Keywords

JEL Classifications

The study of transportation in urban areas relates to urban economics and to public economics and finance. The development of cities and their land use patterns cannot be understood without studying the transportation systems that shape them, nor can urban transportation systems be understood independently of the urban economy.

Unique aspects of urban transportation economics relate to demand, capacity and supply, the performance of urban transportation systems, and pricing and finance. We provide a discussion of the key conceptual issues and knowledge in each of these areas of the field and point out some challenges that remain. (For reviews of transportation economics focused less on its relationship to urban economics and more on technical issues internal to transportation, see Arnott and Kraus 2003; Small and Verhoef 2006.)

Demand

The demand for transport is ‘derived demand’. Travel provides utility mostly because it is a means to an end, be it a consumer purchase, getting to work or to recreation. The travel itself usually has a disutility which varies according to the quality, reliability and safety of the transport system or the particular trip. Hence, virtually all transport choices involve a trade-off between the inconvenience and cost of a trip on the one hand and the frequency with which that trip is chosen relative to other trips on the other.

Beginning with the emergence of the telephone, the demands for travel and for communication have become increasingly interlinked in an urban setting. While travel and communication are substitutes because a phone call, fax or e-mail (or a messenger or letter in the pre-telephone days) may reduce the need for a trip, they are also complements because cheaper communication generates higher demand for goods, services and personal contacts. From this higher demand more travel is subsequently derived.

An important aspect of urban travel is the fact that the out-of-pocket cost of travel can be low relative to the value of time expended in that travel. As such, travel competes with leisure and with work as a key activity to which time must be allocated. The dominance of time–cost means that market prices are less important than full opportunity costs in the explanation and measurement of travel behaviour. Values of time vary greatly among consumers since wage rates vary but also because of other factors. Thus, consumers who undertake similar trips frequently incur vastly different opportunity costs.

The demand for urban travel by consumers is derived from a complex set of hierarchically linked choices. At the top of the hierarchy and slowest to change are decisions relating to where to work and where to reside. Lower in the hierarchy and more malleable are choices about the number and type of cars to own including the possibility of dispensing with cars and relying on walking or public transport for some trips. Yet lower on the hierarchy are choices about the destinations and frequency of discretionary trips, the frequency of commuting (to the extent that work arrangements do not require daily commuting), the destination of the commute being implicit in the residence–workplace choice, and the mode of transport (private automobile, taxi, public transit or walking) that will be used on each such round trip. There is the choice of the time of day during which particular trips are made and the trip chaining decision about whether several trips may be combined into a tour. In a tour, the trips are linked sequentially, beginning and ending at the point of origin (for example, the consumer’s home). Finally, also important are the choices of particular routes (of the highway network, for example) on which trips occur. Firms make a similar set of choices. At the top of the hierarchy is the location of the plant vis-à-vis suppliers and markets, followed by the choice of a vehicle fleet and associated decisions of the modes (barge, rail, truck and so on) to use to get products to market or procure them from suppliers.

The study of travel demand using econometric techniques has not yet advanced to the point where a unified theory of travel can be tested that deals simultaneously with all or even most of the levels in the hierarchy. Led by McFadden (1973), transport economists have mostly focused on mode choice: the split of the demand for travel between competing modes such as auto, urban rail or bus and car pooling on a particular trip or round trip. This has resulted in the widespread use of a rich variety of discrete choice models (such as logit, nested logit or probit) that are designed to predict the probability that a randomly selected traveller will choose a particular mode to commute to work. Modelling the choice of residence or job locations has not received much attention from transportation economists. Virtually all studies in this area can be found in the urban economics literature, and some have emphasized the joint choice of residence location and mode of commuting (Anas and Duann 1985). Most travel demand studies focus exclusively on commuting, ignoring the fact that discretionary non-work travel is continually increasing with incomes, car ownership and suburbanization. And the trade off between commuting and discretionary trip-making under a time budget constraint has remained relatively unexamined.

Another area of demand that has received attention is the choice of travel route on a congested highway network. The work of Beckmann et al. (1956) counterfactually conceived traffic on a network as a stationary state process of steady flow, rather than as a system of queues and bottlenecks causing complex flow dynamics. Despite this, the simplicity of the model led to the prolific development of static assignment models by operations researchers. These models use system optimization principles to simulate how travellers choose the least costly route on a congested network. Stochastic cost perceptions have been introduced into this type of stationary state models (Daganzo and Sheffi 1977). More recently, a variety of dynamic simulation models that recognize the queue and bottleneck nature of traffic are under development based on the principle that travellers choose not only a route but also departure–arrival times (Arnott et al. 1990).

Capacity and Supply

One aspect of supply is that most transport is made possible in large part by consumer effort, time, and by inputs purchased by the consumer such as car, gasoline, vehicle maintenance and garage. Viewed this way supply becomes virtually inseparable from demand. As such it would make sense to model a part of the supply decision within a household production context.

Another aspect of supply is that the public sector is involved in the planning, provision and operation of most travel infrastructure. This includes highways as well as buses and urban rail. A key decision variable is capacity, measured as the throughput of passengers per hour that can be transported in a particular direction at a given time during the day on a particular facility. This throughput determines the user’s travel time. Also important, however, are the safety, privacy, reliability and quality of the travel time and its components such as in-vehicle time, waiting at a station, searching for parking and time walking to and from stations and parking lots.

The key supply decision is the quantity of streets and highways and public transit rights-of-way. Road capacity relative to demand determines, in part, the level of traffic congestion in an area. Since land is the prime input in roads, more road building reduces the land available for other uses such as housing or production, raising the market price of land in such uses. In turn, the price of land chiefly determines how much land is allocated to create road capacity in an area. Thus, the most congested areas are also the ones where land is the most expensive. With extremely expensive land as in Tokyo, London, Paris or downtown New York, the substitution of capital for land results in tunnelling for transit systems (subways) and even for some roads.

An important supply question is whether economies of scale exist in congested highway traffic flow. Congestion occurs when the vehicles sharing the same road segment at the same time reach a critical value relative to road capacity. The addition of one more vehicle then begins to delay the other vehicles. The total cost experienced by the vehicle stream increases by a marginal cost that is higher than the cost privately born by the marginal vehicle’s passengers. The difference between this social marginal cost and the private average cost is the monetary value of the sum of delays the marginal vehicle imposes on all the vehicles travelling with it on the road segment. The evidence seems to suggest that this congestion process exhibits no economies of scale at least at a crude level. Scaling up (or down) road capacity and the volume of traffic in the same proportion, would not increase the average cost of travel. Capital costs of highways, on the other hand, were found to exhibit significant scale economies by the engineering estimates of Meyer et al. (1965), but since then Keeler and Small (1977) and others have found statistical evidence of weak or virtually no scale economies.

In contrast to highways, rail-based public transit systems are subject to scale economies and, more importantly, to economies of density. As more passengers use these systems (for example by reducing headways between successive trains), the per-passenger total average cost comes down because of the high fixed costs involved in system construction and maintenance. This is the chief reason why such rail infrastructure is uneconomical in US cities below some critical size such as one million or more people, or in suburban areas of low land use densities where the passengers’ time-costs of accessing transit stations can be high (Kim 1979).

System Performance

The reconciliation of supply and demand results in system performance. Unlike other markets in which price is the only salient outcome of market performance, in transport the outcomes include travel time, the level of congestion or travel delay, air pollution from car traffic, accidents, system reliability (that is, the variability of travel time from day to day or hour to hour), and pecuniary and non-pecuniary externalities caused by the transport system. While travel time, congestion and reliability costs are primarily born by the travellers, air pollution and accidents have costs that are born by travellers as well as by non-travellers. Thus, the economic performance of a transport system cannot be measured completely without evaluating the social costs and benefits created by these external effects.

The purely pecuniary externalities of transport are pervasive. For example, as noted, the creation of transport capacity has a direct effect on the supply and price of land available for other uses and can thus cause land scarcity. But this is only the direct effect of capacity provision. The indirect effect on land values and land use is quite different and works at both the extensive and intensive margins. At the extensive margin, cities endowed with more road and transit capacity can expand to land areas that were previously inaccessible. At the intensive margin, transport systems work by changing the relative accessibility of land parcels. Areas that are made relatively more accessible than before gain value, while areas made relatively less accessible lose value. As a result of these shifts, windfall gains and losses in land markets should be among the chief measures of transport system performance evaluation. The aggregate change in land values can be positive or negative. Since most land is owned by consumers (such as homeowners) transport system changes play an important role in redistributing private wealth and public revenues from ad valorem property taxes by changing an existing pattern of accessibility.

Transportation, land use and land prices are the central foci in urban economics and a variety of models have been developed. Virtually all of these assume that all jobs are located at a predetermined centre, an anachronism given that current downtowns in US cities contain no more than ten per cent of the jobs. Versions of this basic model based on linear programming have been developed to model road capacity provision and transit investment in congested cities (Mills 1972; Kim 1979).

Other pecuniary externalities centre on the improved discretionary mobility enabled by transport systems. Such mobility improvements have received praise as well as criticism. Improved mobility enables easier, cheaper and more frequent contacts among firms and among firms and consumers. This should result in positive social benefits enhancing productivity and boosting economic growth. It has been noted in the large literature on spatial mismatch that central-city minorities in the United States who are less-mobility enabled, are at a disadvantage competing for suburban employment. While discrimination and suburban land use exclusion cause minorities to be clustered and socially cloistered in central cities, lower car ownership may also hinder their ability to compete for distant suburban jobs.

Improved mobility induces economic agents to locate in a more spread out pattern, substituting cheaper outlying land for more expensive, centrally located land. The resulting urban land use pattern, common in the United States, has been dubbed ‘urban sprawl’. Sprawl has been blamed for a variety of ills stemming from the increased dependence on cars and reduced pedestrian mobility that sprawled land use promotes. Among such perceived ills, for example, is the alleged demise of social and neighbourhood cohesion and the rising obesity of American children and adults.

Pricing and Finance

In practice, urban roads and transit systems are subsidized. In the United States a large part of the cost of highways and roads comes from general income taxes. The rest of the cost comes from taxes on gasoline and taxes on real property. Urban rail systems are also heavily subsidized with fares covering only about half of the operating and maintenance costs. Hence, for all forms of urban transport with the possible exception of unregulated taxis and jitneys, market-based user fees and marginal cost prices do not play the role they do in other markets.

What does economic theory tell us about how urban transport systems should be priced and financed? The answer will be different for highway and rail systems, primarily because the latter are subject to economies of scale.

The congestion externality is key in highway pricing and investment (Vickrey 1969). Economic efficiency requires that each traveller pay his full marginal social cost on each road segment that he uses. As we saw earlier, the full marginal social cost includes the monetary value of the delay each traveller imposes on his cotravellers. This is higher where congestion is high, falling to zero where congestion is not present. It has been shown that if congestion tolls can be properly calculated and levied on travellers, then with no economies of scale in roads, the tolls collected from the vehicles using a particular road segment would in the long run cover the amortized costs of road construction and maintenance. The only requirement is for road planners to build more (less) road capacity where toll revenue exceeds (falls short) of these amortized costs.

The congestion toll has three coincident theoretical interpretations. First, it is a Pigouvian tax (Pigou 1947) because it levies, on the source of a negative externality, a tax that closes the gap between the social marginal cost and the private average cost. In this role, the toll causes travellers to economize on travel by internalizing the negative externality they create. Second, because a road can be viewed as a (congested) public good, the congestion toll in the long run serves to equate the marginal benefit of road capacity with the marginal cost of supplying it, the Samuelson rule for the optimal finance of a public good (Samuelson 1954). The toll itself is a marginal benefit since it measures the reduction in total travel cost that would be realized if one more unit of road capacity were to be added, while the marginal cost of the capacity is the cost of purchasing the additional capacity. Third, since the aggregate toll revenue from the road segment is equal to the land rent the road would fetch in an alternative use, the aggregate toll is equivalent to a confiscatory tax on the owners of the land, the Henry George rule (George 1879). On the view that the land used for roads is privately owned and operated by competitive or contestable firms, the Pigouvian pricing described above would be the outcome of profit maximization, and the aggregate toll revenue would confiscate the profits of these private road owners. On the alternative view that the land used for roads is owned by society, the congestion tolls are the fees travellers pay society for the right to use the road, and in the long run these fees add up to the rent on land, provided land markets are competitive.

Keeler and Small (1977) empirically estimated what congestion tolls should be in the San Francisco Bay Area on the assumption of fixed land use. The effects of tolls on urban form have been studied within the naïve theoretical urban model that assumes all jobs are at a central point (Arnott and MacKinnon 1978) or a central point and a suburban ring (Sullivan 1983). Simple simulations based on such models show small efficiency gains of up to one per cent of income for reasonably congested cities. Recent studies, based on modern assumptions of completely dispersed employment, show similar efficiency gains (Anas and Xu 1999). All of these studies show that congestion tolls could significantly reduce travel times. But the welfare benefits of tolls would come mostly from changes in travel mode and the timing of travel during the day, rather than from land use adjustments.

Congestion tolls have become more popular in recent years and have seen such prominent implementation as in central London. But the correct calculation of first-best tolls is a quagmire. Chief among the difficulties is the fact that one must know how the value of travel time is distributed among travellers using the same road segment. If I share the road with higher (lower) income drivers, the toll on me should be higher (lower). Without knowledge of the distribution, accurate first-best tolls cannot be computed because values of time vary so widely among people. A second difficulty is that road use varies enormously throughout the day, requiring that first-best tolls should similarly vary. The problem is simplified somewhat by dividing the day into peak and off-peak periods. A third difficulty is that the technology used to detect congestion and calculate tolls should not be so obtrusive on travel as to create more congestion than the tolls would alleviate. Automatic vehicle identification by several means is feasible and not expensive. This may contribute to a wider use of tolls in the future.

Although the calculation of first-best tolls is highly daunting, a number of second-bests are available. Tolls levied on major roads but not on local roads may be effective second-bests. A tax on the market price of parking in heavily congested destinations such as the downtowns of major cities would achieve some of the efficiencies of first-best tolls. Taxes on gasoline are not nearly as effective, because gasoline usage is not closely related to the congestion created on a trip. Such taxes heavily penalize driving on congestion-free roads, for example.

Unlike highways, rail transit should be priced as a regulated natural monopoly. Since marginal cost is below average cost at any scale, marginal cost pricing ensures efficiency but requires a subsidy to the transit operator to cover fixed costs. Thus for transit systems, theory tells us that fares should be set to cover variable operating costs, while other taxes should be used to purchase the fixed inputs, including land (right-of-way). The debate then, should be about what these other taxes should be. Considerable evidence exists showing that land around transit stations appreciates in value after a transit investment is announced or constructed. Anas and Duann (1985) used an empirically estimated general equilibrium model to predict prior to construction that residential property values around the proposed stations of the Chicago Midway line would increase, with the increase sharply tapering off with distance from the stations. They estimated that the aggregate increase could pay for about 40 per cent of the construction cost. McMillen and McDonald (2004) used ex post data on housing sales and confirmed that these predictions were accurate. Taxing such windfall gains is one source of revenue for fixed facilities, although there are practical complications about how to accurately measure and document the land value appreciation in a legal-administrative context.

Transportation as a Tool to Shape Land Use

It has been observed that the underpricing of road travel, especially as it relates to the unpriced congestion externality makes travel cheaper than its marginal cost. This not only causes excessive urban expansion but also induces planners to use faulty cost-benefit measures and thus invest in too much road building as argued by Kraus et al. (1976). Excessive road capacity in turn reinforces the excessive urban expansion.

In view of the many pecuniary externalities of transportation, and since perfect pricing is not possible, a combination of judicious capacity provision and land-use zoning to ensure better accessibility to main roads and rail lines could have significant benefits. Such economies of transport–land use interdependence may be possible to exploit in urban planning and urban design at the level of smaller areas and neighbourhoods. Boarnet and Crane (2000) have examined whether land use policy and urban form can significantly affect travel behaviour in such settings. Similar concerns exist at the macro urban level (Gordon et al. 1989). In the future, planners could use such knowledge when major decisions are made on how much capacity to supply, where to supply it and how much to restrict development around it. More often than not, however, when urban planners intervene with land use controls they may fail to find the golden rule, causing distortions in land markets that could outweigh the efficiencies that can be gained by influencing travel.

See Also