New Studies in Chicago, LA and San Francisco [With LEM Analysis Results
Chair, Sierra Club Transportation Committee
For Presentation at the Air & Waste Management Associations 90th Annual Meeting
& Exhibition, June 8-13, 1997, Toronto, Ontario, Canada
Motor vehicles are the single largest source of air pollution and a major source of
water pollution. In the San Francisco Bay Area, for instance, on-road vehicles emit 40% or
the reactive organic gases, 38% of the NOx and 69% of CO.(1) In the U.S., tailpipe
emission standards have not been tightened for years, with no regulations to tighten them
in sight. After heavy lobbying by auto manufacturers, California relaxed its rule that
required new car sales to include a few percent of Zero Emission Vehicles. Its hard
to be optimistic that emissions standards will improve much in the near term.
After leaving the assembly line auto emission systems deteriorate and must be
maintained to keep emissions low. Yet legislators shy away from requiring more effective
inspection and maintenance programs, such as centralized inspection, loaded-mode tests,
higher mandatory repair limits and diesel standards, in all state non-attainment areas.
Talk radio even raised a firestorm in San Francisco against cleaner-burning reformulated
gasoline. Nor are evaporative emissions or entrained dust even addressed by tailpipe
controls. There appear to be real political limits to more effectively cleaning up the
Yet, auto ownership and vehicle miles traveled (VMT) continue to grow. U.S. VMT grew at
rates well over 3 percent per year during the 1980s, and is forecast to increase 25
percent per capita between 1990 and 2010.(2,3) Hidden subsidies to motor vehicle use are
estimated to range from $3 to $7 per gallon of gas, yet discussion of charging drivers the
full cost of driving by raising gas taxes, converting free to toll roads, instituting smog
fees or eliminating free parking runs into widespread opposition from corporate
interests.(4-9) Present auto subsidies include road construction and maintenance not
covered by gas taxes; police, fire and ambulance services to motorists; taxes lost on land
cleared for freeways; free parking; pollution; noise and vibration damage to structures;
global warming; protecting the petroleum supply line; production subsidies; sprawl and
loss of transportation options; uncompensated auto accidents; and congestion.
Even mild measures aimed at reducing single occupant vehicle commuting meet formidable
resistance. Lobbyists stripped from federal law the employee commute options mandate that
required employers to design programs that would modestly reduce single occupant vehicle
commutes in non-attainment areas. The California legislature even denied air districts the
right to implement such trip reduction ordinances. Few employers offer non-driving
commuters cash payments equal to the free parking they give drivers (parking cash-out).
Can Urban Form Be Harnessed to Reduce Driving?
Given the problems confronting reductions in tailpipe emissions, even as mileage and
congestion continue to increase, what alternative approaches are promising? Recently, many
urban planners and architects have suggested that returning to traditional city and
neighborhood design could reduce auto ownership, VMT and cold starts. Traditional cities
grew up around pedestrians, and residents could conveniently walk, bicycle or take transit
to most jobs. These neighborhoods had corner markets and other nearby commerce. They were
so complete that many residents seldom left them. These areas are now the central and
adjacent areas of our older cities and some of the older suburbs. Some are abandoned.
Could restoring these areas benefit air quality?
The hypothesis to be tested is that increased convenience, pedestrian amenities and
transit service in our neighborhoods would decrease driving and cold starts by increasing
accessibility. Increasing density and allowing mom and pop groceries, restaurants and
other neighborhood businesses in residential areas, and locating these areas close to
employment centers, increases accessibility for residents.
But would such design effectively reduce driving in modern America, where unfettered
motoring down an empty country lane is our transportation standard...or so new-car
commercials would have us believe. The issue is controversial, with homebuilding-industry
sponsored studies purporting to show that the suburbanization of jobs results in
suburbanites having shorter commutes, and presumably driving less than residents of
traditional neighborhoods in central cities.(10,11)
To address this issue we first identify the characteristics of traditional
neighborhoods. These areas have relatively high densities: ranging from row houses on
narrow lots through 3 to 4 story walk-ups to high-rise elevator apartments, condos and
co-ops--at 20 to 200 households/residential acre. They are proximate to major job and
shopping concentrations. They have a wealth of local shopping and services, and sidewalks
and other amenities to encourage walking. They have excellent access to transit. Various
mixes and matches of these characteristics have been identified and measured by the
pioneers in this research, including Bob Cervero, Michael Replogle and Larry Frank.(12-14)
Now that we have identified the characteristics of traditional neighborhoods we can
test the hypothesis that they lead to lower auto ownership and driving. The evidence to
test this hypothesis has been rather scant, often based on such inappropriate measures of
total driving as commute times, even though commutes are only 1/4 of total trips, and
walking commutes would show up as longer commute times, implying longer trips or more
traffic congestion. Rather than analyzing neighborhoods, some studies dichotomize central
cities versus suburbs, even though some suburbs are older and traditional, while some
central cities such as Houston or Phoenix dont qualify as traditional neighborhoods.
This practice reflects the difficulty of obtaining detailed data on the wealth of
neighborhoods that populate our urban areas. To get around this problem some studies have
compared neighborhoods that are selected to be representative of traditional/central city
or modern suburban areas.
Recent Research In California
My own 1991 and 1994 research falls prey to the last question--of representativeness.
In the first study, for the Natural Resources Defense Council (NRDC), I analyzed five
communities in the San Francisco region selected to cover the full range of neighborhoods,
from traditional to suburban.(15) They included 100 household/residential acre northeast
San Francisco (Nob, Russian and Telegraph Hills, Chinatown and North Beach), where I live;
the whole of San Francisco at medium-high density; medium density Rockridge district of
Oakland, with a BART station; the densifying suburb Walnut Creek, also with a BART
station; and a low density modern suburb San Ramon, at 3 households/residential acre. I
found that high residential density, nearby shopping, good transit and a good walking
environment go together. And, alternatively, sprawling suburbs isolate stores into
shopping centers, have poor transit service, and often don't even have sidewalks. The
co-variance of all these variables increases the difficulty of disentangling their
effects, but does allow density to serve as a surrogate for the group of variables to some
I obtained automobile ownership from the 1990 U.S. census. Accurately measuring VMT is
usually a major shortcoming of travel studies. Where would you get accurate measurements
of thousands of peoples driving? We used odometer readings recorded when auto owners
took their cars in for smog-checks. I found that residents of higher density communities
drive less. Comparing the extremes, the Nob Hill area was found to have 32 times higher
household density, and 200 times higher local shopping than suburban San Ramon, while only
about 1/4 the household auto ownership and VMT.
The 1994 study included additional San Francisco neighborhoods and Los Angeles, San
Diego and Sacramento neighborhoods.(16) These 27 neighborhoods were identified with the
help of the Association of Bay Area Governments, the Southern California Association of
Governments (SCAG), the San Diego Association of Governments and the Sacramento City
Planning Department. These neighborhoods were selected to represent the communities in the
regions based upon the judgement of regional planners.
Both studies found that the average resident in a neighborhood will drive 20 to 30
percent less per capita or per household than the residents of another neighborhood half
as dense. In other words, if you live in a neighborhood twice as dense as your
sisters neighborhood, on average you and your neighbors will drive 20 to 30% less
than she and her neighbors, whether you both live in low density areas, middle or high.
While this seems like a modest difference, it can really add up. The Nob Hill area is
32 times denser than San Ramon, or 5 doublings, giving it only 32% as much VMT if driving
drops 20% as density doubles, or 17% as much at 30% reduction. The 0-5
household/residential acre densities of most post-WWII suburbs could be doubled 7 times to
equal many Manhattan neighborhoods 400 households/residential acre. At 30% reduction
in driving as density doubled, the Manhattan families would drive only 8% as much as the
suburbanites. Thats a huge decrease in air pollution. This general pattern has been
found in New York, Chicago, Toronto, and across U.S. and British cities, however using
data for whole boroughs, cities, or urbanized areas, or using transportation modeling to
estimate VMT, rather than direct measurements.
The second study gave similar results, and covered sufficient neighborhoods to get
statistical significance. The strongest reductions in driving were found to be due to
residential density, followed by the quality of transit service. Doubling residential
density was shown to lower auto ownership 16%. Doubling residential density was shown to
lower VMT 16%, while doubling public transit service reduced VMT an additional 5%. When
density was considered as the surrogate for all the variables, doubling residential
density resulted in a 20% reduction in VMT. The other variables, local shopping and
pedestrian friendliness, acted in the predicted direction but with less force.
The present study, again sponsored by NRDC, along with the Center for Neighborhood
Technology (CNT) in Chicago, and the Surface Transportation Policy Project in Washington,
DC, expands the previous studies to include every neighborhood in the San Francisco, LA
and Chicago areas. The measures of traditional neighborhood characteristics are
residential density, center proximity, local shopping, transit service and
pedestrian/bicycle friendliness. Auto ownership is again obtained from the census and VMT
from odometer readings taken during smog checks.
The study has produced preliminary results using density as a surrogate for the cluster
of co-variant variables, and crude measures of transit service in the Chicago area. These
results use household travel surveys to measure auto ownership, VMT and trips by private
vehicle. The surveys were compiled by the regional transportation planning agencies--the
Chicago Area Transportation Study and the Metropolitan Transportation Commission--from
1-day travel diaries. In a household travel survey a randomized sample of households is
contacted and asked to participate. A fraction of those contacted agree to participate,
and a fraction of these actually complete and return the diaries in usable form. Commonly
around half the households initially contacted end up participating successfully. If there
is a response bias, it is probably weighted toward the better educated households. The
resulting sample of 10 to 20 thousand households represents 1% of the regions total
households. In the San Francisco area there are fewer households in the few high density
zones than would be desirable for some multi-variate analyses. Yet these surveys are
adequate to strongly support the earlier analyses.
The non-linear regression analysis for Chicago, again using density as a surrogate for
the cluster of neighborhood characteristics, gives a 9% reduction in auto ownership and
14% reduction in VMT every time density doubles. Regressing density and transit service
together gives a 15% reduction in VMT every time density doubles, with a further 7%
reduction in VMT every time transit service doubles.
Average daily driving and vehicle trips were calculated for households in the San
Francisco household travel survey, at 0-2 households/residential acre, 2-5, 5-10, 10-50
and above 50. The results show a 21% reduction in auto ownership, 30% reduction in VMT and
23% reduction in vehicular trips every time density doubles, Figures 2 and 3. At a 30%
drop in VMT as density doubles, the Nob Hill areas 32 times higher density predicts
that it would have 17% of San Ramons VMT/household. Vehicle trips per household,
while not a precise measure of cold starts, are our best approximation. And at a 23%
reduction in vehicle trips as density doubles, the Nob Hill neighborhood produces only 27%
as many cold starts per household as San Ramon. These results indicate that an urban
development strategy of building traditional neighborhoods could substantially reduce
driving and vehicular emissions.
Some very preliminary analyses of SCAGs travel analysis zones in Los Angeles use
census data and a preliminary estimate of VMT based upon smog-check odometer readings.
They show an 18% reduction in both VMT and vehicle ownership as density doubles.
But could income effects be masquerading as density effects? Studies have shown that
wealthy households own more cars and drive more. Could it be that the wealthy have all
moved to the suburbs? Not that anyone who had priced million dollar Nob Hill condos would
believe. Lets again visit MTCs household travel survey. It confirms that the
wealthy drive more, Figure 4, but shows that they drive 29% less as density doubles,
compared to 37% less for the middle class and 35% less for the poor. The impact of density
is strongly confirmed. The tragedy is that even the poor living at low density have to
invest so heavily in auto use.
Applying These Predictions To Existing Built-up Regions
Ok, the skeptic says, but our suburbs are already built and we are not going to tear
them down to rebuild in traditional neighborhoods! Thats true, at least in the short
term. However 50 to 75 year-old suburban housing is often a candidate for demolition and
replacement. So there is plenty of potential in the long term. But consider only the short
term--what benefits a policy of spurring traditional development in a built-up region
might yield in a mere 15 years.
As the Metropolitan Transportation Commission (MTC) was developing its 20-year Regional
Transportation Plan (RTP) for the San Francisco Bay Area in 1993, it soon became obvious
that no non-highway alternatives were to be considered. So the Regional Alliance For
Transit (RAFT) -- a transit and environmental coalition -- proposed to define a
pedestrian- and transit-oriented alternative. MTC analyzed it in their modeling system.
RAFT sought to increase accessibility to transit, especially by foot and bicycle. So
RAFT assumed the same total regional population and job growth as MTC, but clustered the
projected growth of households and employment after 1995 around transit stations at higher
densities. By developing in areas with existing infrastructure -- schools, public
facilities, streets and utilities, the region could save itself up to $25 billion on
construction of these facilities. And save some 200 square miles of forests, grasslands
and farmlands that MTC's alternative would sacrifice to development of residences,
commerce, industry and local streets.
RAFT designed transportation improvements which eliminated nearly all of MTC's 500 new
highway lane-miles, and put the savings into cost-effective public transit. It stressed
rail improvements in the urban corridors, including extending the CalTrain, which now runs
from south of San Jose to San Francisco, a mile and a half to San Francisco's downtown
Transbay Terminal and electrifying the system, new light rail lines in San Francisco and
Santa Clara counties, new heavy rail on existing tracks to link Santa Clara light rail to
East Bay BART and on to Livermore, Sacramento and San Joaquin county, and new commuter
rail in Marin and Sonoma counties. It also included new electric trolley-buses in the East
Bay urban corridor between Hayward, Oakland and Richmond, and express buses on I-80 and
I-680. Having improved CalTrain service to San Francisco Airport, it eliminated the
parallel, more expensive, BART extension. It converted some freeway lanes to bus/carpool
Further, in order to slightly reduce the $3 to $7 per gallon of gas subsidies to
driving, it assumed parking cash-out, whereby non-driving employees receive the cash value
of their unused "free" parking. California law allows county congestion
management agencies (CMA) to mandate parking cash-out. So the RAFT alternative included
this one market-based measure along with the increases in density and transit service.
MTC input this alternative into its modeling system, see Figure 1. They predicted that
by 2010 RAFT's alternative would reduce regional VMT 6% below MTC's alternative, saving
the average family 1,148 miles of travel annually. This is worth $379 annually to the
average family, using FHWA estimates of auto costs.(17, 18) RAFT would save 350,000
gallons of fuel daily and cut mobile source particulate emissions 10%, carbon monoxide 4%,
reactive organic gases 5% and nitrogen oxides 5%. Further, RAFT would cut congestion by
13%. These results come after only 15 years of changing to traditional development
patterns. Consider the magnitude of the impact after 50 years of these patterns!
Further, RAFT would boost transit passengers regionwide by 24% over MTC's alternative,
including Muni Metro 29%, CalTrain 167%, Santa Clara light rail 76%, East Bay heavy rail
(Amtrak) 152%, and AC Transit 39%. It would boost BART patronage 15% higher than MTC's
plan even without a BART extension from Colma to the San Francisco Airport.
MTCs modeling system is conservative in estimating the impacts of density and
increased transit service. It does not model the impacts of increasing pedestrian and
bicycle friendliness in neighborhoods. Yet it shows substantial reductions in driving and
pollutant emissions from increased density and improved transit services. Strongly applied
nationwide, could such policies not only reduce, but reverse the growth of VMT/capita?
What Can We Do? Be Practical
Thats fine, you might say, but what can we as air quality officials, industrial
leaders and ordinary citizens do? Consider the following list of practical actions,
including many suggestions from Urban Ecology.(19, 20)
1. Lobby congress to reauthorize an Intermodal Surface Transportation Efficiency Act
(ISTEA) that builds pedestrian, bicycle and transit facilities rather than highways,
mandates analyses like the RAFT alternative, and requires strong informed public
participation in transportation planning.
2. Strongly link the Clean Air Plans with land use and transportation.
3. Include an air quality element in City Comprehensive Plans.
4. Legalize second, or "in-law," units in single family areas in order to put
basements, garages or unused rooms to use.
5. Revise zoning laws to eliminate side and front yard setbacks, so housing can be
built to the sidewalk and the land used more efficiently.
6. Revise zoning laws to allow markets, restaurants, video rental stores and other
neighborhood businesses to locate along major roads in residential areas.
7. Revise zoning laws to replace off-street parking minimum requirements with maximum
parking spaces allowed, and severely restrict the ground coverage of parking facilities.
8. Site new development near transit, stores and services.
9. Publicly support infill housing developers.
10. Bring back downtown living. Encourage apartment and condo development downtown.
11. Build multi-family housing in the parking lots near the shops in shopping centers.
12. Redevelop "brownfields", or abandoned industrial land.
13. Encourage labor-intensive industrial districts.
14. Halt sprawl development on the periphery by establishing urban growth boundaries.
15. Protect and increase transit service rather than building roadways.
16. Enhance bikeways and footpaths to encourage non-automobile travel.
17. Make streets livable by adding stop signs and slowing traffic, and adding trees,
plants, lights and signage.
18. Promote interesting diversity by including a variety of housing types, unit sizes,
rents and prices.
19. Work with existing residents and neighborhood groups to meet their needs and learn
Sprawl has resulted in increased auto ownership, driving and cold starts. Infill of
traditional neighborhood developments into already built-up areas could reduce auto
ownership, vehicle miles traveled and cold starts, reducing auto emissions. Citizens and
public officials can take specific, incremental actions to promote this type of
1. Proposed Final Bay Area 94 Clean Air Plan; Bay Area Air Quality Management
District: San Francisco, Nov. 1994, Table 1.
2. Saving Energy in U.S. Transportation; Office of Technology Assessment, U.S.
Congress, July 1994, OTA-ETI-589, p9.
3. Energy and Transportation, Task Force Report; The Presidents Council on
Sustainable Development, 1996, U.S. G.P.O.: 1996-404-680:20028, p35.
4. Ketcham, B.; Komanoff , C. Win-Win Transportation: A No-Losers Approach To Financing
Transport in New York City and the Region; Komanoff Energy Associates: New York City, 9
5. MacKenzie J.; Dower R.; Chen D. The Going Rate: What It Really Costs To Drive; World
Resources Institute: Washington DC, June 1992.
6. Moffet J.; Miller P. The Price of Mobility; Natural Resources Defense Council: San
Francisco, 6 November 1991.
7. Vorhees M. The True Costs of the Automobile to Society; Boulder CO, 4 January 1992.
8. Saving Energy in U.S. Transportation; Office of Technology Assessment, U.S.
Congress, July 1994, OTA-ETI-589.
9. Delucchi, M. In Access; University of California at Berkeley, Spring 1996, pp7-13.
10. Gordon, P.; Richardson, H.W. The Case for Suburban Development; Building Industry
Association of Northern California and Home Ownership Advancement Foundation, March 1996.
11. Hayward, S.F. Preserving the American Dream; Pacific Research Institute for Public
Policy: San Francisco,
12. Cervero, R.; Kockelman, K. Travel Demand and the Three Ds: Density, Diversity and
Design; University of California at Berkeley, July 1996.
13. Frank, L.; Pivo, G. Relationships Between Land Use and Travel Behavior in the Puget
Sound Region; Washington State Department of Transportation: Olympia, Sep. 1994.
14. Replogle, M. Montgomery County Comprehensive Growth Policy Study, Volume 2;
Montgomery County Planning Department: Maryland, July 1989.
15. Holtzclaw, J. Explaining Urban Density and Transit Impacts on Auto Use; Natural
Resources Defense Council: San Francisco, Jan. 1991. In California Energy Commission
Docket No. 89-CR-90.
16. Holtzclaw, J. Using Residential Patterns and Transit to Decrease Auto Dependence
and Costs; Natural Resources Defense Council: San Francisco, June 1994.
17. Dahms, L. Memorandum to Work Program Committee: Regional Alliance for Transit
Proposal for RTP Track 1; Metropolitan Transportation Commission: Oakland CA, 13 May 1994.
18. Cost of Owning & Operating Automobiles, Vans & Light Trucks: 1991; Federal
Highway Administration, U.S. Department of Transportation: Washington D.C., Table 4.
19. Urban Ecologist; Urban Ecology: Oakland CA, 1996 Number 1, p5.
20. Urban Ecologist; Urban Ecology: Oakland CA, 1996 Number 4, p7.
Figure 1. Advantages of the RAFT Regional Transportation Plan over MTCs Plan.
- Vehicle Miles of Travel 6% less
- Annual automobile operating cost $379 saving per household
- Vehicle hours of travel (congestion) 13% less
- Fuel consumption 9% savings
- Carbon monoxide 6,900 tons annual reductions
- Reactive organic gases 660 tons annual reductions
- Nitrogen oxide 1020 tons annual reductions
- Particulates (PM10) 1200 tons annual reductions
- Regional population and job growth No difference
- Rural and natural land urbanized 200 sq. miles saved
- Infrastructure costs for new development Up to $25 billion saved
12/97 Supplement: Summary of LEM Analysis
The statistical analysis of the 316 Dram-Empal zones in Chicago, 1700 Travel Analysis
Zones in Los Angeles and 1099 Travel Analysis Zones in San Francisco showed a strong
response of household auto ownership and VMT to density, as shown in the following
figures, giving R2 (amount of variance explained) of 85% and 86%, respectively, in
Chicago; 56% and 63% in Los Angeles; and 63% and 63% in San Francisco.
Further, the detailed statistical analysis gave similar equations in all three
metropolitan areas. In Chicago, Los Angeles and San Francisco, 96%, 79% and 90%,
respectively, of the variance in vehicle ownership is explained by residential density,
per capita income, household size and transit service. Similarly, 93%, 80% and 87% of the
variance in annual household VMT is explained by residential density, per capita income,
household size, transit service and pedestrian/bicycle friendliness.
Up to Top | Printer-friendly version of this page