Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 117
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Appendix B Theory, Experience, and Estimated Effects In 1920, long before hundreds of thousands of motorists were faced with the daily frustration of congested routes like the Santa Monica Freeway or Shirley Highway, economists developed the basic theory of road pricing (Pigou 1920; Knight 1924). In its simplest formulation, road pricing would require motorists using the roadway to pay for the congestion they cause other motorists. In the short run the optimal toll would reduce congestion to its most efficient level. Congestion pricing theory holds that congestion tolls not only would optimize the use of the current road system and generate substantial net savings in the short run, but in the long run, and if other assumptions hold, they also would generate just enough revenues to provide for demand in the future (Mohring and Harwitz 1962). In the 1960s, some four decades after the theory was developed, economists began suggesting that the theory be applied to solve the growing congestion problem (Vickrey 1959; Walters 1961). In 1975, Singapore imposed the first congestion pricing system. The pros and cons of Singapore's congestion pricing approach, which has been praised for its traffic constraint but criticized as a model for congestion pricing, are described in the second section of this appendix. Congestion pricing has been proposed in the United States before. These proposals, however, did not advance very far. Despite fairly compelling arguments in favor of congesting pricing from economic theory, political opposition caused by concerns about the potential effects on the
OCR for page 118
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion poor, opposition by downtown businesses, and resistance to the concept of charging motorists directly for the use of roads has stymied the progress of such proposals. In the first section of this appendix, the theory of congestion pricing is summarized. In the second section, the experience with congestion pricing abroad is reviewed, and in the third, estimates are provided of the possible effects of congestion pricing if applied in U.S. metropolitan areas. THEORY An Overview The traditional approach to resolving congestion problems has been to expand capacity by adding new lanes or building new roads. This approach is problematic for both obvious and subtle reasons. Among the more obvious reasons, (a) in most urban areas addition of more lanes would be very costly because of the high cost of real estate, especially at a time when governmental agencies at all levels are short of funds, and (b) in some areas capacity cannot be enhanced because of concerns about air pollution or community opposition, or both. Among the more subtle reasons is the recognition that building new capacity induces new demand or shifts in existing demand that soon congest the new facility (Downs 1962). Economists offer an alternative to building our way out of congestion that would instead change the behavior of some road users. They observe that whenever the price of using some scarce, valued good does not increase as demand increases, that good will be in short supply. Shortages will be acute if supply cannot be readily enhanced. This is typical of goods in industries with high capital or fixed costs. Throughout the economy, when demand for some commodity or service exceeds supply, the price tends to rise until demand and supply are in balance. The basic theory of congestion pricing for roads has changed little since Knight's (1924) formulation.1 Speed–traffic flow curves plotted by traffic engineers indicate that as the volume of traffic on a road approaches design capacity, speeds and traffic flow decline sharply ( Figure B-1). Once capacity has been reached, the addition of motorists into the traffic stream causes the flow of vehicles per lane per unit of time to decrease, resulting in the 1 This discussion draws heavily from work by Morrison (1986) and Hau (1992b). Hau provides an extensive, nonmathematical treatment of congestion pricing theory.
OCR for page 119
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion FIGURE B-1 Derivation of the speed-flow curve (Morrison 1986). backward-bending curve in Figure B-1.2 A general interpretation may clarify this relationship. Consider a road with a design speed of 97 km/hr (60 mph) on which traffic is moving at an average speed of 80 km/hr (50 mph). At that speed, a traffic lane would accommodate 900 to 1,000 vehicles per hour. As traffic increases, speeds might fall to 56 km/hr (35 mph), but the lane can actually carry more traffic each hour—nearly 1,900 passenger cars—because at slower speeds the space between vehicles is reduced. As capacity is reached, however, the lane starts to become so crowded that speeds fall below 56 km/hr and capacity diminishes sharply. At very high volumes, the risk that traffic flow will degenerate to stop-and-go conditions is quite high. The shape of the speed-flow curve as demand approaches capacity provides a key clue to the theory of congestion pricing. This theory can be illustrated by examining how delay increases as the flow of vehicles per lane-hour increases. The speed-flow curve shown in Figure B-1 can be inverted from kilometers per hour, which measures speed, to hours per kilometer, which measures trip duration (Figure B-2a). Speed is at its maximum in low traffic volume; hence delay is at its minimum. But as the number of drivers in a traffic stream increases, the average delay at each level of flow increases. This is shown by the curve labeled AVD in Figure B-2a. Because the average delay at each level of flow is increasing, it follows that the contribution to delay by each additional driver is increasing 2 Although the backward-bending portion of the curve is shown as a smooth parabola, traffic engineers consider the entire area highly unstable. Once the volume is at or near capacity, any interruption can cause significant declines in flow, resulting in a rapid transition from peak flow to stop-and-go traffic (Transportation Research Board 1985).
OCR for page 120
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion FIGURE B-2 Derivation of the cost-flow relationship (Morrison 1986). (a) Trip duration; (b) trip cost. even more; this marginal delay is illustrated by the curve labeled MRD in Figure B-2a. The marginal delay is the increment in delay for each incremental increase in traffic volume. As can be seen by examining these two curves, each additional driver's contribution to delay is much larger than the increased average delay that he or she experiences (the marginal delay curve is much higher than the average delay curve and is rising faster). The average delay can be converted to a dollar cost by multiplying it by the average value of time. After variable operating costs are added to cover fuel and maintenance, it becomes an estimate of short-run average variable cost (AVC in Figure B-2b). Similarly, the marginal delay curve is converted to a short-run marginal cost (SRMC) curve, which gives the cost to all drivers of adding one more driver to the traffic stream during the same time period (Figure B-2b). As with the delay curves, these two cost curves are closely related and based on the same information, but are computed differently. Average variable costs are the total variable costs (TVC) per unit of flow at each level of traffic flow (F) (or TVC/F) and the short-run marginal cost is equal to the change in total variable costs with each change in traffic flow [(TVC1 – TVC2)/(F1 – F2)]. In the absence of a congestion toll, the quantity of traffic flow will be that which occurs at the point where the average cost curve intersects the
OCR for page 121
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion demand curve (Figure B-3). As can be seen, marginal costs are considerably higher than average costs at this point. Additional drivers joining a congested traffic stream, absent a congestion toll, may only be aware of the average cost they will experience and are largely unaware of the increased costs they are imposing on all other drivers. The theory of congestion pricing has been aptly described by Mohring and Anderson (1994): Urban travelers both experience congestion and contribute to it. Inducing the operator of a vehicle—any vehicle—to remove it from a traffic stream would save not just its occupants' own time but also the time cost they would otherwise impose on other travelers by adding to the road 's congestion level. This time cost is the change in time per trip the departed vehicle would have produced times the number of vehicles that change would have affected. FIGURE B-3 Optimal congestion toll and welfare loss (Morrison 1986).
OCR for page 122
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion A commonly invoked rule of safe road behavior is that drivers should stay one car length behind the vehicles they follow for each ten miles an hour of travel speed. If all travelers follow this rule and all would travel at 60 miles per hour (i.e., would take one minute to travel a mile) on an otherwise unused expressway, there would result the relationships between the instantaneous ratio of actual traffic volumes to “ideal” capacity (about 2,000 vehicles per lane-hour on an expressway) and the average (AC1) and marginal (MC1) travel times per mile that are given by the solid curves in Figure 1. Curve AC1 depicts the travel-time costs that individual travelers directly experience; curve MC1 includes these costs plus those that each vehicle operator imposes on others by adding to congestion. Curve AC1 illustrates a commonly observed phenomenon of urban-expressway travel: maximum traffic flow occurs at about 30 mph. Above 30 mph, lower speeds result in increased traffic flows. In the top, backward-bending portion of AC1 where it takes more than two minutes a mile, however, further speed reductions lower traffic flows. Figure 1. Relationships bertween Volume/Capacity Ratios and Travel Time
OCR for page 123
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Peak-period travel does not take place at a constant rate but, rather, gradually increases to a peak then decreases. Someone traveling at the peak of the peak period is more likely to experience the backward-bending portion of curve AC1 than is someone who travels at the beginning or end of the peak. Still, both peak-of-the-peak and fringe-of-the-peak travelers do almost always get where they are going; a peak-of-the-peak trip just takes longer. We have, therefore, used marginal and average travel-time relationships similar to those given by the dashed curves, AC2 and MC2 in Figure 1 in deriving the toll estimates that are reported in Section V.6 6 The formulae for the solid and dashed average cost curves are, respectively, N/K = 4(1 – ṫ/t)ṫ/t and t = 1 + (N/K)4 where N/K denotes the ratio of vehicle volume to ideal capacity and t and ṫ are respectively actual travel time per mile and travel time per mile at a zero N/K ratio. The marginal delay cost imposed on other drivers is made up of costs that are external to a motorist's choices. The existence of external costs in a transaction indicates that decisions are being made without accounting for their full effects on others. When highway capacity is allocated according to the average cost, the total loss to society due to these external costs is simply the sum of all delay costs in excess of the most efficient level. This loss is represented by the shaded area above the demand curve between the average and marginal cost curves (Figure B-3). This shaded area represents the excess delay imposed on all motorists because of the consumption of highway capacity by drivers in excess of their willingness to pay if charged the marginal cost. The optimal traffic flow would occur at the point where the short-run marginal cost curve intersects demand, but without somehow making drivers internalize the external costs they impose on others, the optimal flow will not occur. The optimal flow would be achieved by requiring each additional driver to, in effect, compensate all others for the delay he or she causes by joining the traffic stream. This toll (t in Figure B-3) would equal
OCR for page 124
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion the difference between the average variable cost and the marginal cost at that point. When motorists are faced with a toll that represents the full social cost of using a congested highway, those who value the use of the less-congested road at that time will pay the toll and save time. Those who place less value on the use of the highway at that time will make other choices, such as delaying their trip, traveling on another route, traveling by transit, or deferring the trip. In the short run, defined as that period during which capacity is fixed, the time savings that result from the imposition of the optimal toll makes society as a whole better off. Over the long run, the willingness of travelers to pay the full marginal cost would give a better indication of the demand for new capacity. It can be shown that pricing highway capacity investment needs according to their marginal cost will generate revenues just sufficient to cover the cost of providing new facilities (if other assumptions discussed below hold) (Mohring and Harwitz 1962; Keeler and Small 1977). Theoretical and Practical Challenges Although theory provides a compelling case for congestion pricing, it has been challenged on at least three fronts.3 As described in the following three sections, (a) the assumptions incorporated by theory have been questioned, (b) concerns have been raised about the effects on low-income road users, and (c) concerns have been raised that the pricing system adopted in practice will not meet the efficient pricing criteria assumed by economists. Regarding the last concern, even one of the strongest proponents of congestion pricing has noted that “experience with the pricing of public services is not such to give confidence that in practice a close approach to an efficient optimum can be achieved” (W. Vickrey, unpublished paper, 1992). Questions About Assumptions As indicated in the previous section, congestion pricing has a strong theoretical rationale for maximizing the efficient use of existing capacity. In addition to its short-run benefits, marginal cost congestion pricing 3 Hau (1992b) discusses the full range of theoretical challenges to congestion pricing.
OCR for page 125
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion should also lead to an optimal allocation of resources in the long run. This additional advantage has been questioned. It can be challenged whether an optimal allocation of resources will occur with congestion pricing because of potential violations to three assumptions: perfect competition, no scale economies, and a specified income distribution. As described below, there are practical responses to the criticisms that these assumptions are not perfectly met. Moreover, it has been shown that whereas the conditions listed are sufficient conditions for marginal cost pricing to be optimal, they are not necessary (Ng 1977). One argument against marginal cost pricing optimally allocating resources in the long run comes from the theory of the “second best”; that is, market imperfections in aspects of the economy not governed by market pricing can still lead to inappropriate price signals and less-than-optimal resource allocation. The validity of this argument in the case of urban transportation depends on the degree to which pricing in the economy approximates the competitive norm assumed by microeconomic theory and on the existence of substitutes for transportation. The former argument could apply in limited sectors, but Meyer et al. (1965) judge the economy as a whole to be fairly competitive and conclude that substitutes for transportation seem to be a limited concern. Though they acknowledge that empirical evidence is less than compelling, Meyer et al. suggest that “second best” problems are easily exaggerated and conclude that adoption of marginal cost pricing in transportation would probably improve the general welfare. In the long run, in the case when the supply of highway capacity is not fixed, theory also predicts that an optimum toll may be just sufficient to cover the cost of providing the facility. This outcome, however, is more complicated because it depends on the assumption that returns to scale are constant. If returns to scale were increasing rather than constant, for example, congestion tolls would not generate sufficient funds to cover the cost of the facility. If returns were decreasing, marginal cost pricing would generate more-than-adequate revenues. In the latter case, if supply is not enhanced, a long-run income transfer will take place. Either of these two outcomes would violate the optimal resource goal of congestion pricing. Hau's (1992b) review of the studies examining the issue of returns to scale in highway construction indicates that evidence is available on both sides; he concludes that surpluses and deficits would coexist. Other economists (Meyer and Gomez-Ibanez 1981) conclude that economies of scale probably offset diseconomies of scale. Regardless of the existence of scale econ-
OCR for page 126
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion omies in the long run, however, short-run congestion pricing will still generate the most efficient use of the existing capacity. Hau (1992b), among others, argues that the revenues earned from congestion pricing should be reinvested in the transportation sector to appropriately respond to transport demand. Diversion of the funds to nontransportation uses may help compensate groups initially opposed to congestion pricing but could also result in less efficient expenditures on transportation facilities. The use of the revenues is a key issue in developing a congestion pricing proposal that satisfies economic criteria. However, that congestion tolls cover capital and operating costs on an “optimally designed, constant-returns-to-scale road network ” does not mean that all the revenues should be spent on capacity. In the United States most capital costs for highways have been paid on a “pay as you go” basis. Thus revenues earned from pricing such highways need not be earmarked for additional investments in capacity. Concerns About Regressivity Congestion pricing is often advocated as a fairer method of providing for highway capacity than the current user tax system because it requires those motorists who wish to use the capacity during the peak period to pay a premium. A limitation on the improved fairness of congestion pricing is that motorists have different incomes and different abilities to pay (although it should be noted that the existing sources of highway finance, mostly gasoline and property taxes, along with sales taxes in some areas, are also regressive). 4In economic welfare theory, the lack of equally distributed impacts is dealt with by making payments to those disadvantaged by a change in policy from the revenues generated by the policy change. In this case, lower-income highway users could be compensated by using some of the revenues earned from congestion pricing. Although the adverse impact on the poor as a group can be mitigated with side payments, compensating all adversely affected individuals would defeat the purpose of congestion pricing, since its success depends on the shift by some users to other modes or times for their trips. Compensation to adversely affected groups could be provided, however, without undermining the effect of congestion pricing on demand. But the nature of this compensation is central to achieving the 4 Of course, in the private economy air travelers and restaurant customers also have different abilities to pay, but there is little social concern about charging for these services on the basis of peak demand rather than ability to pay.
OCR for page 127
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion welfare gain. In the short run, Small (1983) and Hau (1992b) have shown that before the funds earned from congestion pricing are reinvested, the majority of highway users are made worse off by congestion tolls. Judicious uses of the revenues can reverse this loss. Practical Concerns Many students of congestion pricing have discussed the difficulty of valuing social costs (in this case the value of time) and in gaining acceptance of any estimate (Zettel and Carll 1964; Meyer et al. 1965; Altshuler 1979). This lack of certainty makes estimates of congestion tolls somewhat speculative, but some practical solutions may be available. Despite lack of consensus among analysts, estimates of the value of time are routinely used in developing the benefit estimates that underlie transportation construction projects. In addition, Wohl and Hendrickson (1984) suggest that experimenting with alternative congestion prices until the market is cleared is the simplest way of finding how society values scarce road capacity. This solution, however, is complicated by two factors. Given the problems associated with increasing transit fees and tolls, it may be unrealistic to expect public authorities to be able to experiment with prices. Second, congestion pricing is meant to reduce congestion to its most efficient level, not eliminate it. Because of the vagaries of weather and the unpredictability of some factors influencing demand for peak-period travel, some congestion will occur. The cost of trying to respond to all such influences such that congestion did not occur is likely to exceed the benefits. Hence there is a point at which a certain degree of congestion is efficient. Determining when the efficient level of congestion has been obtained, however, may be a subjective and controversial enterprise. Whether the compromises necessary in the real world would reduce or vitiate the potential benefits raises fundamental concerns. Vickrey (1959; unpublished paper, 1992), for example, has argued that an efficient pricing system would require smooth changes in price from the peak to the off peak (and vice versa) and should vary with changes in actual traffic conditions. In this scenario, drivers with different values of time would shift their travel times to match the toll they were willing to pay. This scenario would also reduce the problem associated with abrupt changes in the fee. When abrupt changes occur, such as when daily restrictions on high-occupancy-vehicle (HOV) lanes are eliminated, motorists are given an incentive to try to game the system by queueing up on connecting routes (and often blocking traffic) to take advantage of the change. A
OCR for page 129
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion proposals for congestion pricing in the United States, however, have not advanced very far. Singapore In 1975, the government of Singapore imposed the first congestion pricing system (Button and Pearman 1985, 39–46).6 This system, along with the introduction of congestion pricing on Route A1 south of Paris in 1992, are the only examples of roadway congestion pricing.7 In 1975, a 6.2-km² (2.4-mi2) area in the central city was marked off, and any motorist wishing to enter this area during the morning peak period had to display a windshield sticker, which cost approximately U.S.$1.65 per day initially. Stickers could be purchased from post offices and road-side booths. The fee was enforced by officers at about 30 entry points to the zone. Other policies were also implemented at roughly the same time to discourage automobile ownership and use. Taxes on vehicle ownership and parking rates in the downtown area were increased sharply. To encourage more transit use, park-and-ride lots (10,000 spaces) were provided on the periphery of the downtown area, and minibuses were provided to serve these new lots. The short-term effects were substantial. Traffic into the downtown area declined more than 45 percent during the morning peak. No fee was imposed for the afternoon peak, which remained quite congested until 1989, when the afternoon was priced as well. Congestion on the circumferential routes around the downtown area also increased because of the diversion of through traffic. Many of the former automobile occupants shifted to carpools or traditional bus service (Table B-1). At first, many drivers altered the timing of their trips so that they arrived just before or just after the end of the peak period. Problems in the period before 7:30 a.m. were judged by the authorities to be minor and no actions were taken, but congestion immediately 6 >This discussion about Singapore is largely drawn from Watson and Holland's (1978) evaluation, from Button and Pearman's (1985) description and literature review, and a more recent, extensive review in Annex 1 of Hau's (1992a) paper on congestion charging mechanisms. 7 Apparently Milan, Italy, has recently instituted a permit system for entering the central area. Although little detail has been published about Milan's system, Orski (1992) reports that automobile trips into the city center have declined by 50 percent and that of this group, 41 percent of motorists shifted to public transit. It appears, however, that this is a system of rationing, not pricing.
OCR for page 130
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion TABLE B-1 Mode Used for Work Trips to Restricted Zone from Vehicle-Owning Households Before and After Adoption of Area Pricing (Watson and Holland 1978, Figure 5.1) Share of Traffic (%) Mode Before Area Pricing After Area Pricing Change Bus 33 43 +10 Shuttle — 3 +3 Motorcycle 7 6 -1 Other 4 2 -2 Solo driver 32 20 -12 Carpool 8 19 +11 Taxi passenger 16 7 -9 following the pricing period was sufficient to cause them to extend the end of the period from 9:30 to 10:15 a.m., which reduced the congestion. The substantial investments in park-and-ride lots and minibuses did not facilitate a significant mode shift; the shuttles from the lots gained only 3 percent of total trips, which was far less than their capacity. Singapore is currently planning to replace its windshield sticker system with an electronic pricing system (Richards 1992). An initial evaluation of Singapore's congestion pricing experiment, financed by the World Bank, judged the proposal to have been successful because it met the goal of significantly reducing automobile congestion (Watson and Holland 1978). Button and Pearman (1985) provide a rough calculation of the economic return based on the initial changes in travel time; they estimate a 15 percent return, even when the expenditures on fringe parking lots and feeder buses, which proved far in excess of demand, are included. Although downtown congestion was reduced substantially during the morning, there have been criticisms. There was some evidence that downtown commercial activity left the area, but Button and Pearman (1985) argue that the evidence is equivocal. There was also a recession under way, which could also explain the relatively minor reduction in businesses in the downtown area. Toh (1977) criticized the plan because it charged too high a fee and argued that, instead of being a traffic management tool, it became a revenue-earner for the government. The diversion of traffic to the ring
OCR for page 131
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion road also caused considerable delay to travelers on that route, and the time losses to these travelers offset the time savings enjoyed by commuters into the downtown area. Wilson (1988) argues that the number of individuals with increased travel times exceeded those with time savings; his calculations of before-and-after utilities indicate a net loss in social welfare. Wilson (1988) attributes this to the high, flat fee charged.8 According to Hau (1992a), however, a substantial reduction in the peak-hour fee in 1989, when the program was substantially revised, suggests that this criticism is now less valid. Furthermore, Wilson 's analysis does not account for the benefits derived from the uses of the revenues (see footnote 8). The subsequent improvements in the ring road also reduced the time losses to travelers using this facility, which, over time, improved the net benefits of the program (Gomez-Ibanez and Small forthcoming). The imposition of congestion fees had clear and substantial effects on the traffic entering the downtown area. This benefit has not been disputed, but doubts about the transferability of Singapore's experience to the West have been raised. Singapore has a one-party, quasi-dictatorial government (Morrison 1986; Altshuler 1990). Also, Singapore does not have a separate city government that is distinct from the national government, which “avoids the problem of overlapping jurisdictions between urban and sub-urban areas that are common, for example, in the United States” (Watson and Holland 1978). United States Between 1973 and 1978 the U.S. Department of Transportation (DOT), through the Federal Transit Administration (formerly the Urban Mass Transportation Administration), encouraged U.S. cities to participate in congestion pricing experiments as a means to reduce congestion. Higgins 8 Wilson attempts to include both the loss in utility from pricing and the gain in utility from using the revenues by recomputing utility using the price and income that prevail after the policy's adoption. However, the assumed utility function does not fully account for income, but rather contains income only in an auxiliary manner as an indicator of value-of-time variations. As explained by Viton (1985) for precisely this kind of situation, the utility function estimated as part of a mode-choice model is necessarily incomplete in accounting for the influence of income, the missing part being that which is independent of mode. Viton shows how this can be taken into account; by not doing so, Wilson in effect omits most of the beneficial effects of the uses of revenue in his comparison of various scenarios. This is why his estimates of utility after the policy are barely affected by whether revenues are returned fully, partially, or not at all.
OCR for page 132
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion (1986), one of the Urban Institute staff engaged by DOT to assist the cities that initially responded to Secretary Coleman's invitation, reviewed the individual cases and explained why all efforts foundered. Some of the cities invited to participate refused at the outset but gave different reasons: a lack of congestion (Rochester, New York), fear of controversy (Atlanta, Georgia), and concern about the effect on a struggling downtown area (Baltimore, Maryland). Of those that agreed to participate, in Berkeley, California, the mere possibility of a study resulted in widespread and distorted media coverage, and the mayor and City Council quickly withdrew. After expressing initial interest, the Madison, Wisconsin, City Council declined to participate for a variety of reasons, including concerns about regressivity, appearance of being coercive, and potential harm to the downtown area. The study in Honolulu, Hawaii, progressed the furthest, but policy makers never expressed much support for the concept and, after the first phase of the study was completed, expressed no interest in continuing it. During this same period, congestion pricing was seriously considered in a major transportation planning exercise involving the California Transportation Commission and the California Department of Transportation. Public hearings on the plan, however, led to considerable criticism from well-organized groups such as the Automobile Club of Southern California (Button 1983). A revision to the plan that reduced the role of pricing as a policy tool did not keep then-Governor Brown from undercutting the entire exercise by eliminating any further budget support for it. Examination of the efforts to impose higher tolls and congestion tolls on major toll bridges serving San Francisco and New York, and the failure to adopt congestion tolls, reveals some significant problems (Button 1983). In San Francisco, bridge users complained that they did not have alternate routes and therefore that the tolls would be unfair. Others argued that bridge users were not the cause of the downtown congestion. Tolls to restrain use were perceived as inappropriate for facilities paid for by users, and automobile user groups objected. In both San Francisco and New York, downtown merchants, already feeling intense competition from suburban malls, opposed efforts to increase bridge tolls, even though they could have benefitted from reduced congestion. Altshuler (1990) cites similar controversy over a 1986 proposal for areawide pricing in Manhattan. Despite the theoretical benefits of congestion pricing, concerns about fairness, motorist opposition to the concept, and fears of downtown merchants blocked implementation of these plans.
OCR for page 133
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion ESTIMATED EFFECTS Estimates of the level of congestion toll and the likely impact on travel in the United States have been made in several studies (Table B-2).9 Keeler and Small (1977) estimated long-run optimal congestion tolls for the San Francisco Bay Area. Relying on the earlier work by economists such as Herbert Mohring, Robert Strotz, and William Vickrey, they developed an optimal pricing and investment model. The demand for and peaking characteristics of the San Francisco freeways were calculated from state data on vehicle travel. With these estimates, the model optimized capacity and estimated optimal long-run tolls. The tolls were to vary with the time period and type of route. If these estimated tolls were expressed in 1990 dollars, they would range between $0.05 and $0.36/VMT and would average about $0.15/VMT (Small 1992). Although not estimated by Keeler and Small (1977), the revenue potential of fees this large is quite substantial; if such fees were applied to congested areas of greater Los Angeles, for example, they would raise about $3 billion annually (Small 1992). To show how short-run optimal tolls would affect different income groups, Small (1983) used a conditional logit model to estimate mode choice for three classes of income groups (high, middle, and low) in response to congestion tolls applied to highly congested facilities. He estimated that to reduce delays for trips involving round-trip delays of 6, 12, and 30 min would require fees of $0.27, $0.98, and $2.20 (in 1972 prices), respectively, per automobile passenger round trip. When the revenues earned were applied to reducing the impact on the low-income group as opposed to being reinvested in highway capacity, all income groups were made better off. Gomez-Ibanez and Fauth (1980) estimated the effects of areawide licensing and parking fees on travel for the Boston area. Effects were derived from travel demand forecasts (logit model) using a range of elasticities for transit and automobile use. Their results suggested that areawide licensing based on the Singapore model would reduce trips into the center city the most and would also increase transit trips the most. A $1.00 areawide license was estimated to reduce trips to or through the central area by 44 percent. Recent estimates of congestion tolls were made for the Southern California Association of Governments (Urban Institute and KT Analytics 9 Many other studies have been conducted abroad to estimate the impact in specific cities.
OCR for page 134
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion TABLE B-2 Estimated Effects of Congestion Pricing Policies: Results of Selected U.S. Studies Author Type of Pricing Toll Estimated Effects Keeler and Small (1977) Optimal long-run toll $0.05 to $0.36/VMTa (avg of $0.15/VMT) Balance of supply and demand Kraus, Mohring, and Pinfold (1976) Optimal long-run toll $0.01 to $0.13/VMT in $1976 Annual welfare losses of at least $1 billion nationwide Small (1983) Optimal short run $0.27, $0.98, $2.20/passenger round trip (in $1983) Eliminate delays of 3, 12, and 30 min Gomez-Ibanez and Fauth (1980) Area licensing or parking $1.00/day (in $1980) Automobile trips to central Boston reduced 40 percent, transit trips up 28 percent, regional VMT reduced 3 to 4 percent Shoup and Willson (1992a,b) Cash option in lieu of parking subsidyb NA Commuter solo driving in Los Angeles area reduced 20 percent, commute VMT reduced 17 percent Urban Institute and KT Analytics (1991) All freeways in Los Angeles region $0.15/VMT VMT reduced 4 to 6 percent, average commute trip down 10 to 15 min; annual revenues, $2.5 to $2.7 billion Area pricing at 20 to 30 major activity centers $2.00/day Commute trips reduced 10 percent, trip times reduced 6 to 10 min, annual revenues, $280 million
OCR for page 135
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Kane and DeCorla-Souza (1992) All routes in a metropolitan area with 1 million commute trips Average of $0.13/VMT VMT reduced 2.5 percent; average time savings of 3.6 min/trip; annual revenues, $200 million; hydrocarbons reduced 15 percent Cameron (1991) All routes and employee parking $5.50/vehicle/day VMT reduced 6.5 percent; fuel consumption reduced 9.5 percent; VOCs and nitrogen oxides reduced 9 to 10 percent All routes only $0.15/VMT or $3.00/day VMT reduced 5 percent; fuel consumption reduced 9 percent; VOCs and nitrogen oxides reduced 8 percent Viton (1980) Optimal short run $0.15/auto-mile, $0.25/bus-mile, $0.33/truck-mile Optimizes traffic flow NOTE: 1 mi = 1.6 km. VMT = vehicle miles of travel. NA = not applicable. aAs updated by Small (1992). bNot really pricing but reduction in existing subsidy.
OCR for page 136
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion 1991). Probable effects of imposing facility pricing on 1280 km (800 mi) of congested urban freeways in the Los Angeles area suggest that tolls of $0.15/VMT would increase travel speeds by 10 to 20 percent, reduce annual vehicle travel by 4 to 6 percent, generate annual revenues of $2.5 to $2.7 billion, and cost about $80 million to $160 million to implement. Kane and DeCorla-Souza (1992) provided rough estimates of the effects on traffic speed and emissions of a congestion fee averaging $0.13/VMT applied to the principal arterials in a hypothetical urban area with 1 million daily commute trips during the morning and evening peak. They estimated that solo driving would decline by 30 percent, peak-hour vehicle miles of travel would decline by 10.5 percent (total VMT by 2.5 percent), and average commute trips would decline by 3.6 min. Total peak-hour hydrocarbon emissions would decline by 15 percent, resulting in a 3.6 percent decline in total emissions. The tolls would generate about $700,000 per day or about $175 million annually. Cameron (1991) estimated that a combination of pricing strategies for the Los Angeles area would have substantial effects on congestion and air quality. Charging a peak-hour fee of approximately $3.00 per day per vehicle and significantly increasing employee parking (by $2.25 per day on average) was estimated to reduce VMT by 6.5 percent, mobile source reactive organic gases (the precursors of smog) and nitrogen oxides by 9 to 10 percent, and automobile energy consumption by 9.5 percent. (These estimates are reductions in the travel, energy, and pollution levels that are forecast for the Los Angeles area for 2010 in the absence of any change in policy.) A congestion fee alone of $3.00, which would average about $0.15/VMT, would reduce VMT by 5 percent and reactive organic gases by 8 percent. Repetto et al. (1992) build on a model developed by Douglass Lee for the U.S. Department of Transportation's 1982 Final Report on the Federal Highway Cost Allocation Study. The model estimates the effects of congestion fees on urban roads nationwide. Because some roads are uncongested, no fees would be assessed; on congested urban roads the fees would range from $0.10 to $0.36/VMT. Such fees would reduce peak-period congestion by 22 percent and result in net welfare gains of about $10 billion currently and $21 billion in 1999. Although Shoup and Wilson (1992a, 1992b) did not explicitly conduct a congestion pricing study, they estimated that employer-subsidized parking is roughly equivalent to about $3.00 per commuter per day, which is about what a $0.15/VMT congestion fee would equal if applied to an average round-trip commute of 20 mi. They applied a logit model to
OCR for page 137
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion estimate that subsidized parking encourages demand for both solo driving and parking by commuters that is about one-third more than what the demand would be without these subsidies. For the Los Angeles area, they estimated that offering employees the option of receiving a cash allowance in lieu of subsidized parking would reduce commuter solo driving 20 percent and would reduce their VMT by 17 percent. Because peak-period driving accounts for about 40 percent of total daily travel, this would imply an overall VMT reduction of about 7 percent. The results of the studies reviewed in this section are fairly consistent, even though the methods used and the locales studied vary. The estimates of potential benefits suggest that congestion fees that averaged roughly $0.15/VMT would reduce peak-period traffic by up to 30 percent in the best case, with reductions in overall travel, energy consumption, and emissions of roughly 5 percent. REFERENCES Altshuler, A. 1979. Congestion. In The Urban Transportation System: Politics and Policy Innovation, MIT Press, Cambridge, Chap. 9. Altshuler, A. 1990. Discussion of C. Winston, How Efficient Is Current Infrastructure Spending and Pricing? In Is There a Shortfall in Public Capital Investment? Proceedings of a Conference sponsored by the Federal Reserve Bank of Boston, Conference Series No. 34, pp. 206–213. Button, K. 1983. Road Pricing—An Outsider's View of American Experiences.Transport Reviews , Vol. 4, pp. 73–78. Button, K., and A. Pearman. 1985. Congestion Pricing—Theory and Practice. In Applied Transport Economics: A Practical Case Studies Approach, Gordon and Breach Science Publishers, New York and London, Chap. 3. Downs, A. 1962. The Law of Peak-Hour Expressway Congestion. Traffic Quarterly , Vol. 16, pp. 393–409. Cameron, M. 1991. Transportation Efficiency: Tackling Southern California's Air Pollution and Congestion. Environmental Defense Fund, New York City; Regional Institute of Southern California. Else, P.K. 1986. No Entry for Congestion Taxes? Transportation Research, Vol. 20A, No. 2, pp. 99–107. Giuliano, G., and M. Wachs. 1992. A Comparative Analysis of Regulatory and Market-Based Transportation Demand Management Strategies. In Papers Presented at the Congestion Pricing Symposium, June 10–12, 1992,U.S. Department of Transportation, pp. 6-1 to 6-15. Gomez-Ibanez, J., and K. Small. Forthcoming. NCHRP Synthesis of Highway Practice: Road Pricing for Congestion Management: A Survey of International Practice.TRB, National Research Council, Washington, D.C.
OCR for page 138
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Gomez-Ibanez, J. 1992. The Political Economy of Highway Tolls and Congestion Pricing. Transportation Quarterly, Vol. 46, No. 3, pp. 343–360. (For summary, see Presentation Summary, In Exploring the Role of Pricing as a Congestion Management Tool, Searching for Solutions: A Policy Discussion Series, No. 1, U.S. Department of Transportation, March.) Gomez-Ibanez, J., and Fauth, G. 1980. Downtown Auto Restraint Policies: The Costs and Benefits for Boston . Journal of Transportation Economics and Policy , Vol. 14, pp. 133–153. Hau, T. 1992a. Congestion Charging Mechanisms: An Evaluation of Current Practice . Policy Research Working Paper. WPS 1071. World Bank, Washington, D.C. Hau, T. 1992b. An Economic Analysis of Road Pricing: A Diagrammatic Approach. Policy Research Working Paper. WPS 1070. World Bank, Washington, D.C. Higgins, T. 1986. Road Pricing Attempts in the United States. Transportation Research, Vol. 20A, No. 2, pp. 145–150. Kane, A., and P. DeCorla-Souza. 1992. Regionwide Toll Pricing: Impacts on Urban Mobility, Environment, and Transportation Financing. In Papers Presented at the Congestion Pricing Symposium, June 10–12, 1992, U.S. Department of Transportation, pp. 5-1 to 5-11. Keeler, T., and K. Small. 1977. Optimal Peak-Load Pricing, Investment, and Service Levels on Urban Expressways.Journal of Political Economy, Vol. 85, No. 1, pp. 1–25. Kraus, M., H. Mohring, and T. Pinfold. 1976. The Welfare Costs of Nonoptimum Pricing and Investment Policies for Freeway Transportation. American Economic Review , Vol. 66, No. 4, pp. 532–547. Knight, F. 1924. Some Fallacies in the Interpretation of Social Cost. Quarterly Journal of Economics, pp. 582–606. Meyer, J., J. Kain, and M. Wohl. 1965. Pricing, Subsidies, Market Structure, and Regulatory Institutions . In The Urban Transportation Problem , Harvard University Press, Cambridge, Mass., Chap. 13. Meyer, J., and J. Gomez-Ibanez. 1981. Traffic Congestion. In Autos, Transit, and Cities, Harvard University Press, Cambridge, Mass., Chap. 11. Mohring, H., and D. Anderson. 1994. Congestion Pricing for the Twin Cities Areas. Metropolitan Council of the Twin Cities Metropolitan Area. Mohring, H., and M. Harwitz. 1962. Highway Benefits: An Analytical Framework. Northwestern University Press, Evanston, Ill. Morrison, S.A. 1986. A Survey of Road Pricing. Transportation Research, Vol. 20A, No. 2, pp. 87–97. Ng, Y. 1977. Towards a Theory of Third-Best. Public Finance, Vol 32, No. 1, pp. 1–15. Orski, K. 1992. Congestion Pricing: Promise and Limitations. Transportation Quarterly, Vol. 46, No. 2, April, pp. 157–167. Pigou, A. 1920. The Economics of Welfare. 1st ed. Repetto, R., et al. 1992. Green Fees: How a Tax Shift Can Work for the Environment and the Economy. World Resources Institute, Washington, D.C. Richards, M. 1992. Road Pricing: International Experience. In Papers Presented at the Congestion Pricing Symposium, June 10–12, 1992, U.S. Department of Transportation, pp. 2-1 to 2-10.
OCR for page 139
CURBING GRIDLOCK: Peak-Period Fees To Relieve Traffic Congestion Shoup, D., and R. Willson. 1992a. Commuting, Congestion, and Pollution: The Employer-Paid Parking Connection . In Papers Presented at the Congestion Pricing Symposium, June 10–12, 1992. U.S. Department of Transportation, pp. 7-1 to 7-21. Shoup, D., and R. Willson. 1992b. Employer-Paid Parking: The Problem and Proposed Solutions. Transportation Quarterly, Vol. 46, No. 2, April, pp. 169–192. Small, K. 1983. The Incidence of Congestion Tolls on Urban Highways. Journal of Urban Economics , Vol. 13, pp. 90–111. Small, K. 1992. Using the Revenues from Congestion Pricing.Transportation, Vol. 19, No. 4, pp. 359–382. Toh, R. 1977. Road Congestion Pricing: the Singapore Experience. Malayan Economic Review, Vol. 22, pp. 52–61. Transportation Research Board. 1985. Special Report 209: Highway Capacity Manual. National Research Council, Washington, D.C. Urban Institute and KT Analytics. 1991. Final Report: Congestion Pricing Study. Southern California Association of Governments, Los Angeles, April. Vickrey, W. 1959. Statement on the Pricing of Urban Street Use. Joint Committee on Washington Metropolitan Problems, U.S. Congress . Hearings on the Transportation Plan for the National Capital Region . Nov. Viton, R. 1980. Equilibrium Short-Run Marginal-Cost Pricing of a Transport Facility . Journal of Transport Economics and Policy, Vol. 14,pp. 185–203. Walters, A. 1961. The Theory and Measurement of Private and Social Cost of Highway Congestion. Econometrica, Vol. 29, No. 4, pp. 676–699. Watson, P., and E. Holland. 1978.Relieving Traffic Congestion: The Singapore Area License Scheme. Working Paper 281. World Bank, Washington, D.C., June. 286 pp. Wilson, P. 1988. Welfare Effects of Congestion Pricing in Singapore. Transportation, Vol. 15, No. 3, pp. 191–210. Wohl, M., and C. Hendrickson. 1984. Some Practical Pricing Problems. In Transportation Investment and Pricing Principles, John Wiley and Sons, New York, Chap. 13. Zettel, R., and R. Carll. 1964. The Basic Theory of Efficiency Tolls: The Tolled, the Tolled Off, and the Un-Tolled. In Highway Research Record 47, HRB, National Research Council, Washington, D.C.
Representative terms from entire chapter: