A Model to Estimate Passenger Vehicle Fleet Composition,Vehicle Miles Traveled,and Fuel Consumption

Abstract

A significant portion of funds in the United States for road maintenance and improvement are the levy of a fuel tax per gallon of fuel sold. The government mandated improvements in fuel efficiency of vehicles and the greater proportion of hybrid or nongasoline-powered vehicles expected to be used for passenger transportation are anticipated to adversely affect such revenues. In this study, readily available public domain data on new vehicle sales and survivability data are used to develop estimates of the future fleet composition by specific vehicle categories, vehicle miles traveled by the vehicle category, and fuel consumption by the vehicle category. It is then used to develop estimates. The model takes into consideration the emerging classes of hybrid and alternative fuel vehicles, which were not adequately captured in the historical data. This methodology could be used to support policy and operations analysis related to highway financing and infrastructure management.

Keywords

highway financing Highway Trust Fund gasoline tax fuel tax trust fund highway revenue projection VMT vehicle fleet composition

Introduction

Providing high quality of road infrastructure in the United States is challenging because of the inability to keep the system capacity on pace with the increases in demand. Over the past decade, maintaining and improving the road transportation infrastructure in the United States have become increasingly challenging primarily due to the shortage of financial resources required for the same. The most significant source of revenues at the federal level in the United States for road maintenance and improvement is from the levy of a fuel tax per gallon of fuel. If the funding shortage is not addressed, this situation is expected to worsen in the years to come because of improved fuel efficiency of vehicles, greater proportion of hybrid or nongasoline-powered vehicles used for passenger transportation, and increasing costs to maintain and improve roads. Failure to maintain and improve the nation’s road infrastructure would have significant affects on the overall productivity and economy of the United States and also adversely affect the mobility needs of the populace.

Several studies (American Association of State Highway and Transportation Officials [AASHTO], 2007a; AASHTO, 2007b; AASHTO, 2007c; Congressional Budget Office [CBO], 2008a; CBO, 2011a; Boarnet, 1999; National Cooperative Highway Research Program, 2006; Transportation Research Board, 2006; Wachs, 2003, 2005, 2006; Puentes & Prince, 2003; Seggerman et al., 2010; Vasudevan, 2008) indicate that the current system of gasoline tax–based system needs to be reevaluated or restructured. Many such studies present estimates of revenue generated for the United States based on projected gasoline sales using historical data and on existing polices or in some cases other potential options. U.S. Government Accountability Office (2006) discussed how the Treasury estimates revenues to the Highway Trust Funds (HTF). This study compared HTF estimates based on CBO and the U.S. Department of Transportation (DOT) for years from 2006 to 2011. Although the trend of the estimates was very similar and was consistent for the past years, the projected estimates showed a diverging trend over the study period. However, both the models were consistent to show a negative balance for HTF by the year 2011. In a statement before the Committee on Finance, U.S. Senate, the status of HTF in the year 2011 was presented (CBO, 2011b). This report shows that the shortfall of funds, when outlays are considered, was consistently increasing over the years up to 2011. After 2011, the shortfall is projected to increase significantly. This means that important infrastructure projects will have to be delayed or even abandoned due to lack of funds.

Hijiamiri and Wachs (2010) estimated impacts of hybrid vehicle and electric vehicles on transportation finance in the United States. Studies documented in the literature often do not provide details of methodologies used to arrive at their estimates. These efforts typically do not explicitly document the process to estimate vehicle fleet composition (or mix) or vehicle miles traveled (VMT). However, these estimates directly affect fuel consumption and, in turn, revenues from the sale of such fuel. The absence of such documentation makes it difficult for state or local agencies to estimate the impacts of changes in the corporate average fuel efficiency (CAFE) standards and the proportion of hybrid and alternative fuel vehicle (AFV) in the fleet on their transportation related revenue. Planning for alternative scenarios and developing policy options require quantitative analyses of such factors. Thus, there is a need to document the methods used to develop estimates of future passenger vehicle fleet mixes, their annual VMT, and their corresponding annual fuel consumption. Once such a model is developed, agencies, at either the state, regional, or local levels, could evaluate the impacts of policy changes and various vehicle technological changes on fuel consumption and hence on projected revenues from fuel sales. This will help such stakeholders develop long term strategies to effectively address the corresponding issues.

Objective

This article documents an approach to model for future years the passenger vehicle fleet composition along with VMT and fuel consumption by specific passenger vehicle categories. The model uses data readily available in the United States. Such a model is expected to enable analysts as well as local and state agencies and other stakeholders to analyze the impacts of policies on the vehicle fleet mix, and hence on fuel consumption. The model is structured to enable users to modify its component steps based on their needs and available resources.

Literature Review

The topic of highway financing has received attention from transportation professionals and law makers for many years. A review of the literature identified many related studies conducted within the past few years. Key among these are the works of Wachs (2003, 2005, 2006, 2009), Sciara and Wachs (2007), Puentes and Tomer (2009), Waid, Turochy, and Sisiopiku (2009), Regan (2008), Delucchi (2007), and Pirog and Fischer (2008). Baker et al. (2010) discuss a model developed for projecting revenues from a state VMT fee using data from the state of Texas. In this article, the authors incorporate relationships among various parameters including population growth, VMT, average fuel efficiency, and legislative changes to develop a model. Although these studies estimate deficiencies of the existing system and develop innovative methods for highway financing, they are short on documenting the step-by-step process(es) used to estimate existing/future vehicle fleets and impacts of hybrid and AFV on overall fuel consumption.

A model to estimate the passenger vehicle fleet, the corresponding VMT and fuel consumption by vehicle category is presented and its use is illustrated using U.S. national data.

Method

The new model presented herein uses new vehicle sales data and survivability information to estimate the passenger vehicle fleet composition, distribution of VMT by vehicle category, and fuel consumption by vehicle category. The model is called new sales survivability (NSS) model, and it has six key steps:

Step 1: Establish historical vehicle fleet composition

Step 2: Determine historical VMT by vehicle category

Step 3: Quantify historical fuel consumption

Step 4: Estimate future vehicle fleet composition

Step 5: Estimate future VMT by vehicle category

Step 6: Estimate future fuel consumption

Steps 1 to 3 are for estimating vehicle fleet, VMT by vehicle category, and fuel consumption by vehicle category based on historical data, whereas Steps 4 to 6 are for estimating future vehicle fleet, VMT by category, and fuel consumption by category. As Steps 1-3 are based on historical data, they result in a calibrated set of equations wherein the model estimates are reconciled with independent data. Figure 1 shows a flow diagram for the NSS model. The individual steps are described next.

Figure 1.

Flow Diagram of the NSS Model.

Step 1: Establish Historical Vehicle Fleet Composition

To estimate VMT and fuel consumption by vehicle categories in the NSS model, the first step is to establish the vehicle fleet composition for any year in the past. Typically, in each state in the United States, the state Department of Motor Vehicles (DMV), or an agency with a similar role, maintains vehicle registration information. However, due to privacy considerations, disaggregated details of these data are not readily accessible. In such cases, an alternate process is required to estimate the vehicle fleet composition. New vehicle sales data are more readily available from government agencies. So the vehicle sales information is a potential starting point for this process. One approach to this method is included in Figure 1. In this case, the vehicle fleet composition is determined using new vehicle sales data and vehicle survivability estimates. Survivability estimates provide information regarding vehicle life expectancy (or longevity of use in service based on registration with the DMV). Several factors affect the survivability of vehicles, including crash risks of the region considered and vehicle obsolescence. Survivability data are typically developed by government agencies and insurance companies. Using these two data sets, and classifying vehicles into a finite number of categories, the vehicle fleet composition in any year can be estimated using Equations 1 and 2.

N_{i k} = \sum_{j = n_{1}}^{i} S a l e s_{j k} \cdot S_{(i - j + 1) k}

where, N_ik is the number of category k vehicles in use (registered) in year i, Sales_jk is the new vehicle sales for vehicle category k for year j (j varies from year n₁ to i), S(i-j+1)k is survivability of vehicle category k from year j in year i (i.e., probability of survival), and n_i is the earliest year of vehicle sale used in the analysis.

The vehicle fleet for year i is estimated using Equation 2.

V F_{i} = \sum_{k = 1}^{n_{2}} N_{i k}

where, n₂ is the total vehicle categories considered The accuracy of the vehicle fleet estimated using this model needs to be verified. This can be done using an independent aggregated data source—for example, those from government agencies’ vehicle registration data. The total number of vehicles obtained from the estimated fleet mix is compared with vehicle registration data and the vehicle fleet composition in individual categories needs to be reconciled and adjusted. To reconcile the differences between the estimated and the vehicle registration data (e.g., from a government agency such as Federal Highway Administration [FHWA]), the percentage of vehicle fleet in the year I is calculated using the Equation 3:

P N_{i j k} = \frac{S a l e s_{j k} . S_{(i - j + 1) k}}{\sum_{j = 1982}^{2005} S a l e s_{j k} . S_{(i - j + 1) k}}

where, PN_ijk is the percentage of adjusted vehicle in the year i of the vehicle manufactured in the year j belonging to vehicle category k.

Using the calculated values of all vehicle types and categories, the adjusted vehicle fleet is estimated by multiplying the percentage of cars or light trucks by the total number of vehicles registered in these categories respectively using the Equation 4.

V F_{i} = \sum_{j = 1981}^{i} \sum_{k = 1}^{6} P N_{i j k} . R_{i k}

where, R_ik is the total reported vehicles registered in the year i for vehicle category k from government sources (e.g., annually published information in the U.S. DOT’s Highway Statistics report).

Step 2: Determine Historical VMT by Vehicle Category

In this step, the VMT by vehicles in each of the categories are estimated for the past years included in the modeling process using the adjusted vehicle fleet estimated in Step 1. This is computed based on the adjusted number of vehicles in year i for category k and the average VMT in year i in category k from year j. The VMT for the entire vehicle fleet in year i is estimated using Equations 5 and 6.

V_{i k} = \sum_{j = n_{1}}^{i} A N_{i k} . E V M T_{(i - j + 1) k}

where, V_ik is the estimated VMT by vehicle category k in year i, AN_ik is the adjusted number of vehicles in year i for category k, and EVMT(i-j+1)k is the estimated average annual VMT in year i for vehicles in category k from year j.

The total VMT in year i for the entire vehicle fleet is estimated using Equation 4.

V_{i} = \sum_{k = 1}^{n_{2}} V_{i k}

Similar to Step 1, the accuracy of the estimated VMT needs to be verified. As with vehicle registration data, state government agencies such as the Departments of Transportation publish estimated annual VMT for various vehicle categories. Using this number, the estimated VMT by fleet category needs to be adjusted. This step is very similar to the adjustments made in Step 1. First, the percentage of vehicle fleet is calculated using Equation 7 (similar to Equation 3)

P V_{i j k} = \frac{A N_{i k} . E V M T_{(i - j + 1) k}}{\sum_{j = 1981}^{i} A N_{i k} . E V M T_{(i - j + 1) k}}

The adjusted VMT by vehicle fleet is estimated by using Equation 8.

A V_{i k} = \sum_{j = 1981}^{i} \sum_{k = 1}^{6} P V_{i j k} . R V_{i k}

where, RV_ik is the total reported VMT in the year i for vehicle category k from Transportation Statistics.

Step 3: Quantify Historical Fuel Consumption

The fuel consumption by vehicles in each of the categories in the years past is estimated based on adjusted VMT values estimated in the previous step using Equations 9 and 10.

F C_{i j k} = \frac{A V_{i j k}}{F E_{i j k}}

where, FC_ijk is the average fuel consumed by vehicle category k from year j in year i, AV_ijk is the adjusted average VMT in year i for vehicles in category k from year j (from Step 2), and FE_ijk is the average fuel efficiency in year i for vehicle category k from year j.

F C_{i} = \sum_{k = 1}^{n_{2}} \sum_{j = n_{1}}^{i} \frac{A V_{i j k}}{F E_{i j k}}

where, FC_i is the fuel consumed in year i by vehicles in all model years of all vehicle categories.

Step 4: Estimate Future Vehicle Fleet Composition

The passenger vehicle fleet composition is estimated using the estimated vehicle sales. This requires developing models to estimate future vehicle sales. Vehicle sales depend on several factors including the economy, fuel and operating costs, fuel efficiency, incentives, and vehicle availability. These factors need to the included in estimating future vehicle sales by vehicle categories. Once the vehicle sales data are estimated, the vehicle fleet composition could be calculated as shown in Step 1.

Step 5: Estimate Future VMT by Vehicle Category

Similar to Step 2, VMT by vehicle in each of the categories is calculated in Step 4. The equations developed in Step 2 are used in this step.

Step 6: Estimate Future Fuel Consumption

Fuel consumption by each of the vehicle categories in the future is estimated using the equations developed in Step 3.

Illustration Using National Data From the United States

The steps discussed in the previous section to estimate vehicle fleet, VMT by fleet, and fuel consumption, using the NSS model, are illustrated using national data for passenger vehicles from the United States. Although national data are showed here for illustration, this model could be used with data for other spatial extents such as at a state or local government jurisdictional level.

Step 1: Estimate Vehicle Fleet Based on Historical Data

As shown in Figure 1, this step is divided into several steps. They are as follows:

Step 1A: Develop Inventories of the Vehicle Fleet

Vehicle fleet data such as the year of manufacture, make, model, vehicle type, and fuel efficiency are critical data elements that need to be incorporated into the model. In a technical report by the U.S. Environmental Protection Agency (EPA), Heavenrich (2006) lists historical data of the total number of vehicles sold by year for the years 1975 to 2006. The data provided include the proportion of vehicles in five major types: cars, station wagons, vans, sport utility vehicles (SUVs), and trucks. Each of these categories is again categorized into three other categories based on size: small, medium, and large. For each of these categories, the information provided includes average fuel economy and key engine properties. This report also provides fuel efficiency based on lab tests as well as based on road data. Using these data, vehicle fleet inventories could be developed. A report by the U.S. Department of Energy (DOE; Davis & Diegel, 2007) also lists vehicle inventory for the United States for the past few years. This inventory again categorizes vehicles into two basic types: cars and light trucks. This report lists the average age of vehicles for each the two types. This information could be used along with EPA data for verification and validation purposes.

Step 1B: Estimate Vehicle Fleet for Registered Vehicles Using Vehicle Sales Information

This is the key step in identifying VMT and fuel consumption by each vehicle type and vehicle category. This step is divided into the following steps.

Step 1B-a: Categorize vehicle sales into vehicle categories

As indicated in Step 1A, the vehicle sales information obtained from the EPA report (Heavenrich, 2006) divides vehicles into five different types (sedans, station wagons, vans, SUVs, and trucks), each of which is again divided into three categories, (small, medium, and large vehicles) based on their size. This results in a total of 15 categories. Although this classification system would be good for a detailed study, the 15 categories are not necessary to meet the objective of this article, which is to develop a first-generation model. For illustrative purposes, this study uses the following types: cars and trucks, each of which is further classified as small, medium, or large. This results in a total of six categories. Cars type includes sedans and station wagons, whereas trucks include vans, SUVs, and pickup trucks. Table 1 shows new vehicle sales data based on these revised categories. This table shows a notable hike in small light truck category for the year 1986. This is mainly because of the change in small pickup sales reported in that year.

Table 1.

Passenger Vehicles Sales Data (Based on Limited Categories).

Model year	Small car	Medium car	Large car	Small truck	Medium truck	Large truck
1975	4,565,000	1,920,000	1,752,000	215,000	481,000	1,290,000
1976	5,386,000	2,449,000	1,886,000	234,000	530,000	1,847,000
1977	5,865,000	2,774,000	2,659,000	312,000	576,000	1,935,000
1978	4,991,000	3,842,000	2,341,000	358,000	744,000	2,171,000
1979	4,715,000	3,693,000	2,386,000	470,000	601,000	2,016,000
1980	5,135,000	3,244,000	1,065,000	528,000	328,000	1,006,000
1981	4,495,000	3,174,000	1,064,000	423,000	347,000	1,050,000
1982	4,421,000	2,420,000	978,000	403,000	593,000	917,000
1983	4,249,000	2,545,000	1,209,000	381,000	1,054,000	863,000
1984	6,127,000	3,135,000	1,412,000	651,000	1,554,000	1,140,000
1985	6,015,000	3,118,000	1,657,000	705,000	1,779,000	1,183,000
1986	6,551,000	3,075,000	1,387,000	1,022,000	2,108,000	1,219,000
1987	6,814,000	2,611,000	1,306,000	821,000	2,462,000	851,000
1988	6,959,000	2,398,000	1,378,000	683,000	2,606,000	1,270,000
1989	5,836,000	2,830,000	1,352,000	617,000	2,611,000	1,206,000
1990	5,159,000	2,526,000	1,123,000	508,000	2,171,000	1,124,000
1991	5,241,000	2,236,000	1,046,000	460,000	2,720,000	868,000
1992	4,583,000	2,257,000	1,267,000	421,000	2,603,000	1,040,000
1993	4,838,000	2,496,000	1,122,000	419,000	3,104,000	1,231,000
1994	4,926,000	2,195,000	1,293,000	557,000	3,600,000	1,552,000
1995	5,388,000	2,691,000	1,315,000	493,000	3,649,000	1,607,000
1996	4,287,000	2,528,000	1,075,000	343,000	3,524,000	1,386,000
1997	4,592,000	2,548,000	1,195,000	620,000	3,217,000	2,286,000
1998	3,940,000	3,120,000	912,000	576,000	3,809,000	2,100,000
1999	3,997,000	3,322,000	1,059,000	527,000	3,815,000	2,496,000
2000	4,334,000	3,128,000	1,665,000	501,000	4,151,000	2,795,000
2001	4,277,000	2,716,000	1,416,000	471,000	3,427,000	3,302,000
2002	4,037,000	3,015,000	1,252,000	551,000	3,399,000	3,863,000
2003	4,036,000	2,654,000	1,261,000	458,000	3,741,000	3,624,000
2004	3,574,000	2,680,000	1,283,000	417,000	3,773,000	3,982,000
2005	3,531,000	3,002,000	1,443,000	225,000	3,524,000	4,242,000
2006	3,677,000	2,854,000	1,733,000	112,000	4,073,000	4,224,000

Step 1B-b: Summarize registration data

Highway Statistics are reports published annually by the FHWA (2006). They provide registration information for all vehicles (both commercial and private) separately by state for each year. In this step, the registration data are summarized for the years from 1981 to 2005 from the Highway Statistics report for the corresponding year.

Step 1B-c: Identify vehicle fleet composition for the registered vehicles by year

The registered vehicle information provided in the Highway Statistics does not contain information on the fleet composition. However, such information is very important from the point of view of fuel consumption, as fuel efficiency is dependent to some extent on vehicle size and type. A study by National Highway Traffic Safety Administration (NHTSA, 2006) estimated the survivability and mileage information based on vehicle age (Table 2). The information is divided into two types: cars and light trucks. Using this information, the vehicle fleet composition of registered vehicles in any year for each category could be estimated using Equation 1.

Table 2.

Passenger Vehicle Survivability and Vehicle Miles Traveled.

	Survivability		Vehicle miles traveled
Vehicle age	Sedan	Light trucks	Sedan	Light trucks
1	0.9900	0.9741	14,231	16,085
2	0.9831	0.9603	13,961	15,782
3	0.9731	0.9420	13,669	15,442
4	0.9593	0.9190	13,357	15,069
5	0.9413	0.8913	13,028	14,667
6	0.9188	0.8590	12,683	14,239
7	0.8918	0.8226	12,325	13,790
8	0.8604	0.7827	11,956	13,323
9	0.8252	0.7401	11,578	12,844
10	0.7866	0.6956	11,193	12,356
11	0.7170	0.6501	10,804	11,863
12	0.6125	0.6040	10,413	11,369
13	0.5094	0.5517	10,022	10,879
14	0.4142	0.5009	9,633	10,396
15	0.3308	0.4522	9,249	9,924
16	0.2604	0.4062	8,871	9,468
17	0.2028	0.3633	8,502	9,032
18	0.1565	0.3236	8,144	8,619
19	0.1200	0.2873	7,799	8,234
20	0.0916	0.2542	7,469	7,881
21	0.0696	0.2244	7,157	7,565
22	0.0527	0.1975	6,866	7,288
23	0.0399	0.1735	6,596	7,055
24	0.0301	0.1522	6,350	6,871
25	0.0227	0.1332	6,131	6,739
26		0.1165		6,663
27		0.1017		6,648
28		0.0887		6,648
29		0.0773		6,648
30		0.0673		6,648
31		0.0586		6,648
32		0.0509		6,648
33		0.0443		6,648
34		0.0385		6,648
35		0.0334		6,648
36		0.0290		6,648

The next step in finalizing the fleet mix is to compare the sum of estimated vehicle fleet mix from NSS model with the vehicle registration data obtained in Step 1. Figure 2 provides a comparison of registered vehicle data and vehicle fleet data obtained in the previous step for each of the years from 1981 to 2005. It can be seen that the vehicle estimate obtained based on vehicle sales data under estimated vehicle registration data between 8% to 21% for cars and 6% to 24% for light trucks for various years.

Figure 2.

Comparison of Passenger Vehicle Registrations Data from the NSS Model and FHWA.

Step 1C: Adjust Vehicle Fleet Based on Registration Data

Figure 2 shows that although the vehicle registration data from NSS model and from FHWA show similar patterns, there is a notable difference between these two. This needs to be reconciled. As shown previously this is done Equations 3 and 4.

Step 2: Estimate VMT by Vehicle Fleet Based on Historical Data

This step is to estimate the VMT by vehicles in individual categories. FHWA (2006) provides a summary of total VMT by state for each year. In this step, the overall VMT is apportioned to vehicle fleet as follows:

Step 2A: Estimate VMT by the Vehicle Fleet. In this step the overall VMT is apportioned to vehicle fleet using Equations 5 and 6. The same process is performed for all the years from 1981 to 2005. This activity estimates VMT by vehicle fleet based on sales data and survivability data. Similar to the comparison made for registered vehicles, it is important to compare the VMT estimates from the model with similar estimated published in Highway Statistics (FHWA, 2010). Figure 3 compares VMT data from NSS model and data from Highway Statistics for each of the years from 1981 to 2005. It is seen that the total VMT estimates based on vehicle sales data and that reported by Highway Statistics follow very similar trends over the years. However, when the cars and light trucks are considered separately, there are some notable differences, especially for light trucks.

Figure 3.

Comparison of Passenger Vehicles VMT Data from the NSS Model and FHWA.

Step 2B: Adjust VMT by Vehicle Fleet Using VMT Data by FHWA

Figure 3 shows that although the VMT based on the registration data and the fleet mix data shows similar trends, they are not identical. Therefore, the next step is to reconcile the differences. As discussed earlier, Equations 7 and 8 are used to reconcile and adjust the estimates.

Step 3: Estimate Fuel Consumption Based on Historical Data

A basic step in identifying the impact of fuel efficient vehicles on fuel tax–based revenue is estimating the historical fuel consumption rate by each of the vehicle categories. The distribution of VMT by vehicle category estimated in the previous step is used to estimate fuel consumption by vehicle category. The U.S. EPA (Heavenrich, 2006) publishes fuel efficiency by vehicle category for each year. This report categorizes vehicles into over 30 categories. The first step is to consolidate these into the six categories identified in Step 2. Using the consolidated vehicle categories, the fuel efficiency (i.e., gas mileage) for vehicles in each of the categories is estimated by calculating the weighted average using adjusted data for the corresponding vehicle fleets.

Fuel consumption by vehicles in each category is estimated using Equations 5 and 6. The summary from each year is compared with the gasoline sales data published in Highway Statistics. Figure 4 provides a comparison of the gasoline sales data from estimates and those published in Highway Statistics. It is evident that there are discrepancies between the general trends for the estimated fuel consumption and reported fuel consumption. Furthermore, the discrepancy between the two increases with the year. This could be because of the fuel efficiency data used for estimation. For this estimation, it is assumed that the fuel efficiency of vehicles would remain the same, irrespective of the age of vehicle. However, both the model estimates and the reported values from Highway Statistics show very similar trends, which validate the estimation process. These estimates could be greatly improved if the fuel efficiencies of vehicle categories were to be adjusted for the age (and/or miles driven) of the vehicle.

Figure 4.

Comparison of Fuel Consumption Data from the NSS Model and FHWA.

These steps estimate the distribution of vehicle mileage and fuel consumption by various vehicle classes. These steps also address the accuracy of these estimates by comparing them with the Highway Statistics data.

As described in the previous sections, Steps 1 to 3 illustrated the methodology to estimate vehicle fleet, VMT, and fuel consumption by vehicle category. As these were based on historical data, it was easier to compare and validate the model. Figures 2, 3, and 4 showed that the estimated from the model were comparable to historical data. However, Steps 4 to 6 address processes of developing estimates for the future years. There are several uncertainties involved in each of the steps. Vehicle sales, fuel efficiency, and fuel consumptions are dependent on several factors, such as the economy, gasoline prices, and technological advances. Therefore, it is important to recognize these factors in developing such estimates. In this study, models developed in Steps 1 to 3 are slightly modified and are used in Steps 4 to 6.

Step 4: Estimate Future Vehicle Fleet Based on Models Developed

Step 4A: Develop a Model to Estimate New Vehicle Sales by Category

The CBO of the U.S. Congress (CBO, 2008b) published a report which analyzed the impact of high gasoline prices on new vehicles sales and VMTs. Using data for the three years 2005 to 2007, this report identifies that for a 20% increase in gasoline price, there is a +2.6% shift in new vehicle sales toward sedans from light trucks. This report also shows that the sale prices of used (preowned) SUVs and pickups decreased and the prices of used sedans increased. These data show a clear consumer trend over the 3-year period toward purchasing sedans, which have relatively better fuel efficiencies. This was a relationship which could not be established from the historical vehicle sales data for the years 1976 to 2005. While accounting for VMT, the report shows that a 20% increase in gasoline price results in a significant decrease (–0.40%) for weekday traffic and an insignificant increase (+0.12%) for weekend traffic. These findings are used to estimate various scenarios of vehicle fleet and fuel consumption. Equation 11 shows percentage new vehicle sales by category for a future year:

P N N_{i k} = P N N_{(i - 1) k} . {1 + (\frac{(G P_{i} - G P_{i - 1})}{G P_{i - 1}} . \frac{C_{k}}{0.20})}

where, PNN_ik is the proportion of new vehicles of category k sold in the year i, PNN_(i–1)k is the proportion of new vehicles of category k sold in the year i–1, GP_i is the gasoline price for the year i, GP_i–1 is the Price of gasoline for the year i–1, and C_k is the change in proportion of vehicle of category k for a 20% increase in gasoline price.

One of the issues in using the CBO report is that the categories identified in the CBO report are different from the one used in NSS model. Therefore, the vehicle categories reported were recategorized into the six categories used in this article. The proportion of vehicle fleet for new vehicles for any year i is estimated using Equation 12:

P N N_{i} = \sum_{k = 1}^{6} P N N_{(i - 1) k} . {1 + (\frac{(G P_{i} - G P_{i - 1})}{G P_{i - 1}} . \frac{C_{k}}{0.20})}

The next step is estimate fleet mix for a future year based on these equations. This starts with identifying the existing fleet mix in terms of the six categories considered in this study. Using the new vehicle sales data, the proportion of vehicles in each of these categories is calculated.

Once the fleet mix distribution model is determined, the next step is to estimate gasoline price for future years, which has tremendous uncertainties (but the accuracy of gasoline price estimates is not the focus of this article). The price of gasoline has varied considerably over the past few years. Although it is extremely difficult to predict gasoline prices, for illustration purposes, it is assumed that the gasoline price increases proportional to the Consumer Price Index (CPI) annually after 2009. Using the estimated gasoline prices and Equations 11 and 12, the proportions of new vehicle sales by categories are estimated. To convert these to the number of vehicles, the total number of new vehicles sold for each of the years needs to be estimated.

Historical data for the years from 1981 to 2005 show that new vehicles sold account for an average of 7.1% of total registered vehicles. For illustration purposes, it is assumed that for any future year, this proportion would remain at 7.1%. The next step is to estimate vehicle registration data for a future year. Historical data do not show a consistent trend for the change in vehicle registration. However for illustration purposes, it is assumed that the total number of registered vehicles would increase by 0.5% every year. Once the total number of registered vehicles is estimated for a year, the corresponding number of new vehicles for that year is estimated by multiplying the total number of registered vehicles with the proportion of new vehicles (i.e., 7.1%).

Once the new vehicles sales are estimated, the vehicle fleet for these new vehicles is estimated using the information on the proportions estimated using Equation 13.

V F N N_{i} = \sum_{k = 1}^{6} P N N_{i k} . P N V_{i} . R_{i}

where, VFNNi is total number of new vehicles by vehicle category in year i, PNV_i is the proportion of new vehicles sold in year i based on vehicle registration data, and R_i is the estimated total number of vehicles registered in year i.

Step 4B: Estimate Distribution of Hybrid Vehicles in the New Vehicle Fleet

The vehicle fleet mix estimated thus far accounts for six major categories of vehicles, without considering alternate fuel vehicles and hybrid vehicles. When considering vehicle fleets and VMT, it is important include these vehicle categories in the fleet composition, although they are within the six categories considered. This estimation is described in the following steps.

Step 4B-a: Estimate new hybrid vehicle sales

The U.S. DOE—Energy Efficiency and Renewable Energy website (U.S. DOE, 2008) lists the vehicle sales information of hybrid vehicles in various categories for the years 1999 to 2007. Although hybrid vehicles were available in 1999, only from the year 2000 did their sales show any notable numbers. The total numbers of hybrid vehicles sold per year are shown in Figure 5. These vehicle types are summarized into six vehicle categories. Table 3 summarizes the vehicle sales into the six vehicle categories. Data in this table show that of all hybrid vehicles sold in the United States, a predominant portion has been cars. Further within the cars type, small cars and medium cars account for more than 70% of all vehicles sold in all the years and more than 80% for most of the years. This provides a sense of the consumer’s buying trend. Although one could argue that in the previous years, there were no other vehicles available as a choice, yet even after they became available, their sales were not comparable with those of small/medium cars category. Therefore, it could be indicative that the sales of hybrid vehicles are also very similar to the sales of nonhybrid vehicles, with more cars being sold than light trucks with higher gasoline prices.

Figure 5.

Total Number of Hybrid Vehicles Sold per Year.

Table 3.

Proportion of Hybrid Vehicle Sales by Vehicle Category.

	Year
Vehicle category	1999 (%)	2000 (%)	2001 (%)	2002 (%)	2003 (%)	2004 (%)	2005 (%)	2006 (%)	2007 (%)
Small cars	100.0	100.0	100.0	100.0	100.0	95.2	64.1	55.0	60.7
Medium cars	0.0	0.0	0.0	0.0	0.0	1.3	8.0	15.3	19.5
Large cars	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.3
Small light trucks	0.0	0.0	0.0	0.0	0.0	3.6	9.4	9.2	7.1
Medium light trucks	0.0	0.0	0.0	0.0	0.0	0.0	18.4	20.4	12.4
Large light trucks	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Total	100	100	100	100	100	100	100	100	100

Using the trend of historical vehicle sales data, sales data for a future year could be estimated. The historical data show that for all years cars accounted for over 70% of all the hybrid vehicles sold. Hybrid vehicle sales grew well over 30% for most of the years. Although sustaining such growth is difficult, with the gas prices remaining at historically high levels, the growth of hybrid vehicles into the future is expected to increase. For illustrations purposes, it is assumed that the over the future years the sales of new hybrid vehicles will increase annually by 20%, with the proportion of six categories of, small cars, medium cars, large cars, small light trucks, medium light trucks, and large light trucks being 60%, 18%, 2%, 7%, 12%, and 1%, respectively. Therefore, vehicle fleet of new alternate fuel vehicles for a year i could be represented mathematically as:

V F H N N_{i} = \sum_{k = 1}^{6} H N N_{b} . {(1 + r)}^{i - b} . P N H_{k}

where, HNN_i is vehicle fleet of hybrid vehicles for a year i, HNN_b is the total number of hybrid vehicles in base year b, r is the average rate of change in annual sales of hybrid vehicles, and PNH_k is the proportion of hybrid vehicles of category k.

Step 4B-b: Estimate the hybrid vehicle fleet

Once the new vehicles sales data are estimated, the next step is to estimate vehicle fleet composition of the hybrid vehicles for all years considered. For this, models developed in Step 1 are used. As the hybrid vehicles are relatively new, their survivability data are not readily available. Therefore, it is assumed that the survivability of hybrid vehicles is similar to that of regular vehicles. Using these values, estimates are developed for the hybrid vehicle fleets for the years considered.

Step 4C: Estimate Distribution of AFV in the New Vehicle Fleet

Hybrid vehicles are considered in broader terms in the AFV category. However, they use gasoline as their primary fuel, with alternate fuel sources such as electric batteries improving its fuel mileage. In this study, AFV represents vehicles that do not use gasoline at all for their operation. Several models of AFV have emerged during the past decade. Because of the availability of relatively cheap gasoline, comparatively higher purchase costs, and other technology related concerns, they typically did not last long to make an impression as the hybrid has done within a few years of its introduction. The Energy Information Administration (EIA, 2007) publishes data on the vehicle sales data by fuel type. Table 4 shows a summary extracted from EIA’s report. It shows that over the 3 years (2004-2006), on an average about 3% of all new cars, 7% of all new trucks sold were AFV, excluding hybrid vehicles.

Table 4.

Alternative Fuel Vehicle Sales Data (in Thousands).

	Year
Vehicle type	2004	2005	2006
New alternative-fuel cars (excluding hybrids)	138	0 248.5	278.5
Percentage of new alternate-fuel cars	1.8%	3.1%	3.1%
New alternative-fuel light trucks (excluding hybrids)	680.1	565.3	511.4
Percentage of new alternate-fuel light trucks	8.1%	7.0%	7.2%
New alternative-fuel vehicles (excluding hybrids)	818.1	813.7	789.9
Percentage of new alternate-fuel vehicles	5.1%	5.0%	5.0%

Source: Energy Information Administration, 2007.

However, most of the existing AFV predominantly use petroleum based fuels (e.g., methanol and compressed natural gas).This, in turn means that as the petroleum price increases, the costs to operate such vehicles also will increase. This might be a reason why the vehicle sales of this category of vehicles has decreased. However, as the fuel prices remain at relatively high levels, the demand for newer versions of vehicles in this category are expected to increase. The introduction of Honda’s hydrogen fuel vehicles in California (Sabatini, 2007), and all electric cars developed by Tesla motors (Copeland, 2008) are only a few of the new fleet mix expected to enter the market in the coming years. Reports in the popular press indicate that many of all the major (United States based as well as foreign) automobile manufacturers are currently working on some AFV. All these point to the potential growth in this segment of vehicles would show a similar trend, if not more dramatic, as that of hybrid vehicles.

To show the impact of AFV on vehicle fleet composition and fuel consumption, it is assumed that from the year 2009 onwards, the sales of older technology-based AFV would increase at a rate of 0.5% and the newer version, such as, electric and hydrogen cell fuel, would grow at the rate same as hybrid vehicles. The number of new AFV is estimated as follows:

V F A F V N N_{i} = V F A F V N N_{O i} + V F A F V N N_{N i}

where, VFAFVNN_Oi is the vehicle fleet of new AFV with older technologies for a year i and $V F A F V N N_{O i} = \sum_{k = 1}^{6} A F V N N_{O b} . {(1 + r)}^{i - b} . P A F V_{O k}$ . VFAFVNNN_i = VFHNN_i–9is vehicle fleet of new AFV with newer technologies for a year i. AFVNN_Ob is the total new older technology AFV sold in the base year b, r is the average rate of change in annual sales of new older technology AFV PAFV_Ok is the proportion of AFV of category k, and HNN_i–9 is the total number of new hybrid vehicles sold in 9 years back.

This means that the number of new newer technology-based AFV sold will be equal to the number of hybrid vehicles sold in the year 2000.

Step 4D: Adjusting the New Vehicle Fleet to Accommodate Hybrid and AFV

The vehicle fleet estimated in Steps 4A and 4B did not include either hybrid or AFV. However, Steps 4C and 4D shows that it is extremely important to consider them in the vehicle fleet so as to better estimate VMT, and hence fuel consumption by vehicle category. The total number of new vehicles sold in each year is adjusted to accommodate these vehicle categories. Therefore, the revised vehicle fleet of regular fuel vehicles is calculated as follows:

M V F N N_{i} = \sum_{k = 1}^{6} (V F N N i_{k} - V F H N N_{i k} - V F A F V N N_{i k})

where, MVFNN_i is the revised vehicle fleet of regular fuel vehicles.

Step 4E: Develop Vehicle Fleet Based on Estimated New Vehicle Sales Data

In this step, new vehicle sales data estimated from Steps 4A to 4C are used with models developed in Step 1 to estimate estimated vehicle fleet of all vehicles (both new and existing vehicles).

Step 4F: Adjust Vehicle Fleet Based on Estimated Registration Data

The process illustrated in Step 1C is used to adjust the vehicle fleet by category. These adjusted values are used for the remaining estimations.

Figure 6 shows the adjusted summary of vehicle fleet estimates for years 2006 to 2025. This shows clearly that hybrid vehicles and AFV are expected to account for significant portions of the vehicle fleet in the future.

Figure 6.

Estimated Vehicle Fleet Based on the NSS Model.

Step 5: Estimate Future VMT by Vehicle Fleet

Step 5A: Estimate VMT by Vehicle Fleet

VMT for the future vehicle fleet is estimated by adopting the process illustrated in Step 2 using the adjusted vehicle fleet data. As VMT data for hybrid vehicles and AFV are not available, the data discussed in the preceding steps for regular fuel fueled vehicles are used as proxy.

Step 5B: Adjust VMT by Vehicle Fleet Data

An examination of the average VMT per registered passenger vehicle for the years 1981 to 2005 shows an increase at an average rate of 1.04% annually. Using this value for the estimated number of vehicles registered yearly, the total VMT of all vehicles is estimated. Using the model illustrated in Step 2B, the estimated VMT by vehicle categories are adjusted for each of vehicle categories.

Figure 7 shows the adjusted summary of estimates of VMT by vehicle classes for years 2006 to 2025. Similar to Figure 6, this also shows that hybrid vehicles and AFV are expected to account for significant portions of the VMT in future years.

Figure 7.

Estimated VMT by Vehicle Classes based on the NSS Model.

Step 6: Estimate Future Fuel Consumption

The fuel consumption is estimated for each of the vehicle classes, viz. regular fuel and hybrid vehicles separately. As AFV do not consume gasoline-based fuel, this vehicle class is not considered herein.

Step 6A: Estimate Fuel Consumption of Regular Fuel Vehicles

In this study, to estimate fuel consumed, it is assumed that the average fuel efficiency of the regular gasoline fuel vehicles increase by 3.0% annually for each vehicle category, without considering hybrid or alternate fuel vehicles. This value is selected based on the revised CAFE (NHTSA, 2009) standard of average fuel efficiency improvement of 4.5% each year until 2015. Although the 3% used in this study is lower than the CAFE recommended rate, as all the major car manufacturers in the United States are developing hybrid and AFVs, a 3% improvement in regular fuel vehicles would be sufficient, and realistic, to comply with the revised CAFE standards, once hybrid and AFVs are added to the fleet by each of the automobile manufacturers. Fuel consumption for each of the vehicle categories is estimated using models developed in Step 3.

Step 6B: Estimate Fuel Consumption of Hybrid Vehicles

With improvements in technology, hybrid vehicles are also expected to improve their fuel efficiency. In this study, for illustration purposes, it is assumed that their fuel efficiency improve by 2.0% annually. Fuel consumption by each of the hybrid vehicle categories is estimated using models developed in Step 3.

Figure 8 shows the summary of estimates of fuel consumption by vehicle classes for the years 2006 to 2025. This shows that due to the increases in vehicle presence and VMT distribution of hybrid vehicles and AFV, fuel consumption decreases significantly in the future years.

Figure 8.

Estimated Fuel Consumption by Vehicle Classes based on the NSS Model.

Conclusions and Recommendations

The objective of this article was to develop and illustrate a process to estimate passenger vehicle fleet composition, VMT by vehicle categories, and fuel consumption. A simple model, termed the new sales survivability (NSS), is illustrated in this article to estimate these items based on readily available data. Estimates generated by the NSS model for vehicle fleet composition, VMT distribution by the vehicle fleets, and total fuel consumption follow reported trends when compared with the historical data. However, the estimates of fuel consumption show a discrepancy. This could be because of not factoring in the changes vehicle fuel efficiency based on the age and mileage of vehicles. The estimates of future passenger vehicle fleets show that hybrid vehicles and AFV would represent significant portions of the fleet by the year 2025. Similarly, the estimated VMT by these vehicle categories is expected to grow significantly over the years, decreasing the fuel consumption substantially. The impacts of corresponding reductions in highway revenues would be tremendous. Based on CBO projections (CBO, 2011), the HTF would grow from 36.9 billion in 2011 to 40.9 billion in 2021 because of the projected increase in gasoline and diesel fuel consumption. The estimates of HTF based on the fuel consumption from this study show that these studies could be underestimating impacts of alternative vehicles. The estimates from the NSS model shows clearly that the issues associated with the sustainability of HTF over time could be much more severe than what has been reported by various federal agencies. These differences in estimates could be attributed to the assumptions built in the model on the AFV sales. However, they also indicate potential uncertainties in projections and estimates.

Although in the illustration presented in this manuscript used U.S. national data, this model could be easily adapted by state or local agencies to develop estimates for their own jurisdictions. Access to some specific data would reduce the computations required significantly. For example, for most of the state and local agencies, vehicle registration data could be readily accessed from Departments of Motor Vehicles. By having these data, Step 1 could be eliminated. In this study, although the overall model is universal, all the submodels were developed for a general scenario for illustration purposes. Also, the sub models to estimate future vehicle sales were developed based on several assumptions. Although these assumptions would increase uncertainties in the model results as seen in the illustration presented in this article, the quality of estimates from this model could be improved by using input data based on current/appropriate information and an improved knowledge of changes in key assumptions. The assumptions and key input data used in the NSS model merit further examination. These processes is data intensive and time consuming. However, these processes can easily be replicated using spreadsheet-based applications. Obtaining appropriate data is the key activity in the modeling process. Once all the required data are obtained, the subsequent steps can be completed with about 80 to 100 person hr of effort. As it is recommended that inputs be updated periodically to fine-tune the results, it would be efficient to develop an automated tool, which enables faster data updates and model validation. Although additional time would be consumed initially for tool development, such a tool would help expedite the process and save labor costs in the future.

The results from the NSS model could be used to support decision making related to highway financing, transportation infrastructure, operations, and safety project planning efforts. Such activities could benefit from access to better estimates of the vehicle fleet and VMT distribution by the vehicle categories, improved. The model is first-generation tool, and it could be enhanced, particularly with regard to some of the key inputs and assumptions, and further disaggregating the vehicle categories. Likewise, it could serve as a starting point for modeling diesel/commercial vehicle fleets.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Bios

Vinod Vasudevan is a faculty of Civil Engineering at the Indian Institute of Technology Kanpur (IITK). He has over 12 years of experience in the areas of traffic safety, transportation policy, operations, and risk assessment. Prior to joining IITK in 2010, he worked at the University of Nevada, Las Vegas for over 6 years as a research engineer.

Shashi Nambisan is the Director, Institute for Transportation and a Professor of Civil Engineering at Iowa State University in Ames, Iowa. Over the past 21 years, he has led efforts on more than 150 research and education projects that have addressed local, statewide, regional and national issues in transportation and infrastructure systems management related to policy, planning, operations, safety, and risk analysis. Among the awards and honors he has received is a proclamation by the Governor of Nevada designating January 31, 2007 as the “Professor Shashi Nambisan Day” in recognition of his leadership role in and contributions to enhancing transportation safety.

References

American Association of State Highway and Transportation Officials. (2007a, September). A new vision for the 21st century (Publication code: TIF4-12). Washington, DC: Author.

American Association of State Highway and Transportation Officials. (2007b, March). Surface transportation policy recommendations (Publication code: RP-TIF5-1). Washington, DC: Author.

American Association of State Highway and Transportation Officials. (2007c, September). Revenue resources to fund transportation needs (Publication code: TIF4-12). Washington, DC: Author.

Baker

Goodin

Meyland

(2010, January 10-14). Model for projecting revenues from a state vehicle-miles-traveled fee. Proceedings of the 89th Annual Meeting, Transportation Research Board, Washington, DC.

Boarnet

M. G.

(1999). Road infrastructure, economic productivity, and the need for highway finance reform. Public Works Management & Policy. Advanced online publication. doi:10.1177/1087s724X9934002

Congressional Budget Office. (2008a, May). Issues and options in infrastructure investment (Pub. No. 3145). Washington, DC: Congress of the United States.

Congressional Budget Office. (2008b, January). Effects of gasoline prices on driving behavior and vehicle markets (Pub. No. 2883). Washington, DC: Congress of the United States.

Congressional Budget Office. (2011a, January). The budget and economic outlook: Fiscal years 2011 to 2021 (Pub. No. 4235). Washington, DC: Congress of the United States.

Congressional Budget Office. (2011b, May). The highway trust fund and paying for highways (Pub. No. 4375). Washington, DC: Congress of the United States.

10.

Copeland

M. V.

(2008, July 10). Tesla’s wild ride. Fortune Magazine. Retrieved from http://money.cnn.com/2008/07/10/technology/copeland_tesla.fortune/index.htm

11.

Davis

S. C.

Diegel

S. W.

(2007). Transportation energy data book (26th ed., Report No. ORNL-6978). Washington, DC: Office of Planning, Budget Formulation and Analysis, Energy Efficiency and Renewable Energy, US Department of Energy.

12.

Delucchi

M. A.

(2007). Do motor vehicle users in the US pay their way? Transportation Research Part A. Policy and Practice, 41, 982-1003.

13.

Energy Information Administration. (2007). Light-duty vehicle sales by technology type. Retrieved from www.eia.doe.gov/oiaf/aeo/supplement/pdf/suptab_45.pdf

14.

Federal Highway Administration. (2006). Highway Statistics 2005. Washington, D.C.: Office of Highway Policy Information. Retrieved April 15, 2007, from http://www.fhwa.dot.gov/-policy/ohpi/hss/hsspubs.htm

15.

Federal Highway Administration. (2010). Highway statistics 2009. Washington, DC: Office of Highway Policy Information. Retrieved from http://www.fhwa.dot.gov/policyinformation/statistics/2009/index.cfm

16.

Heavenrich

R. M.

(2006, July). Light-duty automotive technology and fuel economy trends: 1975 through 2006 (Report EPA420-R-06-011). Washington, DC: US Environmental Protection Agency, Office of Transportation and Air Quality.

17.

Hijiamiri

Wachs

(2010). Hybrid electric vehicles and Implications for transportation finance. Public Works Management & Policy, 15(2), 121-135.

18.

National Cooperative Highway Research Program. (2006, December). Future financing options to meet highway and transit needs (Final report for NCHRP Project 20-24[49]). Washington, DC: Transportation Research Board.

19.

National Highway Traffic Safety Administration, National Center for Statistics and Analysis. (2006). Vehicle survivability and travel mileage schedules (DOT HS 809 952). Retrieved from http://www-nrd.nhtsa.dot.gov/Pubs/809952.pdf

20.

National Highway Traffic Safety Administration. (2009, August 20). Laws and regulations, Corporate Average Fuel Efficiency (CAFE). Washington, DC: U.S. Department of Transportation. Retrieved from http://www.nhtsa.dot.gov/portal/site/nhtsa/menuitem.43ac-99aefa80569eea57529cdba046a0/

21.

Pirog

Fischer

J. W.

(2008). Transportation fuel taxes: Impacts of a repeal or moratorium. Washington, DC: Congressional Research Service.

22.

Puentes

Prince

(2003). Fueling transportation finance: A primer on the gas tax. The Brookings Institution, Washington DC, USA.

23.

Puentes

Tomer

(2009). Untangling transportation funding. Washington, DC: Brookings Institution.

24.

Regan

E. J.

(2008). National transportation policy commission sees major role for tolling and pricing. Washington, DC: International Bridge, Tunnel and Turnpike Association.

25.

Sabatini

(2007, November 30). Honda sees a hydrogen future. The Wall Street Journal. Retrieved from http://online.wsj.com/article/SB119636009026608248.html

26.

Sciara

G. C.

Wachs

(2007). Metropolitan transportation funding: Prospects, progress, and practical considerations. Public Works Management & Policy, 12, 378-394.

27.

Seggerman

Williams

Lin

Fabregas

Nelson

Nicholas

(2010, January 10-14). Impact fee methodology emphasizing vehicle miles traveled. Proceedings of the 89th Annual Meeting, Transportation Research Board, Washington, DC.

28.

Transportation Research Board. (2006). The fuel tax and alternatives for transportation funding (Special Report 285). Washington, DC: National Research Council, National Academy Press.

29.

U.S. Department of Energy. (2008, July 23). Alternative fuels & advanced vehicles data center, energy efficiency and renewable energy. Retrieved from http://www.afdc.energy.gov/afdc/data/vehicles.html#afv_hev

30.

U.S. Government Accountability Office. (2006, April). Highway trust fund: Overview of highway trust fund estimates (GAO-06-572T). Washington, DC: U.S. House of Representatives.

31.

Vasudevan

(2008, December). Policies for highway financing: Gasoline taxes and other alternatives (Dissertation report). Las Vegas, NV: University of Nevada.

32.

Wachs

M. A.

(2003). A dozen reasons for raising gasoline taxes. Public Works Management & Policy, 7, 235-242.

33.

Wachs

M. A.

(2005). Improving efficiency and equity in transportation finance. Washington, DC: Brookings Institution.

34.

Wachs

M. A.

(2006). Quiet crisis in transportation finance: Options for Texas. Santa Monica, CA: RAND.

35.

Wachs

M. A.

(2009). Financing transportation infrastructure in California. Santa Monica, CA: RAND.

36.

Waid

J. C.

Turochy

R. E.

Sisiopiku

V. P.

(2009, January 11-15). Innovative financing sources for state highways: Alabama revenue from a vehicle mileage road user fee. Proceedings of the 88th Annual Meeting, Transportation Research Board, Washington, DC.