«THE FULL COST OF INTERCITY HIGHWAY TRANSPORTATION DAVID M. LEVINSON* and DAVID GILLEN Institute of Transportation Studies, University of California ...»
Transpn Res.-D, Vol. 3, No. 4, pp. 207±223, 1998
# 1998 Elsevier Science Ltd. All rights reserved
Pergamon Printed in Great Britain
THE FULL COST OF INTERCITY HIGHWAY TRANSPORTATION
DAVID M. LEVINSON* and DAVID GILLEN
Institute of Transportation Studies, University of California at Berkeley, McLaughlin Hall, Rm. 109, Berkeley, CA 94720, U.S.A.
(Received 25 August 1997; in revised form 30 November 1997) AbstractÐIn this paper we review the theoretical and empirical literature on the cost structure of the provi- sion of intercity highway transportation and specify and estimate our own cost functions. We develop a full cost model which identi®es the key cost components and then estimate costs component by component: user costs, infrastructure costs, time and congestion costs, noise costs, accident costs, and pollution costs. The total long run average cost is $0.34 per vehicle km traveled. The single largest cost category is free¯ow travel time. While the marginal cost of infrastructure is higher than its average cost, indicating that new construc- tion is increasingly expensive, the marginal cost of driving (user ®xed and variable costs) is less than the average cost, indicating that by increasing travel the user can spread his ®xed cost of a vehicle over more trips without penalty. # 1998 Elsevier Science Ltd. All rights reserved Keywords: Full cost, externalities, social cost, highway expenditures, transportation economics
1. INTRODUCTION There has been a great deal of recent interest in identifying and measuring the full costs of trans- portation, particularly highways (see for instance: Keeler et al., 1975; Fuller et al., 1983; Quinet, 1990; Mackenzie et al., 1992; INRETS, 1993; Miller and Moet, 1993; IBI Group, 1995; IWW/ INFRAS, 1995; Delucchi, 1996; Levinson et al., 1996). This debate questions whether various modes of transportation are implicitly subsidized and to what extent this biases investment and usage decisions. While environmental impacts are used to stop new infrastructure, the full costs to society of transportation are not generally calculated for ®nancing projects or charging for their use.
In this paper we review the theoretical and empirical literature on the cost structure of the pro- vision of intercity highway transportation and specify and estimate our own cost functions. In de®ning this framework we distinguish between internal (private) and external (social) costs, long and short run costs, and average and marginal costs. We also explore the various scale and scope economies that arise in the provision of transportation services.
In general, highway segments produce two outputs: trac ¯ow which requires capacity in terms of the number of lanes, and standard axle loadings which require durability in terms of the thickness of the pavement. As early as 1962, Mohring and Harwitz demonstrated that the ®nancial viability of an infrastructure facility, under optimal pricing and investment, will depend largely upon the characteristics of its cost function. To quote Winston (1991): ``If capacity and durability costs are jointly characterized by constant returns to scale, then the facility's revenue from marginal cost pricing will fully cover its capital and operating costs. If costs are characterized by increasing returns to scale, then marginal cost pricing will not cover costs; conversely, if costs are characterized by decreasing returns to scale, marginal cost pricing will provide excess revenue.'' The cost characteristics for infrastructure providers include scale, scope and economies. Scale economies refer to the size of a facility; for example, is it cheaper per lane to build three lanes than it is to provide two? If so, there are economies of scale in the provision of highways.
*Author for correspondence. E-mail: email@example.com
Small et al. (1989) refer to scope economies in highways when both capacity (number of lanes) and durability (the ability to carry heavier vehicles) are supplied.
This paper will proceed by ®rst discussing some relevant aspects of economic theory, including external and internal costs and economies of scale and scope. Next is a development of the full cost model which identi®es the key cost components. This is followed by estimation of costs component by component: user costs, infrastructure costs, time and congestion costs, noise costs, accident costs, and pollution costs. The paper concludes by summarizing the full costs of highway infrastructure and drawing some general points about the magnitude of each component. For space reasons, the comparison of the social cost results from this work with other studies are not contained in this paper, the interested reader is referred to a companion paper: The Social Costs of Intercity Passenger Transportation: A Review and Comparison of Air and Highway, (Levinson et al., 1998), which treats the subject in more detail.
2. ECONOMIC THEORY
2.1. External and internal costs Economics has a long tradition of distinguishing those costs which are fully internalized by economic agents (internal or private costs) and those which are not (external or social costs).
Agents (individuals, households, ®rms, and governments) in interrelated markets interact by buying and selling goods and services, as inputs to and outputs from production. The costs and bene®ts voluntarily interacting agents convey or impose on one another are fully re¯ected in the prices which are charged. However, when the actions of one economic agent alter the environment of another economic agent, there is an externality1. More formally, ``an externality refers to a commodity bundle that is supplied by an economic agent to another economic agent in the absence of any related economic transaction between the agents'' (Spulber, 1989). The essential distinction which is made is harm committed between strangers which is an external cost and harm committed between parties in an economic transaction which is an internal cost.
When estimating external costs, we are using the estimated amount of economic damages produced by the externality, rather than the cost of preventing that damage in the ®rst place. Rational economic actors would choose the lower of prevention costs or damage costs when costs are internalized. This should bias the results upward (if there were a cheaper prevention measure, it could be used, but if prevention were more expensive, then the actors would accept damages, the cost value we use here).
2.2. Economies of scale and scope The long run average cost curve, formed by the envelope of the short run average cost curves, often decreases over a broad range of output as size of the producer expands in both output and capacity, giving rise to economies of scale2. The presence of economies at the relevant range of producer size means that the larger the size of the producer, the lower the average or per-unit cost of output. If there were signi®cant scale economies, it would imply that fewer and larger producers (highway authorities) would be more ecient.
Typically, a highway is used to produce a large number of conceptually distinct products, differentiated by time, space, and quality generating joint and common costs. The presence of joint and common costs gives rise to economies of scope, the cost characteristic that a single ®rm multiproduct technology is less costly than a single product multi-®rm technology. Whether scope economies exist and the extent to which they exist depend upon both the number of products and the level of each output.
1 An action by which one consumer's purchase changes the prices paid by another is dubbed a `pecuniary externality' and is not analyzed here further; rather it is the non-pecuniary externalities with which we are concerned.
2 Another note of terminology should be mentioned. Economies of scale is a cost concept, returns to scale is a related idea but refers to production, and the quantity of inputs needed. If we double all inputs, and more than double outputs, we have increasing returns to scale. If we have less than twice the amount of outputs, we have decreasing returns to scale. If we get exactly twice the output, then there are constant returns to scale. In this study, since we are referring to costs, we use economies of scale. The presence of economies of scale does not imply the presence of returns to scale.
The full cost of intercity highway transportation 209
3. FULL COST MODEL
The method we use to estimate the full cost (FC) of highway travel combines elements from a number of sources, including User Costs (CU), Infrastructure Costs (CI), Environmental Costs (CE), Noise Costs (CN), Accident and Safety Costs (CA), and Time Costs (CT). Each of these costs is a function of various parameters, which may include usage of the system. Thus, in many ways, full cost depends upon demand, so we examine both the function and the range of point estimates based upon assumptions of demand and other factors.
First, we measure total costs borne by users of the system (CUT). These include the cost of vehicle ownership (as measured by depreciation) and the cost of operating and maintaining the vehicle (including gas, tires, repairs and such). Costs borne by users also include the costs of taxes and insurance. Although the cost of taxes and insurance are borne by users, they are transfers to other cost categories (infrastructure, accident and safety). The transferred costs are subtracted from user costs, they are labeled user transfers (TU).
The next category is infrastructure costs. Here we look at state level expenditures, including federal transfer payments as well as the expenditures of lower levels of government. Highway travel, like other modes, is wrought with common and joint costs between dierent trip classes and vehicle types. Using econometric analysis, we estimate the short and long run average as well as the marginal cost (government expenditure) per vehicle kilometer traveled accounting for dierent vehicle types.
Finally we add social costs which include damage to the environment (CE), which is the monetized consideration for pollution and property damage in addition to the estimated costs of global climate change; the decline in property value due to noise (CN); and the full cost of accidents (CA), regardless of incidence. While noise and environmental damage costs are pure externalities, in that their incidence falls on those outside the system, accident and congestion costs are in¯icted by one system user on another. Time costs (CT) are divided into two components, one re¯ecting free¯ow travel time, the other re¯ecting the increase in time due to congestion (other users). The full cost is
then computed with the following formula:
4.1. A model of car price The cost of operating a vehicle depends upon numerous factors, many of which are decided by the user. An important such factor is the size of the vehicle. In 1995, the most popular cars were intermediates, and that is the type assumed in this analysis of cost. The operating costs considered in the analysis include gas, oil, maintenance and tires. Insurance costs (®re/theft, collision, and property damage/liability) and license, registration, taxes, and depreciation are typically considered transfers (at least in part) and must not be double counted, and so are not considered here, but rather in later sections. For instance, the full cost of accidents can neither be considered a solely social cost nor solely a private cost. Insurance simply transfers part of the ®nancial incidence of accidents from drivers to an insurance pool. Similarly, license, registration, and taxes pay for part of constructing, maintaining, and operating the highway system. We can express this intricate
cost accounting system as a series of equations:
CUT Y=User Operating Cost ($/yr) as a function of output (Y);
TU Y=User Transfer Costs ($/yr);
CUN Y=Net User Costs ($/yr);
Cg =Cost of Gas ($/km);
Co =Cost of Oil ($/km);
Ct =Cost of Tires ($/km);
Cf =Cost of ®re and theft (insured) ($/yr);
Cp =Cost of property damage and liability (insured) ($/yr);
Cc =Cost of collision (insured) ($/yr);
Cl =Cost of licenses, fees, and taxes ($/yr);
Cd AY Y=Cost of depreciation ($/yr) as function of years and output;
Y=Output in distance traveled per year (km);
A=Age (years over which car is depreciates), for purposes of our analysis A 1 when determining annual depreciation;
1, 2=coecients from price model discussed in the following section.
Since we are dealing with a single output product, vehicle trips, we can apply basic economics to
®nd the average and marginal costs per unit distance (Y) (km):
where ACUN =Average Unit Cost; MCUN =Marginal Cost.
The hypothesis of the user cost model is that the cost of depreciation increases with age and distance traveled.
4.2. Results It is known that depreciation occurs for two main reasons: wear and tear on the vehicle and changing demand. Demand for an aging (unused) vehicle is replaced by the demand for a newer vehicle which comes equipped with more technologically advanced features. Demand is also aected by changing preferences. In order to estimate the various cost control components of depreciation, and thus to distinguish between average (stand-alone) cost or the marginal (incremental) cost, we developed a database of used car asking prices from an internet site for used car trading selecting Honda Accords and Ford Tauruses. A model with the following form was estimated using ordinary least squares regression, the results are shown in Table 1.
where P= asking price (current $); A=Age of automobile=1996, Model Year; Y=Distance Traveled per Year (miles) for that particular car; M=Make 1 if the car was a Ford Taurus, 0 if it was a Honda Accord; x=model coecients.
The implication of this is that the car loses $0.023/vmt in value and loses $1351 in value per year. This also implies that Tauruses sell for $2740 less than Hondas, all other things being equal.