestion reduction benefits on surface streets were ignored. Unlike tolls on major highways, which do nothing to reduce congestion on surface streets and may in fact make it worse due to diversion of traffic from tolled facilities, variabilization can reduce congestion on all roads. Also the strategy can result in reductions in privately borne crash costs as people reduce their driving. (These costs were not included in the external cost estimates in the analysis). Delucchi s estimates of total nationwide annual accident costs in the U.S. priced in the private sector are about $85 to 110 billion (in 2000 dollars). Thus, even a 1% reduction in crashes due to reduced driving

12 DeCorla -Souza 12 would result in crash reduction savings of about $1 billion annually (with a present value of a 20- year stream of savings of more than $10 billion at a 7% discount rate). Despite the large public expenditures that could be economically justified to implement variabilization strategies especially in the absence of other effective and politically acceptable pricing initiatives, they may actually be obtained without the type of huge public expenditures usually necessary for conventional capacity expansion or adding new priced lanes. For example, the construction costs in large urban areas (excluding right-of-way costs) average about $8.8 million per lane mile in year 2000 dollars for high cost lane additions (15). Assuming a total capital cost (including right-of-way) of $10 million per lane mile and a 20-year life, the annualized capital cost for a typical lane addition in large urban areas is about $1 million per mile at a discount rate of 7%. The estimates using upper bound external costs in Table 2 suggest that net social benefits (i.e., total social benefits less highway agency costs) for a capacity expansion alternative involving adding a free general purpose lane in high cost situations will be far less than for a variabilization policy, even if highway operation and maintenance costs and environmental costs of highway development are ignored. (Note that Delucchi s cost estimates only include environmental costs of highway use.) Thus, in high cost situations, and especially where environmental costs are high (such as in built up areas), variabilization strategies may be a more efficient tool to address congestion, make better use of expensive highway resources, manage the operation of the highway system, and postpone the need for highway expansion. However, where population, economic development, and consequent transportation needs are growing, the demand curve will shift to the right, and demand management strategies such as variabilization will not be sufficient to address access needs. In such cases transportation capacity enhancements will be needed, although variabilization can still play a significant role in postponing the need to expand capacity and in ensuring efficient use of added capacity after it is put in place. CONCLUSIONS The analysis in this paper has demonstrated that strategies which convert the fixed costs of vehicle use to variable costs can be an efficient solution to address congestion, make better use of expensive highway resources, manage the operation of the highway system, and postpone the need for highway expansion. The analysis suggests that a nationwide variabilization policy which adds 10 cents per mile to the perceived variable cost of vehicle use (without increasing total costs) could produce a 20-year stream of congestion-reduction benefits conservatively estimated at over $44 billion. ACKNOWLEDGEMENTS The author would like to acknowledge valuable comments received from Todd Litman of the Victoria Transportation Policy Institute, Kiran Bhatt of K.T. Analytics, Allen Greenberg of FHWA, and anonymous members of the TRB Committee on Transportation Economics. They provided helpful suggestions in improving this paper. 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.

13 DeCorla -Souza 13 REFERENCES 1. National Association of Insurance Commissioners. State Average Expenditures & Premiums for Personal Automotive Insurance in Kansas City, MO. May Litman, Todd. Distance-Based Vehicle Insurance. Victoria Transportation Policy Institute, Victoria, BC. November Guensler, R. and Ogle, J. Commuter Choice and Value Pricing Insurance Incentive Prigram. Georgia Institute of Technology. Atlanta, GA. Submitted to FHWA. May (available at: 4. Broad, Martha and Joseph Fereira, Jr. Proposal for a Usage-based Pricing Simulation in Massachusetts. Submitted to FHWA. June (available at: 5. Minnesota DOT, Managing Congestion Through Consumer Choice in Transportation Costs. Submitted to FHWA. May (available at: 6. US DOT, Report on the Value Pricing Pilot Program. Washington, DC, July (available at: 7. FHWA, Value Pricing Notes. Vol.5, Fall Washington, DC. (available at: 8. FHWA, Value Pricing Notes. Vol.7, Fall Washington, DC (available at: 9. AAA and Runzheimer International. Your Driving Costs Edition US DOT. Participant Notebook for NHI course no , Estimating the Impacts of Urban Transportation Alternatives. Publication No. FHWA-HI Washington, DC. December Delucchi, Mark. The Annualized Social Costs of Motor Vehicle Use in the U.S, : Summary of Theory, Data, Methods and Results. Institute of Transportation Studies, Davis, CA. June US DOT. Memorandum on Departmental Guidance for Valuation of Travel Time in Economic Analysis. Washington, DC. April 9, US DOT. Our Nation s Travel: 1995 NPTS Early Results Report. Washington, DC. September US DOT. Participant Notebook for NHI course no , Advanced Travel Demand Forecasting Course. Publication No. FHWA-HI Washington, DC. December US DOT Status of the Nation s Highways, Bridges and Transit: Conditions and Performance. Washington, DC. (available at: Standing Advisory Committee on Trunk Road Assessment (SACTRA) Trunk Roads and the Generation of Traffic. Conducted for the Department of Transport, London, England. December TRB. Traveler Response to Transportation System Changes. Interim Handbook. TCRP Web Document 12 (Project B-12). Washington, DC. March 2000.

14 DeCorla -Souza TRB. User Benefits Analysis for Highways. Draft Report. NCHRP Project Washington, DC. June OMB. Benefit-Cost Analysis of Federal Programs: Guidelines and Discounts. Circular A-94 revised. Federal Register, November 10, Washington, DC. 20. US DOT. Highway Statistics Publication # FHWA-PL Washington, DC. November LIST OF TABLES TABLE 1 TABLE 2 The Social Costs of Motor Vehicle Use in the U.S. Summary of Estimated Benefits LIST OF FIGURES FIGURE 1 FIGURE 2 FIGURE 3 FIGURE 4 Demand and Price Curves: Base Case. Alternative 1 - Variabilization Policy: (A) Effect on Severely Congested Freeway; (B) Effect on Less Congested Freeway Alternative 2 - Add Free General Purpose Lane Alternative 3 - Add Priced Lane: (A) Effect on Unpriced Freeway Lanes; (B) Effect on Priced Freeway Lane.

15 DeCorla -Souza 15 TABLE 1 The Social Costs of Motor Vehicle Use in the U.S. Lower bound Costs Private (incl. travel time) Public road costs Bundled private Bundled public (except roadway costs) External TOTAL Upper bound Costs Private (incl. travel time) Public road costs Bundled private Bundled public (except roadway costs) External TOTAL Nationwide Cost ($ billion, 1991) 1, ,658 1, ,272 Cost per VMT (1991 cents)

16 DeCorla -Souza 16 TABLE 2 Summary of Estimated Benefits Using Lower Bound External & Bundled Costs Alternative 1: Variablize fixed vehicle costs Alternative 2: Add free general purpose lane Alternative 3: Add priced lane Peak hour (dollars) Daily (dollars) 3,293 12,569 14,863 Annual (million $) Using Upper Bound External & Bundled Costs Alternative 1: Variablize fixed vehicle costs Alternative 2: Add free general purpose lane Alternative 3: Add priced lane ,796 7,688 12,

17 DeCorla -Souza 17 Demand and Price Curves 160 Cents per Vehicle Mile User Cost Curve X1 Demand Curve vehicles Vehicle Operating Cost Peak Hour Traffic Volume FIGURE 1 Demand and Price Curves: Base Case.

18 DeCorla -Souza 18 A. Effect on Severely Congested Freeway Cents per Vehicle Mile "real" cost User Cost Curve X2 User Cost Curve X1 Demand Curve 4,450 Vehicles Peak Hour Traffic Volume B. Effect on Less Congested Freeway Cents per Vehicle Mile Loss of Consumer Surplus "real" cost User Cost Curve X2 User Cost Curve X1 Demand Curve Peak Hour Traffic Volume 3,880 Vehicles 4,000 Vehicles FIGURE 2 Alternative 1 - Variabilization Policy: (A) Effect on Severely Congested Freeway; (B) Effect on Less Congested Freeway

19 DeCorla -Souza 19 Effect of Adding Free General Purpose Lane Cents per Vehicle Mile Consumer Surplus User Cost Curve X Peak Hour Traffic Volume Demand Curve User Cost Curve X3 4,500 Vehicles 6,150 Vehicles FIGURE 3 Alternative 2 - Add Free General Purpose Lane

20 DeCorla -Souza 20 A. Effect on Unpriced Freeway Lanes Cents per Vehicle Mile Consumer Surplus ,700 Vehicles Peak Hour Traffic Volume Prior Demand Curve Shifted Demand Curve 3,780 Vehicles B. Effect on Priced Freeway Lane Cents per Vehicle Mile Consumer Surplus User Cost Curve X4 Toll = User Cost Curve X1 1,600 Vehicles Peak Hour Traffic Volume FIGURE 4 Alternative 3 - Add Priced Lane: (A) Effect on Unpriced Freeway Lanes; (B) Effect on Priced Freeway Lane.