Final report of ITS Center project: Road pricing
For the Center for ITS Implementation Research
A U.S. DOT University Transportation Center
ROAD PRICING
Principal
investigator: Kenneth Button
Center for Transportation Policy, Operations and
Logistics
School of Public Policy
George Mason University
Fairfax, VA
22030
703-993-4647,
kbutton@gmu.edu
March 2004
Disclaimer
The contents of this report reflect the views of the authors, who are responsible for the facts and the accuracy of the information presented herein. This document is disseminated under the sponsorship of the Department of Transportation, University Transportation Centers Program, in the interest of information exchange. The U.S. Government assumes no liability for the contents or use thereof.
CONTENTS
References 98
Project abstract
Road pricing, sometimes called congestion pricing or value pricing, aims to apportion scarce road space by market pricing rather than queuing. As increasing population densities in growing metropolitan areas press on available resources, pricing has been used to deal with land, communications facilities, public transportation service, waste disposal, and building construction. But in the case of roads queuing, not pricing, apportions available capacity. Value pricing could correct this anomaly. Road pricing can also provide useful information to motorists about congestion, as, for example, SR 91 in Southern California, by continuously adjusting toll rates to reflect congestion levels, the system communicates to motorists approaching the route just how bad congestion ahead is. At the same time, there is at present widespread political resistance to road pricing. This study will examine issues of political feasibility of road pricing.
1. congestion pricing: standard theory and the latest advances [1]
1.1 INTRODUCTION
Road traffic congestion is a major problem in all large cities and is becoming increasingly common on some inter-urban highways. The efforts to build a way out of the problem has not proved successful largely because once constructed there has been limited effort to ensure the new capacity is appropriately used. Efforts to attract people out of their automobiles have been equally ineffectual partly because of land-use patterns in the US are seldom conducive to the economic provision of transit and, in any case, many people do not see buses or local rail services as offering comparable services to the automobile at the current prices being charged for road use.
Road Pricing is a simple concept that extends the common practice that is virtually ubiquitous in every other sector of a market economy whereby prices are used to reflect scarcity, and to allocate resources to those that can best use them. In most places road space, even in such supposedly market orientated societies as the US, is in actuality allocated in a manner more akin to the general practices employed in pre-1989 communist Russia, namely by waiting in queues and lines.
While there are some that see merit in using waiting as an allocation device (Barzel, 1974), by and large society finds congestion inefficient and wasteful. After many years of adding capacity in an attempt to reduce congestion, a number of transport and road authorities at national, regional and local level are seeking to move away from the centralist approach to roads policy to one that has at least a veneer of economic rationality underpinning it.
The introduction of Area Licensing to Singapore in 1975 is the classic case study of a pioneering application [2] , but despite the success of the measure it is often seen as not being particularly relevant for other cities, and especially those in Western Europe and North America. It has taken over a quarter of a century for another scheme to essentially replicate it. Certainly there have been toll rings introduced in a number of Norwegian cities. High occupancy/toll lanes have emerged with the opening of Californian’s FASTRACK north of San Diego where there is capacity being priced on a real time basis in lanes that run parallel to free lanes. Although a number of other initiatives have also taken place, they have all generally been driven primarily by the need to find revenues for engineering works rather than as a policy to allocate traffic on extant networks.
Travel demand is mainly a derived demand. Travel is usually demanded not for its own sake but as a means of consuming some other good or service. Because the activities with which transport is associated vary over time, the demand for transport is not constant over time. For example, many towns and cities experience traffic congestion during commuting times (morning and evening), and holiday routes experience seasonal congestion.
Transport infrastructure in the short run has a finite capacity. When users of a particular road begin to interfere with other users because the capacity of the road is limited, then congestion externalities arise. Although some degree of congestion is desirable, or capacity would be under-utilized most of the time, excessive congestion is not. The question then is, what is the optimal level of congestion?
The economic costs of traffic congestion can be calculated using the engineering concept of the speed-flow relationship. [3] If a straight one-way street is assumed, the relationship between speed and flow over time typically looks like the relationship depicted on Figure 1.1. Flow (defined as the number of vehicles or passenger car units [4] , PCU, past a certain point) is dependent upon the number of vehicles entering a road and the speed of traffic. Hence, at low volumes of traffic, high speeds are possible, constrained only by the capability of the vehicle and the legal speed limits. As the number of vehicles trying to enter the road increases, vehicles affect each other’s speeds and slow one another down. As more traffic enters the road, speeds fall but, up to a point, flow will continue to rise because the effect of additional vehicles outweighs the reduction in average speed. This is the congested branch of the speed-flow curve. At the point where increased traffic volume ceases to offset the reduced speed the road’s capacity is reached at the maximum flow or capacity of the link, indicated k on Figure 1.1. At that point the flows turns unstable, with the characteristic stop-start conditions, typical of a traffic jam. If vehicles keep on entering the road further drops in speed and flow will result. This is known as forced flow or hyper-congestion. As flow keeps on decreasing, average speed will eventually increase and jump back to some point along the normal flow portion of the speed-flow relationship.
|
The actual form of the speed-flow relationship depends on a number of factors including width of the lanes, grade, road curvature, speed limit, weather, mix of vehicle types, etc. (Button, 1993; Lindsay and Verhoef, 2002).
Walters (1961) was the first one to translate the backward bending speed-flow function into a cost function. Today’s standard transport economic analysis still uses that framework to compute congestion costs and optimal pricing in static models. Thus, the speed-flow curve can be converted into a time-flow curve. Multiplying time per unit of distance (mile or kilometre) by the value of travel time and adding vehicle operating costs (monetary units per unit of distance) gives the average social cost-flow curve, ASC (Morrison, 1986). This is shown on Figure 1.2.
The ASC is a reverse of the speed-flow curve seen on Figure 1.1, with the positively sloped portion corresponding to the negatively sloped section of the speed-flow curve. As speeds tend to zero, time and therefore costs, tend to infinity (Verhoef, 1999). The MSC curve is also shown. It represents the extra cost the additional user places on the existing traffic flow. The MSC approaches infinity as flow approaches capacity (Morrison, 1986).
 |
 |  |
 |
 |
 |
 |
 |
Figure 1.2 Derivation of the cost-flow relationship
Although the hyper-congested portion of the speed-flow function and its associated backward bending ASC curve with multiple equilibria has recently been the subject of much study (Small and Chu, 1997; Verhoef, 1999, 2003), for economic analysis it is common to ignore it (Button, 1993) and this is what shall be done here. [5]
When a demand curve is added on
to the upward sloping portion of the ASC, Figure 1.3 is obtained. Road users are
assumed to be identical except for their marginal willingness to pay to traverse
the link, represented by the demand curve, D, equal to the marginal private benefit
(MPB) which, for simplicity shall be assumed to be equal to the marginal
social benefit (MSB).
Figure 1.3 The simple diagram of congestion pricing
According to standard economic theory, the optimal flow is qe, where marginal social cost equals marginal social benefit. The actual flow however will typically be q0, because road users ignore the congestion they impose on others.
The ASC and MSC curves reflect the average and marginal generalized costs associated with different flows; they show time and vehicle operating costs borne by road users when making trips. They can be seen as representing social costs, in the limited sense that they are costs to the society of road users.
Any individual driver entering the road will only consider his time and vehicle operating costs, including the congestion costs he will have to bear, which with many users, will be equal to the average cost prevailing at that moment or ASC. Thus the ASC is often referred to as the marginal private cost (MPC), or cost the new user will bear. He will not however take into account the costs that he will impose on other vehicles already on the road. The difference between the ASC and MSC curves at any flow level reflects the marginal congestion cost.
From a social point of view the actual flow, q0, is excessive because the q0th motorist is only enjoying a benefit of q0C but imposing costs of q0M. The additional traffic beyond the optimal level qe can be seen as generating costs q0MDqe but only enjoying a benefit of q0CDqe. This yields what in transport economics is called the social cost of congestion or deadweight welfare loss, given by triangle DMC.
It should be noted that at the optimal flow qe there is still a congestion externality, given by segment DE. The difference is that at this point the congestion externality is internalized, and drivers are paying for the full social costs of their trips.
One idea to optimize the level of congestion and ensure that qe is achieved, is to use the price mechanism to make travelers more fully aware of the congestion externality they impose on others. The idea is that motorists should pay for the additional congestion they create when entering a congested road. [6]
The optimal congestion charge reflects the difference between the MSC and the ASC, as defined in Section 1.2. On Figure 1.3 this optimal congestion charge is equal to DE. That is the charge that equates MSB and MSC at the efficient level of traffic. Congestion pricing generates a welfare gain of DMC. Traffic flow is reduced from q0 to qe, resulting in some motorists not using the road any longer and thus loosing consumers’ surplus of BCD. At the same time however, the road authority collects revenues FDEG. From these revenues, FDBA are a transfer of consumers’ surplus enjoyed by road users to the road authority in the form of revenue. These revenues can (and perhaps should) be returned to road users. [7]
In mathematical terms the following model can be used to compute the efficient congestion charge. If the average social cost per km of a representative vehicle is
(1.1)
where a is the cost per PCU.km (pence per PCU.km), and b is the value of time (pence per PCU.h) and the total social cost of a flow of q vehicles is C = cq, then when an additional vehicle is added to the flow, the total social cost will be increased by
(1.2)
The congestion externality or marginal congestion
cost is given by 
. This externality measured at the efficient level of traffic qe
gives the optimal congestion charge DE on Figure 1.3. It can be numerically estimated
by specifying a speed-flow relationship, v, and making the relevant substitutions
in Equations (1.1) and (1.2). There are innumerable case studies that have sought
to place a value on the optimal road price indifferent circumstances. These vary
in their degree of sophistication and Table 1.1 lists a few of the earliest North
American efforts. Recent work have tended to make use of more sophisticated models
are a less easy to tabulate.
Table 1.1 Some early North American findings relating to road pricing.
| Study |
Place |
Fee at date of Study |
Traffic effect |
Comments |
|
Cheslow (1978) |
Berkely (1977) |
$2.00 per vehicle per morning peak |
Reduce Traffic by 30% |
Part of a transport package including bus provision |
| Spielberg (1978) | Madison (1977) | $1.00 per vehicle per morning peak |
Reduce Traffic by 30% |
Part of a transport package including bus provision |
| Gomez-Ibanez and Fauth (1980) |
Boston (1975) |
$0.5.-1.00 per vehicle per peak period |
Reduce Traffic by 30% | |
| Keeler and Small (1977) | San Francisco Bay Area (1972) |
$0.02-0.09 per vehicle per mile (suburban) $0.06-0.35 per vehicle per mile (central areas) | Optimal tolls, differentiated by road type to allow for differing long run cost | |
| Elliot (1975) | Los Angeles | $0.03-0.15 per vehicle per mile (central areas) |
Reduced congestion by 20-30% |
Also reduces smog significantly |
| Kraus, Mohring and Pinfold (1976) |
Twin-Cities Area (1970) |
$0.03-0.09 per vehicle mile (suburban) $0.05-0.15 per vehicle mile (central areas) | ||
|
Dewees (1978) |
Toronto (1973) |
$0.38 per vehicle mile per peak period | Specifically an estimate of the average marginal social cost of a trip over a range of streets | |
| Mohring (1979) | Minneapolis and St. Paul | $0.66 per vehicle mile per peak period $0.33 per vehicle mile per off-peak period | Estimates of optima user charges |
The standard analysis presented in Sections 1.2 and 1.3 has its roots in work that was carried out in the 19th and early 20th centuries. This section honors that work as without it, today’s understanding of the problem of traffic congestion and congestion charging would probably be different.
1.4.1 The French engineers
The French engineers of the 1840s-1850s were as concerned with how their structures should be financed and used as with their physical construction. Dupuit’s (1844) work on the provision of public goods, their pricing and investment assessment is seminal [8] . Dupuit’s well-known analysis of bridge pricing (actually he has six examples scattered throughout his work and they are not entirely consistent), argues for a pricing structure that maximises utility whilst at the same time covering up-keep and capital costs. With an economically optimal capacity this is essentially what a congestion charge does.
Minard (1850) also recognised this and, although he thought primarily in terms of physical wear-and-tear, his work also had an implicit congestion component in it. He pioneered the fact that travel time-savings have economic value and suggested simple, revealed preference based, ways of measuring it. He did not, however, possibly because of the context of his work that mainly concerned uncongested facilities, explicitly link time savings into his infrastructure pricing concepts.
At the turn of the 20th century, the French engineering concepts were not widely known yet. In part, the originators were outside of mainstream economics and the fact that their material was not widely available in English did not help. Perhaps more importantly, the railway age removed much of the interest in charging for road infrastructure. The track costs issue debate switched to considering the appropriate way for railways to recover their fixed costs without a regulated environment designed to limit their quasi-monopoly powers. Matters akin to the later work on Ramsey pricing and the like came to the fore.
1.4.2 The Pigou debate
Perhaps the most frequently cited name in the development of Road Pricing is that of Arthur Pigou’s. His book The Economics of Welfare (Pigou, 1920) laid the foundation for much of the subsequent academic literature in the field. [9]
Pigou’s overriding objective in The Economics of Welfare was to systematically try to bring positive analysis to an area of study that had previously been dominated by normative arguments. Indeed, economics was still often treated as moral philosophy at the time. He made the case that roads were not being utilized efficiently because users were not being charged for the congestion costs they imposed on others. Many would not use the facility if they had to bear this cost and their resources and time would be more gainfully employed doing something else.
His analysis looks at two competing roads, one wide but with a poor track and hence slow, and the other narrow but with a good track. Traffic will disperse itself between these alternatives up to the point where travel time to the destination is the same irrespective of the road used. He argues that imposing a differential charge on the narrower road will redistribute traffic that does not value travel time savings very highly to the wider, slower road. The result is that aggregate travel time is reduced and society as a whole gains. [10]
The 1960s saw a sudden interest in congestion pricing. Two seminal papers were published by leading academics of the day (Walters, 1961; Vickery, 1963), and in the UK a government study provided the first major policy analysis of road pricing (Ministry of Transport, 1964). The interest grew largely from practical considerations. Automobile ownership was expanding and cities were getting congested. Land-use planning, urban design and infrastructure expansion were in vogue as mechanisms for combating this but were seen as long term and costly. Subsidizing public transport as an alternative to the automobile was under review but had not been widely adopted. In this context, road pricing was seen as a potential means for containing commuter traffic whilst leaving road users options about when, how and where they could travel.
The Vickrey/Walters approach very much followed the lines of Pigou’s analysis. There would be some decision on what the optimal traffic flow should be, essentially determined by estimating the traffic speed-flow relationship underlying the cost curves on Figure 1.3 together with the demand function. The congestion charge would then be an efficient way of attaining the target flow of traffic. It was seen as a fiscal instrument allowing those who gained the most utility from road use to use the facility. The model presented on Figure 1.3 was developed in the 1960s and is still in use today. The basis for that model are of course much older and as stated above, they go all the way back to the work by Pigou (1920) and the engineers and physicists that had been studying the relationship between speed and flow since the 1930s.
Vickrey (1968) combined his analysis with a comparable one for urban public transport. While the analysis was technically more rigorous than that of Pigou’s in that congestion was carefully defined, the argument was essentially the same – roads are being publicly provided and in the absence of a market they should be used to maximise social welfare.
A slightly different conceptual approach within this vein that emerged at the same time as the Vickrey/Walters analysis is to treat roads as club goods (Buchanan, 1965). This approach has the intellectual attraction of isolating the allocation of road space debate from debates centered on more conventional Pigouvian externalities (that are retained in all of the editions of The Economics of Welfare) involving the impacts of traffic on non-road users. This latter group embodies such things as environmental effects that are outside of the market.
A club good approach strictly assumes that a group of people derive sufficient benefit from a facility to provide it, exclude non-members whilst allocating out its use amongst members according to incremental costs (which would include a scarcity cost). [11] Since road users are often willing to pay above their incremental costs to use a road, debates arose about ways in which commercial road builders would operate such a system. It may also be tied into a road charging regime that would involve a ‘membership fee’ such as an annual license, and then user fees that capture wear-and-tear costs plus an element to reflect scarcity and congestion at various times.
1.6 Recent Refinements In SHORT-RUN PRICING
The upsurge of interest amongst academics, as well as practitioners since the 1960s has led to refinements in assessing how the congestion charge should be calculated. As Vickrey (1963) pointed out, there is in ideal circumstances a need to vary the price according to traffic demand and costs. The type of day, the traffic mix, the physical features of the network and local road conditions (such as the weather and accidents) may influence these.
Vickrey (1969) developed a bottleneck model, where congestion is assumed to arise when vehicles queue behind a bottleneck. All commuters wish to arrive at work at a certain time but there is a bottleneck with finite capacity that will not allow all of them to arrive at their preferred time. There are costs associated with early and late arrival, which together with the toll, are added to the cost of the trip, which commuters try to minimise by choosing their departure time. Queuing time evolves during the rush hour and this imposes a time pattern of departures. The evolution of congestion over the rush hour is thus endogenously determined. This bottleneck model was further extended, mainly by Arnott, De Palma and Lindsay (1988, 1990a, 1990b, 1992, 1993, 1994, 1998) to include heterogeneous commuters, route choice, stochasticity in capacity, elastic demand, and time-varying tolls.
A second generic form of the dynamic congestion model, initially developed by Henderson (1974), does not completely eliminate travel delays at the social optimum (Chu, 1975). Both approaches involve the distribution of travel delays and scheduling of costs of the peak, and the duration of the peak in the untolled equilibrium and the social optimum, being determined exogenously.
Road pricing is designed to produce a Pareto optimal use of a facility but this is dependent upon the standard assumptions that surround first-best partial equilibrium analysis and especially that all other prices are equal to marginal costs. Moving into the realm of ‘second-best’, where the assumptions of the Pareto world are relaxed, is less tidy and inevitably the efficient price becomes situation specific.
One particular issue that is often addressed in this context is the pricing of substitutes to road transport, and especially public transport. In many cases subsidies have been given to public transport to attract trip-makers away from congested streets. Irrespective of whether this has effected any significant transfer, this is an instrument that could be used to optimize the modal split of traffic. A road user in the absence of road pricing will only consider the marginal private cost of a trip that is lower than the full marginal social cost of the road space taken up. A subsidy to public transport, accurately reflecting the cross-elasticity of demand between the modes, could theoretically be used to attract sufficient motorists from the roads to limit road use to qe in Figure 1.3.
Other topics within second-best pricing have been the pricing of only part of a road network, or pricing subject to constraints (Verhoef et al, 1996; Liu and McDonald, 1999; Verhoef, 2000, 2002a, 2002b; Small and Yan, 2001). When some routes are priced and some are not road users are left with a choice between a facility where access is unpriced but the service quality poor and one where there is an access price but congestion is optimized. [12] The road price introduced on such routes requires to reflect the opportunity cost of not using the unpriced roads as well as the first best toll on the priced road. Ignoring the cross-effects will lead to sub-optimally high use of the unpriced facility. Marchand (1968) was one of the first to look at this type of situation in terms of setting a charge that weighs the welfare benefits of reducing traffic on the unpriced facilities against the losses from deviating from first-best pricing on the priced route.
Second-best considerations may also become important in the context of complements to, as well as substitutes for, a priced facility. Parking is, not unexpectedly, the most explored complimentary good (Glazer and Niskanen, 1992; Verhoef, 1996, Calthrop et al 2000) although even here the literature is remarkably thin. Indeed because of the complementary nature of road use and parking, parking charges have in some cases been are treated as an alternative to a road price (Verhoef et al, 1995). Although parking charges often involve low transactions costs the circumstances where they can act as complete substitutes for road pricing are very stylized.
In a first-best world there would be optimal parking fees reflecting the opportunity cost of taking up land to ‘rest’ a vehicle as well as the congestion costs associated with cruising around seeking a parking spot. [13] In practice, parking is often provided free or perverse charging regimes are employed. Such regimes are frequently structured to limit certain categories of users such as long stay users irrespective of willingness to pay. Parking fees should ideally be structured to reflect the traffic congestion caused by drivers cruising around looking for a parking place. Parking costs can also affect traffic demand by influencing the lengths of trips made over a network. Low parking fees, being a fixed cost of a trip, are spread more over a longer trip than a shorter one. This may affect congestion in sub-urban areas.
A further input into urban car trips that has received a lot of attention is that of environmental costs. There are few that argue that motor traffic does not make considerable demands on the environment and that the costs involved are external to the market. These effects are both wide-ranging going through the emission of a variety of local, regional and global atmospheric pollutants, to noise, ground water contamination, through to visual intrusion, and in some cases very large. A road price that reflected the total marginal social cost would internalize both the congestion and the environmental externality, and would include a congestion charge and an environmental charge. This is a large subject area, the issues complex, and the literature substantial and is not dealt with here. Fuel taxes however can approximate optimal environmental charges because most environmental externalities are closely dependent on fuel consumption. [14]
There is another side to the second-best debate that should perhaps just be touched upon. There has been a mounting interest in how other road taxes, public transport fares and road investment decisions can be made in the absence of optimal road pricing. In this context Sherman (1971), Bertrand (1977), and Arnott and Yan (2000) have looked at pricing across multi-modal systems; Wheaton (1978) developed ideas of second-best road investment strategies; and Chia et al (1996) examined the appropriate fuel taxation policy to pursue if appropriate congestion charges are absent.
Much of the early work on congestion pricing treated all road users as being identical, Strotz (1965) being an exception. Variations in their income and, linked to this the valuations that they place on travel time saving and the size of vehicles driven, which affects road take, are now appreciated as being important considerations for both efficiency and equity reasons. At the very least, they raise questions about whether the congestion charge should reflect these features. If the concern is purely with efficiency and standard first-best conditions hold then there is no need to consider these variations with time varying congestion pricing. This is because all that matters are the congestion costs imposed. Arnott and Kraus (1992) provide a rigorous proof of this.
Other situations that have attracted attention include those when drivers have different preferences for speed that are unrelated to income. Here the models have often tended to exclude the possibility of overtaking making them particularly relevant to bridge and tunnel cases. The results suggest that the congestion charge needs to be higher for slow vehicles to reflect their impact on slowing to higher speed vehicles. But that the differential, because slow vehicles affect fewer fast vehicles on average and also because the speed of fast vehicles declines asymptotically to that of the slower vehicle, should decrease with the proportion of slow vehicles in the traffic stream. This approach would require differential approaches to congestion pricing depending on drivers’ speed preferences – in practical terms this may be done by having different pricing regimes for cars and trucks. (Dafermos, 1973).
Added to these challenges, even if public transport is not an issue, the network is simple and road users are homogeneous, the standard model of congestion pricing excludes any consideration of implementation and operation costs like for example, the costs of collecting the fees and policing the system. [16] These may be significant and may become entwined with other issues such as how much information the road user has on the ‘product’ that is being purchased (De Palma and Lindsey, 1998).
The idea of using electronic technology to collect a real time, variable congestion charge was initially advanced by William Vickrey (1959) in evidence to a US congressional committee as early as the late 1950s. Despite the largely successful testing of electronic road pricing equipment in Hong Kong (Hau, 1990) [17] , operating costs, issues of privacy and concern over technical reliability have favored area based systems. This means that a road price does not vary directly with traffic conditions but rather location or time of day is used as proxy for congestion levels. The Singapore and the recent London schemes use a combination of these surrogates largely for pragmatic reasons.
The efficiency of these discrete charges is inevitably less than a continuous congestion charge, the loss of efficiency being a function of just how closely the proxies reflect changing congestion levels. This in turns often depends on the number of ‘steps’ involved and the prices charged at each step. Empirical estimates of the extent to which these systems lose efficiency are found in Chu (1999) and Arnott et al (1993, 1998) and they seem to be significant.
A market price serves not only to allocate current facilities optimally but it also provides signals to where capacity should be expanded, and the revenues from the price provide resources for that expansion. In a perfect market situation, effectively Knight’s position, there is no requirement for separate consideration of capacity expansion – the net revenue flows give the necessary guidance. Given the reality of public ownership of most road track and the possibility of internalizing the congestion externality with a corrective congestion charge, there have been a number of studies looking at the link between the optimal congestion charge and road investment strategies (Mohring, 1970).
An important result that provides an additional incentive for the road agency to internalize the congestion externality is that the optimal congestion charge, as defined in Section 1.3, equal to the difference between the MSC and the ASC at the efficient level of traffic, covers the fixed operating cost, equal to the sum of interest, depreciation and policing, provided there are constant returns to scale in road construction and maintenance and capacity can be increased in continuous increments (Mohring and Harwitz, 1962). This result was later expanded to include damage. In this sense, a road charge equal to the sum of the optimal congestion charge and the road damage charge, covers the fixed operating cost as defined above, and the variable maintenance cost, provided there are constant returns to scale in construction, maintenance and use of road capacity (Newbery, 1989).
Under constant returns and optimal pricing, whenever there is economic profit, the capacity of the road should be expanded, and when there is economic loss, the road has been overbuilt and should be abandoned or closed. The optimal pricing and investment decision for roads can thus be dealt with in a single model (Mohring and Harwitz, 1962; Keeler and Small, 1977). When there are diseconomies of scale then at the optimal capacity and with an optimal congestion charge, economic rent will be earned. This reflects a scarcity rent on a fixed factor of production, in this case, the land. Equally, if there are economies of scale, and in some instances when there are large indivisibilities in investments, the optimal congestion charge will not generate sufficient revenue to cover the fixed operating cost. Subsidies will be required and techniques such as cost-benefit analysis come into play in determining their level.
Much, therefore, depends on the view taken about the nature of returns to scale and indivisibilities in road investment. Kraus (1981) argues that the existence of lanes implicitly means that there are indivisibilities in supply. In contrast, Starkie (1982), following the legacy of Keeler and Small (1977) looks at a road network as a whole and argues that within that context, actions such as resurfacing, changes in lane width, etc, can be varied incrementally. Hau (1998) argues that increasing, decreasing and constant returns to scale all exist in road transport and each case should be assessed individually.
Whilst road pricing results in an overall welfare loss to road users, some of those who continue to use the roads do benefit from faster trips. Others, while paying a higher monetary price for their road trips have this at least partly off-set by enjoying time savings. Time is a significant cost component of any trip and its value is therefore of fundamental importance. Indeed different values of travel time and travel time savings can lead to different answers on the viability of a transport project.
The extensive literature on the subject (Small, 1982; Hensher, 1997; Calfee and Winston, 1998; Wardman, 1998; Hague Consulting Group et al, 1999; Jara-Díaz, 2000; Lam and Small, 2001; Whelan and Bates, 2001; Mackie et al, 2001, 2003; Brownstone and Small, 2003) suggests there is ample evidence that the value of time varies amongst population subgroups (Small, 1999) according to income, gender and age, and also with trip purpose, mode used, length of the trip, prevailing conditions of congestion and sometimes weather, and the value given to the travel time variation depends on the sign (loss or saving) and the magnitude, and its ratio to total travel time.
As a consequence of this, each congestion pricing proposal will need to be evaluated on an individual basis, paying particular attention to the socio-economic characteristics and trip patterns of the road users to be affected. The value of the travel time savings will play a substantial role in the net benefits to be derived from any transport project, including congestion pricing schemes.
As Small and Gómez-Ibáñez (1998) observed, ‘Winning political approval for any form of congestion pricing project is difficult in a democracy, even with careful planning.’ There may be a number of reasons for this but the political economy of road pricing has largely centered around the distribution implications of the policy. Indeed Andrew Evans (1992) sees it as the critical practical issue to large-scale application. This is hardly surprising since it is the road user that immediately loses as a result of charging for congestion. While economists tend to focus mainly on the efficiency implications of transport policies, politicians and administrators put more emphasis on the various effects on different groups in society.
Overall the road user would be worse off with congestion pricing, although within that some users who value travel time savings and reliability of service highly would gain. Those concerned with distributional effects essentially accept that congestion pricing may result in a potential Pareto improvement in as much as theoretically those who gain could be ‘taxed’ to yield compensation to those who lose out, and still enjoy a net benefit. The problem however is that congestion pricing projects tend to lack any actual compensation mechanism.
As a first step in introducing incidence into the analysis it is necessary to know exactly who gains and who loses, and in what ways. The distributional impacts will typically depend on the nature of the congestion pricing regime. While there are important theoretical contributions in this field, the studies have been mainly empirical in their approach (Vickrey, 1968; Small, 1983; Richardson, 1974; Segal and Steinmeir, 1980; Foster, 1975; Layard, 1977; Santos and Rojey, 2003). Chapter 12 in this book assesses the welfare and distributional effects of high occupancy/toll lanes and other road pricing policies in Metropolitan Washington DC.
What emerges from the mainly empirical literature on the distributional impacts of congestion pricing is a lack of consensus. Layard (1977) for example, argues that the policy is regressive. Foster (1975) argues that it is progressive. Giuliano (1992) takes a somewhat different position, namely that even if congestion pricing is regressive, the fact is used to disguise other objections to the policy rather than necessarily being important in itself. Santos and Rojey (2003) conclude that the distributional impacts depend on the design of the scheme and the geo-economic characteristics of the town in question (where do people live, where do they work and how do they get to work).
A related topic to the distributional effects is the implications of various ways of using the revenues from a congestion charge. Some, such as Small (1983) and Andrew Evans (1992) see this as the core of the political economy debate.
Newbery (1990, 1994) suggests that revenues from congestion pricing in the UK could to be returned to motorists in the form of lower fuel duties. Motorists as a group would then be made better off as they would pay the same total road user charges but would experience less congestion.
Goodwin (1989, 1990) and Small (1992) both propose that revenues should be allocated in thirds, one third to the reduction of taxes in general, or fuel taxes, or vehicle license fees or to the increase in social expenditure, one third to investment in public transit, and one third to investment in new roads or reimbursement to trip makers. Although arbitrary, this allocation in thirds would aim at compensating losers and promoting improvements in the transport sector.
The Lex Report on Motoring (Lex Services, 1998) based on two surveys carried out in England in 1997 suggests that the few motorists that would favor congestion pricing would do so as long as it were implemented in the form of charges rather than taxes, and the revenues stayed in the transport sector.
Ison (2000) conducted a survey in which, amongst other things, he compared public support for urban road pricing before and after the question relating to the use of revenues had been asked. He found that the acceptability increased from 11.3% to 54.6% after it was explained that revenues would be allocated to specific, clearly established objectives. Harrington et al (2001) found between 7% and 17% increase in support to congestion pricing when use of revenues was specified in a survey they conducted between residents in Southern California.
The 2003 RAC Report on Motoring Making the Most of Britain’s Roads (RAC Foundation for Motoring, 2003) finds that 69% of motorists would be more supportive of road charging if revenues were spent on transport over other public services. 49% would support the idea if revenues were spent on improving existing roads.
In a purely competitive private market, large amounts of revenue would signal that road use was highly valued by users of the roads and the net revenues would go towards enhancement of the road network that would bring down the future cost to motorists of using it. In a second-best situation characterized by a variety of market and government failures across the economy this is not automatically the best way to meet concerns about compensation.
In any case, returning revenues to the transport sector and attempting some form of compensation to lower-income groups can only increase public and political acceptability.
Despite all the energies of academics that have been expended, together with the time of a number of policy makers, the textbook application of congestion pricing has not yet emerged and it probably never will. The few applications of the idea are simplifications of the concept of the Pigouvian charge. This is perhaps inevitable, and the theoretical work should, at best, be viewed as providing important benchmarks against which success is measured.
The introduction and impact of the Area Licensing system in Singapore offered a perspective on the power of the price mechanism to control traffic flows – until then elasticities of demand for road space had been little more than conjectures on academics note pads. As seen below, estimation techniques seem to have improved but there would appear to still be a tendency to under-estimate response levels.
The subsequent experiments with electronic technologies in Hong Kong to make charges more reflective of real-time congestion illustrated that more sophisticated methods could be deployed if desired (Dawson and Catling, 1986). This has been supported in many places where conventional tolling has reverted to rapid pass collecting techniques.
Subsequent years have in many ways seen the initial impacts of the Singapore scheme and the Hong Kong experiment, limited as they were, be diluted. Indeed, this piece actually reflects some of the factors affecting this. Other miracle cures for traffic congestion problems, ranging from intelligent transportation systems that manage traffic more efficiently through the network to the aim of encouraging more teleworking to reduce the number of daily commuters, have come along, and largely gone away again after making a limited, albeit sometimes useful, contribution. None, however, have confronted the basic problem that road users do not take full account of their actions on others.
The London Congestion Charging Scheme initiated in February 2003 (see Chapter 2) however has attracted considerable attention from around the world and may be the landmark for the beginning of a new era on the application of congestion pricing [18] .
The introduction of high occupancy/toll lanes in California provides useful insights into both viable technologies and has generated considerable amounts of new information about the sensitivity of road users behavior when confronted by variable road charges. The Norwegian toll rings (see Chapter 2 for details), initially conceived to generate revenues to fund transport infrastructure, may now be changed into time-varying tolls to manage demand and reduce congestion. There is a new breeze in transport policy that might be announcing that the times are about to change and congestion pricing may finally become an instrument used widely around the world.
It seems almost inevitable that motor traffic will grow well beyond the immediate future. This will impose additional strains on the road infrastructures of most countries. The associated impacts of longer commutes, and higher freight distribution costs have to be borne somewhere in the system. On top of that there are now additional costs that have recently come to the fore, most notably in the US, linked to security. While congestion has traditional economic costs, an excessively congested network also has costs in terms of impeding response to incidents. Reacting to single events by the emergency services often poses problems at present, but in context of a major catastrophic occurrence, large scale evacuation for example from many city centers would be chaotic. This realization may accelerate the adoption of more efficient traffic pricing.
Those responsible for road provision are moving to make use of fiscal tools to manage the normal economic concerns over traffic growth. One can debate why there has recently been an up-surge of interest in congestion pricing. It may reflect a sudden appreciation that roads are really no different to other goods and can be treated in similar ways. Markets may not be possible but at least some effort can be made to make more rational use of road space. It may also reflect a crisis. Other measures have either been designed to increase the available capacity with no form of allocation mechanism other than queuing, or have involved providing alternatives that road users could have used before but declined. Both of these broad approaches have been found wanting and have become ever more expensive to pursue. The shift in emphasis is thus towards treating the root course of the problem rather than the symptoms.
2. USING the revenues from Congestion pricing
2.1 INTRODUCTION
Urban traffic congestion is a major and growing problem not only in cities but also on many major inter-urban arteries. Using the price mechanism to control the problem has now moved on from textbook discussions and to become a central theme of political agendas in many countries. The evidence shows that road pricing can be an efficient mechanism for controlling congestion. Road pricing would also provide the road authorities with a significant fiscal dividend. The issue examined here is how the dividend should be spent.
Advocates of road pricing have long argued that pricing can in practice, as well as theory, improve the use of existing urban road networks and inter-urban systems. Mathematical models of bottleneck pricing, system wide pricing, second best principles and so on, as was seen in Chapter 1, now abound (Button and Verhoef, 1998; Lindsey and Verhoef, 2001; Verhoef, 1998). Technological advances have also made road pricing more relevant as tolls can now be collected electronically with no reduction in traffic flow. Electronic tolls also allow real-time price adjustments to reflect variations in traffic flow.
Simpler systems, such as ring tolls, whereby road users pay every time they cross a cordon, or area licenses still have some advantages, however, such as letting users know in advance of travel what the exact road price will be. Ring systems may also be cheaper although that advantage is fast diminishing as electronic toll technology improves. Simpler regimes may provide important measures along the way to electronic pricing as they illustrate the power of prices to influence travel behavior and create political support for more efficient methods of pricing (Behbehani et al, 1984).
The fact that there is no ideal method of charging for road congestion does not mean that there are no significant social benefits from deploying imperfect systems. Indeed, the power of road pricing as a mechanism for limiting traffic in urban areas has been amply illustrated by the few instances where it has been initiated (Table 2.1). Indeed, judging by the reactions of many of those questioned over the years, efficiency would not seem to be an issue, or at least not an over-riding one. There are wider concerns. [19]
Much less attention has been given to the issue of how to spend road pricing revenues than to calculating optimal toll prices. [20] The expenditure issue is not one that has been entirely ignored although the relevant material is rather fragmented. [21]
Table 2.1. Effectiveness of road pricing and quasi-road pricing, international experience
| Instrument |
Reduction in road usage |
| London (UK) cordon | –30% per day |
| Durham (UK) cordon | –90% per day |
| Bergen (Norway) cordon | –6% to 7% per day |
| Oslo (Norway) cordon | –5% to 11% per day |
| Rome (Italy) cordon | –20% per day |
| Trondheim (Norway) cordon | –5% to 10% per day |
| Seoul (South Korea) peak tunnel toll differential |
– 24% peak reduction |
| New York/New Jersey (US) small peak beridges/tunnel toll differential |
–7% (am) and –4% (pm) |
|
Paris (France) A1 peak toll |
– 13% peak reduction |
| Singapore (1992) evening peak | – 53% in peak-hours |
Much less attention has been given to the issue of how to spend road pricing revenues than to calculating optimal toll prices. [22] The expenditure issue is not one that has been entirely ignored although the relevant material is rather fragmented. [23]
If roads were provided privately and competitively then the inefficient congestion problem would not exist. Road owners would price to maximize profits and in so doing would take into account the costs of congestion each driver imposes on other drivers. [24] In addition the issue of net investment would also be taken care of as prices would accurately convey information about where and when to build new roads. [25]
Prices administratively set by the authorities, however, will rarely mimic those that would be set by the forces of supply and demand. The authorities, for example, must rely on estimates of the underlying market parameters in setting charges and these are inevitably not totally accurate and in any case bureaucratic incentives are unlikely to be coincident with the profit and loss incentives of entrepreneurs. Administrative charges, however, are generically a more efficient mechanism than devices such as queuing and congestion to allocate scarce publicly owned resources (Baumol and Oates, 1975).
The problem of the surpluses can be seen from another, welfare economics based, perspective. What congestion charges do in theoretical terms is to produce a Hicks-Kaldor-Scitovsky improvement in welfare. This means that there would be an aggregate social gain from such charges. What they do not do is produce a Pareto improvement that would require that no one is made worse off by the charges. This is, to reiterate the situation, because the main immediate direct beneficiaries of road pricing are the pricing authorities that take the windfall net revenues of pricing. It is a Hicks-Kaldor-Scitovsky improvement because these authorities could compensate those road users who lose out, pay for the administration of the system and still have surplus revenue. Whether this compensation is actually given is a crucial matter when looking at the implications of the distribution gains from road pricing. It also determines the degree of opposition to it.
Pricing in a market context serves as an indicator of when more of a commodity should be provided and as a mechanism for generating the revenues necessary to expand supply. Hence in the longer term even those initially priced out of the market may have greater opportunities to consume if they are willing to wait and then pay the price that emerges after long-run output is optimized. But with the adoption of the Hicks-Kaldor-Scitovsky criteria in the road pricing context, there is no mechanism that automatically offers a stimulus for such an expansion of road capacity. The way net revenues could be spent is divorced from where they are collected and from the demands of previous and remaining road users.
2.2 THE PROBLEMS OF SPENDING
It's often suggested that road revenues be used to offset license, car or fuel taxes. [26] Sharp (1966), however, pointed out this complicates the issue of pricing since it would cause road users to ‘buy-back’ some of the road space. Reducing fuel taxes, for example, would increase the incentive to drive. If the money is instead dispensed to non-related areas, say to the elderly, then this buy-back would likely be negligible. Given modern information systems and charging technologies it is still not a trivial matter to determine feedback effects but total perfection in setting any price is a utopian notion.
When looking at distribution issues there is sometimes confusion between the positive aspects of describing changes from the status quo with normative consideration of whether the effects of road pricing are in someway ‘inequitable’. Robbins’ (1932) view, that economics should be purged of value judgments, has merit in this respect. Whilst analysis of the implications of road pricing for different groups in society is unquestionably important in understanding why it is not widely used, whether the distribution of these impacts is fair or not is a matter for others.
Sharp touches upon distribution issues. He argues that those potential motorist’s ‘excluded’ from the road would only be as well off after the initiation of road pricing if public transport “was so improved by the decreased congestion that it provided an alternative service which the ‘excluded’ motorists regarded as being as (and no more costly than) the road space which they formerly consumed.” The matter of distribution effects is not, however, dealt with in any detail.
The problem of spending road pricing revenue was resurrected in a little more detail by Richardson (1974). He argued that road pricing has important distribution implications in terms of who is affected. It benefits the wealthy who would gain from the combination of them having a high value of time and the ability to make faster trips. It benefits existing, largely low income, urban bus users because their services would be faster and more reliable. It has adverse effects for middle income groups priced off the road and having to transfer to what they consider inferior public transport. Use of revenue from road pricing is seen by Richardson as a mechanism for, at least partly, compensating this group if it is expended on public transport improvements, especially if they improve service quality.
Gomez-Ibañez (1992) goes to a further level of disaggeregation and looks at eight groups who either directly gain or lose from road pricing. While his findings are not seen as reflecting any judgment about whether the distribution outcome meets any criteria of equity, the wide-ranging impacts are seen as reflecting the difficulties of obtaining political acceptance for road pricing. The degree of disaggregation can be taken further. For example Giuliano (1994) considers implications according to gender, opportunities to work flexi-time, availability of transit, etc. rather than income classes. But whichever way it is looked at, there are inevitable winners and losers and who they are is influenced by the way road pricing revenues are spent.
What these studies tend to neglect is that a full network based system of road pricing would not only increase the price of using congested facilities but would logically reduce the costs of using many uncongested sections (notably inter-urban roads). This was an implicit point made by Newbery (1988) in the context of the UK road network. The overall implications for distribution of impact across the motoring public would then depend on the relative use of different elements of the network as would the total net revenues collected. There is an argument, that in some countries the outcome could be revenue neutral and the distributional issue be reduced to the impacts on various groups of road user (Roy, 1998).
Since road pricing does not meet a strict Pareto criteria and, because it is unlikely that revenues will be used to fully compensate those adversely affected, there is a pragmatic need to ensure that net revenue is spent in ways that make road pricing attractive to a winning coalition of people. Winning coalitions are not simply coalitions of road users but may include a variety of other interests with ‘voting rights’. Since comprehensive road pricing would involve charging for both existing infrastructure and for new infrastructure the issue is not even a static one amongst road users. Successful efforts at initiating quasi-road pricing measures have often involved concessions for certain road user groups (car pools in Singapore), improvements to alternative modes (subsidies to transit in Oslo), increased choice for car users (HOT lanes in California), or additional road capacity (toll rings in Norway).
Arguing for the need to form dominant coalitions of advocacy to gain acceptance for road pricing a number of academic studies have sought to look at ways of in which distributing the proceeds of road pricing may assist in this process. Goodwin (1989) came up with the idea of a ‘Rule of Three’. This, rather arbitrary, rule would involve distributing the proceeds, although not in equal part, to improve the road system, to improve public transport, and to reduce taxation or to be spent as part of general expenditure. The improved road network would appease those forced to pay the road price and in the longer term reduce the number priced from the system. Improved public transport would partly placate those forced from the road as well as gaining the support of existing users. Lower taxes or increased general public expenditure would appeal more generally.
This theme of triple dispensation has been taken up by Small (1992) who looks at ways of using the net revenues from road pricing to improve transportation and make the package more attractive. He argues that regional wide pricing in the Los Angeles region would generate about $3 million annually that could be used to off-set regressive taxes and especially those such as fuel and sales taxes that are used to finance public transportation. And in particular, to gain the support of those in a small number of relatively concentrated geographical areas the revenue could be used subsidize local transportation.
The economic difficulty with putting large sums into public transportation, however, is that it may be wasteful and result in much of the benefits of the system being lost. In the past subsidies have been captured by the suppliers of public transportation services and by the labor operating those services. Studies of the UK bus industry prior to regulatory reform in the 1985 Transport Act indicated that a large proportion of subsidies, up to 50%, was absorbed in lower productivity and higher wages (Bly et al, 1980). There are ways of containing but not eliminating this problem. The regulatory changes that introduced network tendering in London in 1985, whilst not eliminating the problem, would seem to have reduced the seepage. Kennedy (1995) estimates the cost saving emanating from the London regime was about 20% over the previous regime of subsidies.
Many economists dislike the earmarking of revenues, [27] their tying to particular expenditure items, in the ways that Goodwin, Small and others have discussed. Beside the fact that they are arbitrary, the argument is that any rational company in the private sector does not limit itself in the ways in which it spends its revenue flow. It seeks out those activities that generate the highest potential profits. Indeed, if this were not the case then there would be little innovation in the economy as investments would exclusively be channeled back into existing activities. The parallel argument for a public sector undertaking is that government should use any revenue that it gets to finance activities that produce the highest social rate of return.
The analogy of a government budget to the revenues of private firm, however, is flawed. A private firm faces a direct market test, if consumers do not regard its products as good value at the prices that it charges the firm will go bankrupt. Revenues are thus tied to consumer benefits. Taxes, however, are not prices and we have no direct way of assessing whether taxes paid are more than matched by benefits received. Voting is an indirect method of testing whether government is performing at par but voting is at best only a coarse test. Votes are typically registered only every four or more years and they are made over large bundles of goods that the voter cannot untie (a voter, for example, cannot vote for one party's position on the military and another party's position on the environment).
To begin the analysis of earmarking with the argument that it would be inefficient to constrain a private firm to spend its revenues in prescribed fields is to miss the point. The problem that earmarking attempts to address is one that private firms do not face. Earmarking tries to ensure that citizens can see how taxes paid are matched by benefits received.
For these sorts of reasons, Buchanan (1963) and others have advocated the use of earmarking. [28] Earmarking is seen by advocates as more transparent than general taxation and provides the populace with a clearer picture of the options confronting them and allows subsequent monitoring of actions. Accountability is stronger. Earmarking does not necessarily imply putting the revenues from the revenue back into the same activity from which it is raised but rather into some pre-specified expenditure package. From a public policy perspective, it provides ‘voters’ with choices of policy packages.
Earmarking, however, need not work in this way. Lottery funds "earmarked for education," for example, rarely result in more being spent on education just a reshuffling of fungible revenues (Borg and Mason 1990). In these cases earmarking is a form of fiscal illusion (Lee 1997).
Earmarking may also be necessary to create a dominant political coalition. [29] In some cases such coalitions may have some economic logic as when road revenues are earmarked for road building – thus ensuring that road users are not inefficiently subsidizing non-users and non-users not subsidizing users. In other cases, however, especially when the revenue and expenditures sides are not closely related, the coalition may be purely political.
The FASTRACK HOT lanes on an eight-mile section of I-15 north of San Diego are priced on a real time basis and run parallel to free lanes. Stretches of private road in Orange County California are priced in accordance with a pre-established time-table. Opposition from road users to the charge was placated in part by the fact that the revenue from the charges is put into maintaining and improving the road. This benefits not only the user of the FASTRACK facility but also users of the free parallel road that is now also less congested.
The evidence from numerous surveys also generally indicates road users favor the earmarking of net revenues from road pricing. For example, a National Economic Development Office (1991) study in London found that 43% of adults questioned supported road pricing as a stand-alone measure but this figure rose to 63% when hypothecation of revenues for purposes favored by respondents was of part of the package. In particular, 38% wanted it spent on public transportation, 25% on road improvements and 14% on non-transportation services such as the National Health Service.
Jones (1991) surveyed a large number of studies that have looked at attitudes towards road pricing in the UK and found that a high level of public acceptance required that revenues be dedicated to local transport and environmental projects. More recent generic findings are contained in the UK Department of Environment, Transport and the Regions (2000) consultation document, Breaking the Logjam. Forty-four percent of respondents said that road pricing revenues should be earmarked for better public transportation. [30]
The experience of FASTRACK, small as the sample is, indicates that private sector road suppliers find putting surplus revenue back into road infrastructure a commercial proposition, at least with current capacity and demand parameters. It also reflects the fact that prices do not simply ration out scarce resources but also act to stimulate an expansion in those resources. The difficulty with the public sector is that while earmarking to a portfolio of outlays may obtain a political coalition in favor of road pricing, this portfolio may well reduce the efficiency of the regime and be highly wasteful of resources. Earmark systems, even when seen as efficient in curbing government failures to spend wisely in the short-term are notoriously rigid in the face of changing circumstances.
2.3 BERGEN AND OTHER NORWEGIAN SCHEMES
Despite all the difficulties with introducing road pricing, a number of systems with some of the characteristics of road pricing are in place. The Norwegians have introduced from the late 1980s a series of cordon-pricing regimes, ‘toll rings’, around a number of their cities (Trondheim, Bergen and Oslo). These essentially involve payment for entry to the city. The geography of Norway, with its mountainous terrain and small cities with limited access points makes this a relatively straightforward way of charging road users. The monies are collected through both electronic and manual systems. Over time, as practical experience has been gained and as technology has developed, so the systems have become more technically sophisticated. In all cases, the initial technology was not state-of-the-art but well tested to ensure that minimal operational problems would arise. Details of the various to toll ring schemes, together with some data on the cities themselves, are provided in Table 2.1.
The schemes are not strictly road pricing in the Pigouvian sense. They are more in the tradition of tolls. They do not directly relate to levels of congestion and, indeed, in some cases there are discounts offered to frequent users of the systems. As can be seen from Table 1 the charges are also relatively low. The positioning of the toll rings was also not designed to optimize traffic flows. They were located largely to capture revenue in what was seen as a politically equitable fashion. Nevertheless, since a charge is levied in the cordon schemes, it is almost inevitable that there will be some impact on travel behavior and, ipso facto they have implications for congestion.
In all cases there is a large element of earmarking in the ways revenues are used and this has also been seen as having longer-term implications for traffic congestion.
In the case of Oslo where a ring system was initiated in 1990, for example, an argument has been advanced that since the revenue collected is used largely to construct ring road facilities this will ultimately contribute to reducing congestion in the central area of the city. A further 10% of revenue is earmarked for public transportation that, it is argued, has an additional positive substitution effect (Solheim, 1990).
Table 2.1. The Norwegian Toll Rings (data for 1992)
Bergen Oslo Trondheima
Population of urban area (thousands) 300 700 136
Percent inside toll ring 10 28 40
Starting date of toll ring January 1986 February 1990 October 1991
Number of stations 7 19 11
Entry fee for cars (NOK)b
Single trip (manual or coin)c 5 11 10
Per trip (subscription)d
With prepaymente 4.50 7.43 7
Off-peak discount (after 10.00pm) NA NA 2
Monthly passf 100 250 NA
Time charges are in effect
Days Monday-Friday All days Monday-Friday
Hours 6.00am-10.00pm All hours 6.00am-5.00pm
Average daily crossing during toll hours (thousands) 68 204.4 40.5
Percent by subscribers 59 63 85
1992 gross revenue ( million NOK) 63 628 70.7
Notes:
a Figures exclude the pre-existing Ranheim toll station which has higher rates applicable in both directions and at all times.
b For 1992 the exchange rate was NOK1 = $0.16.
c Bergen: all stations manned. Oslo: all stations manned, 8 have coin lanes. Trondheim: 1 station manned, others coin or magnetic card only.
d In Trondheim subscribers are charged for no more than one trip per hour and no mkore than 75 per month. Trondheim subscription rates rose in 1994 for people making 10 or fewer crossing per month.
e charges shown are for the following prepayment quantities. Bergen: booklets of 20. Oslo: 350 trips. Trondheim: NOK 1500 prepayment. A post payment option is also available in Trondheim.
f Six and twelve month passes are available at lower rates.
Source: Small and Gomez-Ibañez (1998)
The Bergen system of cordon tolling was initiated in 1986 (Larsen, 1988) and is perhaps the most studied of the Norwegian rings. The city is small and the initial toll ring had six stations with another added after the completion of a new road link. The evidence from the outset was that, whilst the tolls were nor specifically designed to impact on congestion, they did lead to some limited spreading of peaks. The initial system covered the period 6.00am to 10.00pm each weekday and hence did not differentiate peak flows. The result was a very rapid reduction of traffic in the urban area by some 6% to 7%. Small reductions were initially recorded in Oslo, but within nine months traffic had built up again in the urban core. The Trondheim system has the least impact on congestion because about a third of the region’s drivers live within the toll ring (Ramjerdi, 1994). [31]
Whilst there are technical issues of toll collection technologies and the like that are of interest in the Norwegian case, of more importance is perhaps the political mechanisms by which the authorities gained acceptance for their policies. In all cases there was earmarking of revenues. In the case of Oslo, the introduction of a toll ring was coincidental with the opening of an express bypass that would be financed from toll revenues. Despite this, surveys indicate that only 29% of those questioned in Oslo before the toll rings were introduced had a positive view of the scheme, with 7% in Trondheim. The figures rose to 39% and 20% respectively after a short period of operation. While clearly still not a majority, this does indicate a rapid change of opinion for a large number of people. Similar feelings were expressed in Bergen.
The crucial point about the Norwegian toll rings is rather less the impact they have had on traffic congestion in the short terms, and the support that some of the schemes have given to the technology possibilities, than how one can gain increased political acceptance for what were at the outset unpopular measures. The ideas outline earlier for earmarking road pricing revenues for road improvements, public transportation subsidies and for reduction in other (possibly partly car related) taxes largely fall into this coalition forming framework
3.4 ROAD PRICING IN LONDON
Over the years a number of proposals have come forward for the introduction of road pricing in London. Much of the debate in the Smeed Report of the early 1960s was about London. While attracting public and political attention, the idea was abandoned in favor of institutional and urban planning reforms with a focus on improving public transport.
Road pricing was reconsidered in detail in the 1970s and major studies of the implications of a system of ‘supplementary licensing’ were undertaken. A number of variants were considered but an option involving a daily fee of £2.00 (1973 prices) per vehicle driving on inner London roads between 8.00am and 6.00pm on weekdays was subjected to considerable scrutiny. The simulations indicated that this would reduce peak traffic in the 3.4 square miles of Central London and raise speeds by up to 40% (May, 1975).
The 1980s saw a restructuring of the local government in London and another resurgence of interest in road pricing (May et al, 1990). This also coincided with a shift in national transportation priorities with more emphasis on traffic management and making better use of existing networks. Cordon pricing was examined in detail with London divided into six zones with all-day charges for crossing screenlines between the inner zones and peak charges only for crossing the boundaries of outer zones. Estimates of the most favored toll structure found that traffic peak traffic in London would decline by 15% with a 25% fall in Central London.
Another set of studies were completed in the early 1990s looking mainly at cordon pricing around Central London deploying somewhat higher fees than previous work had assumed (ranging up to £14.00 for a peak inward crossing). The cordon charges were explored in combinations with changes in parking and public transportation policies. The models indicated that traffic speeds would rise in Central London by up to 32% with traffic volume falling by up to 22%. Additional traffic reductions could be attained adopting more cordons but only at the cost of more complex and expensive structures (Bates et al, 1896).
A further restructuring of London’s government has brought about a much less academic approach to road pricing. UK transportation policy shifted in 1998 to explicitly facilitate local government to initiate road pricing within their jurisdictions allowing them, at least at the outset of any scheme, to retain all revenue generated. More germane here, the Greater London Authority Act 1999 conferred road user charging powers in Greater London on the Authority. Since that time the mayor of London has been developing plans to introduce a simple form of road pricing within the capital largely based on the report of the Government Office for London (2000). The report concluded that a daily charge between 7.00 am and 7.00 pm for cars of £5.00 per day (£15.00 for trucks) enforced by digital cameras would reduce traffic by about 12%. The study also found that a daily charge of this magnitude would generate about £200 million annually (later revised down to £120 million, dependent on exemptions).
The outcome was the Central London Charging Scheme that came into effect in February 2003. This introduced the £5.00 for driving in the central zone of London between 7.00am and 6.30pm on weekdays. The charge for commercial vehicles is £5.50 to reflect higher collection costs. The scheme operates with a paper based system using automatic number plate recognition technology via cameras located at cordon entry points or in mobile locations throughout the charging area. Some vehicles categories (e.g., taxis, motorcycles and buses) are exempt. Those who live within the zone enjoy a 90% discount on the charge.
The results for the first year indicate congestion has fallen by 30% within the central zone, that traffic levels are down by 18% with a 30% reduction in the number of cars and that bus patronage has gone up by nearly 30,000 passengers during the morning peak. The latter is partly due to bus reliability increasing – additional wait time at bus stops is down 20% across London and by 20% within the charging zone. There is little impact on traffic levels immediately around the central zone. The greater than expected traffic impact combined with legal issues of enforcement has meant revenues were only £68 million in the initial year [32] , although the figure is anticipated to rise to £80 million to £100 million in future years (Mayor of London, 2003).
Here we are not concerned with whether £5.00 is a reasonable figure, whether the exemptions are justified, whether the enforcement technology is appropriate, or whether the temporal and spatial boundaries. Rather attention is with the implications of £100 million or suddenly becoming available to London’s public authorities and with the distribution of impacts and the ways it is spent. These latter considerations are important because the desk-top studies supporting the daily charge indicated that reduced traffic congestion would, despite any financial payments, generate benefits for bus users and freight transportation and for medium and lower income car users. This, however, runs counter to the general arguments of Richardson cited above and seems somewhat counterintuitive.
The UK legislation only guarantees that a local authority may maintain the rights to this for 10 years. This is clear, if limited, earmarking in the sense that the ring fencing implicitly means the revenues must be spent locally. The Government Office for London study indicated that the traffic benefits from road pricing would be greater if the scheme were complemented by improved public transportation. Its modeling efforts, however, involved only a limited number of scenarios that directly looked at taking monies from road users to fund public transportation. [33] But, equally, the models, largely because of the complexity of London’s transportation system, did not consider potential traffic gains from investing revenues into improved road infrastructure (either in terms of civil engineering schemes or improved telematics and control systems). Nevertheless, and accepting that the proposals are still in their formative stages, the general impression is that the vast majority of monies raised will go to public transportation.
The issue ultimately comes down to whether a fairly crude form of road pricing that reduces congestion in Central London is justified if its political acceptance depends upon revenues collected being earmarked for public transportation. The successful adoption of the Norwegian schemes was largely predicated on the basis that the majority of revenues would go to clearly defined road projects. In some cases public transportation would receive some limited resources. While it may be true that public transportation in London is in need of additional investment, it is also true that the road network is in need of improvements.
Perhaps more important is the potential resource loss that could stem from road pricing revenues crowding out other sources of funding for London’s public transportation. If road pricing does reduce traffic flows by 15% or more over the long term [34] , then the inevitable outward shift in the demand for public transportation would make private investment more attractive. The subsidization of additional investment or the provision of operating subsidies effectively competes with this commercially driven finance. Unless it is clearly demonstrable that the subsidized activities generate more social welfare than those favored by the market then resources that could be utilized with benefit elsewhere are wasted. [35]
Finally, the focus on public transportation as an alternative to car use may neglect other consumer choices. There is the potential substitution of resources to other activities that may be socially prefe