tion in fuel consumption in the Bay Area as a result of an average toll of $0.06/km ($0.10/mi) and a 9.2 percent reduction in the South Coast Air Basin as a result of an average toll of $0.09/km ($0.15/mi) ( Table 3-1 and Table 3-2).

Over the long term, changes in land use patterns could have substantial impacts on air quality and energy in addition to those already cited. Two different arguments are made about the possible effects of congestion pricing on the spatial development of metropolitan areas (Deakin, Vol. 2). One line of reasoning holds that the traditional underpricing of highways has encouraged urban sprawl and that correct pricing would encourage more dense development in and around urban centers. An opposing line of reasoning holds that congestion pricing would facilitate continued decentralization because (a) it would reduce the attractiveness of areas affected by pricing by creating a negative image (especially if competing commercial areas are unpriced) and (b) zoning regulation would prevent landholders from increasing the density of development.

The outcome of this debate has both practical and political implications. At the practical level it is important to anticipate the effects on development in order to be able to design a program that would minimize adverse consequences. At the level of political feasibility, it is important to anticipate the possible resistance of affected interests. When areawide pricing was proposed for the most congested parts of Manhattan in the mid-1980s, for example, retail and commercial establishments in the area reacted with alarm, apparently thinking that increased out-of-pocket costs for travel would reduce the attractiveness of the central business district.

Neither theory nor research on the relationship between the cost of transportation and urban development provides compelling evidence to support whether congestion pricing would have a centralizing or decentralizing effect (Deakin, Vol. 2). The possible impact of congestion pricing on urban form is complicated further by the intended use of the revenues and by steps that businesses might take to minimize the impact. For example, if the revenues are used to compensate adversely affected groups (which might include commercial establishments in an affected area), the land use impact might well be minimized. Alternatively, if the revenues were used to expand highway capacity at a major bottleneck serving the downtown area, it is possible that the improved access would benefit the