Given these facts, and reflecting the long-established interdependence between transportation and land use, some transportation planners have argued that the most effective way to stimulate transit is to reorganize land uses, moving them towards higher densities. Densification in itself does not reduce air pollution (in fact, the opposite; O'Toole, 1999) unless it changes modal choice and/or travel behavior. Downs (1992, p. 84) refers to a simulation where a tripling of suburban/exurban densities was required to reduce suburban/exurban commuting distances by 5 percent. The reason is that current settlement patterns, i.e. the dispersion of both workplaces and residences, allow for a high degree of accessibility even in low-density environments because of the wide choice of locations offered to both firms and households. Newman and Kenworthy (1989, 1990) were the pioneers of the "increasing density will reduce automobile use" (and hence air pollution) argument. Leaving aside their spurious statistical relationship between density (population per acre) and gasoline use per capita,3 even if there is a relationship between density and a non-per-capita variable (e.g. percentage of trips by automobiles), there is a difference between long-established high-density urban settlement patterns and incremental adjustments to an already decentralized low-density environment.4 First, America cannot become Europe. Most of the urban capital stock is in place and is so durable that changes are not easily made, except at the margin (especially the spatial margin). Second, even European countries are complaining about automobile dependence, and transit use has declined in the majority of European cities (OECD, 1995). Even if higher densities can be created at the local level (e.g. via building New Urbanist settlements), Crane (1996) has demonstrated that the results in terms of the impact on automobile travel are indeterminate, perhaps resulting in more, if shorter, trips. Automobile travel is more likely to increase the more elastic is the price-elasticity of trip demand, the stronger is the income elasticity of demand for auto travel, and the more positive the cross-price-elasticity of demand for car trips. The hoped for outcomes of the new land-use-design protagonists require more compact land uses to result in both a modal shift from cars to other modes (e.g. walking) and a reduction in new car trips (zero or low own-price elasticity). Even in the cases where total automobile travel is reduced because any increase in trip frequency is more than offset by shorter trips, air pollution may nevertheless increase because more cold starts and hot soaks outweigh less trip-length-related emissions (Guensler and Sperling, 1993).
The other, more ambitious idea is to integrate both land use changes and transit investments in the same project, via the promotion of Transit-Oriented Developments (TODs). There are very few successful examples of this approach in the United States. TODs take a long time to implement, perhaps 15 years or more (Boarnet and Compin, 1999). As a generalization, they are much more likely to be effective on new greenfield station sites (e.g. the Pleasant Hill BART station in California; Bernick and Cervero, 1997) than at close-in, already developed locations where land assembly is difficult, redevelopment costs may be prohibitive and community opposition is both vocal and obstructionist (Deakin and Chang, 1992). The dilemma is that these problems are more likely to occur along high-density growth corridors where the prospects for attracting riders are highest. Even so, the residential developments around transit stations tend to attract existing transit users rather than former commuter drivers.
A more modest expression of the same idea is to simultaneously promote higher densities and improved "connectivity," presumably encouraging more non-auto trips. Microstudies have suggested that this strategy might reduce VMT, but there is no evidence that it cuts down on the number of automobile trips. Unless this happens, the impact on air quality will be negligible.