Land use and traffic congestion

              Land Use and
Traffic Congestion
Final Report 618
March 2012
Arizona Department of Transportation
Research Center
Land Use
and Traffic Congestion
Final Report 618
March 2012
Prepared by:
J. Richard Kuzmyak
Transportation Consultant, LLC
Silver Spring, Maryland
In Association with:
Caliper Corporation
Newton, Massachusetts
and
PolyTech Corporation
Surprise, Arizona
Prepared for:
Arizona Department of Transportation
in cooperation with
U.S. Department of Transportation
Federal Highway Administration
The contents of this report reflect the views of the authors who are responsible for the
facts and the accuracy of the data presented herein. The contents do not necessarily
reflect the official views or policies of the Arizona Department of Transportation or the
Federal Highway Administration. This report does not constitute a standard,
specification, or regulation. Trade or manufacturers’ names that may appear herein are
cited only because they are considered essential to the objectives of the report. The US
government and the State of Arizona do not endorse products or manufacturers.
Front cover photographs courtesy of Downtown Tempe Community and Valley Metro.
Technical Report Documentation Page
1. Report No.
FHWA-AZ-12-618
2. Government Accession No.
L
3. Recipient's Catalog No.
5. Report Date
MARCH 2012
4. Title and Subtitle
Land Use and Traffic Congestion
6. Performing Organization Code
7. Author
J. Richard Kuzmyak
8. Performing Organization Report No.
9. Performing Organization Name and Address 10. Work Unit No.
J. Richard Kuzmyak
Transportation Consultant, LLC
9509 Woodstock Ct.
Silver Spring, Maryland 20910
11. Contract or Grant No.
SPR-PL 1 (69) 618
13.Type of Report & Period Covered
FINAL REPORT
12. Sponsoring Agency Name and Address
Research Center
Arizona Department of Transportation
206 S. 17th Avenue MD075R
Phoenix, Arizona 85007 14. Sponsoring Agency Code
15. Supplementary Notes
Project performed in cooperation with the Federal Highway Administration.
16. Abstract
The study investigated the link between land use, travel behavior, and traffic congestion. Popular wisdom
suggests that higher-density development patterns may be beneficial in reducing private vehicle dependency
and use, which if true, could hold important implications for urban transportation planning and related goals
such as congestion relief, air quality, and sustainability. However, an important consideration is whether more
higher-density development also exacerbates traffic congestion on adjacent streets and roads simply
because of its concentration of activity. Researchers performed a detailed analysis of the relationships
between higher-density land use and traffic conditions in four Phoenix transportation corridors. The corridors
included three older, high-density, mixed-used urban areas and a more contemporary suburban area with
lower density but high traffic volumes. The analysis showed that the urban corridors had considerably less
congestion despite densities that were many times higher than the suburban corridor. The reasons were
traced to better mix of uses, particularly retail share, which led to shorter trips, more transit and nonmotorized
travel, and fewer vehicle miles of travel (VMT). Also recognized was the importance of a secondary street
grid in the three urban areas, which allows for better channeling of traffic and enables walking. Researchers
developed a set of regression models to quantify the effects of key land use variables on household vehicle
ownership and VMT, illustrating the mitigating effects of higher density, better mix, and better transit
accessibility. Researchers also performed an extensive review of literature on transportation and land use
interaction, and surveyed local officials to elicit information about familiarity with compact, mixed land use
concepts; perceptions of impact on travel and traffic; and desirability of greater proliferation in Arizona’s
metropolitan areas.
17. Key Words
Land use; traffic congestion; smart growth; mixed-use
development; integrated transportation and
land use planning; the 4Ds; traffic mitigation;
density; transit-oriented development; auto
dependency
18. Distribution Statement
This document is available to the
U.S. public through the National
Technical Information Service,
Springfield, Virginia, 22161.
19. Security Classification
Unclassified
20. Security Classification
Unclassified
21. No. of Pages
287
22. Price
23. Registrant's Seal
SI* (MODERN METRIC) CONVERSION FACTORS
APPROXIMATE CONVERSIONS TO SI UNITS
Symbol When You Know Multiply By To Find Symbol
LENGTH
in inches 25.4 millimeters mm
ft feet 0.305 meters m
yd yards 0.914 meters m
mi miles 1.61 kilometers km
AREA
in2 square inches 645.2 square millimeters mm2
ft2 square feet 0.093 square meters m2
yd2 square yard 0.836 square meters m2
ac acres 0.405 hectares ha
mi2 square miles 2.59 square kilometers km2
VOLUME
fl oz fluid ounces 29.57 milliliters mL
gal gallons 3.785 liters L
ft3 cubic feet 0.028 cubic meters m3
yd3 cubic yards 0.765 cubic meters m3
NOTE: volumes greater than 1000 L shall be shown in m3
MASS
oz ounces 28.35 grams g
lb pounds 0.454 kilograms kg
T short tons (2000 lb) 0.907 megagrams (or "metric ton") Mg (or "t")
TEMPERATURE (exact degrees)
oF Fahrenheit 5 (F-32)/9 Celsius oC
or (F-32)/1.8
ILLUMINATION
fc foot-candles 10.76 lux lx
fl foot-Lamberts 3.426 candela/m2 cd/m2
FORCE and PRESSURE or STRESS
lbf poundforce 4.45 newtons N
lbf/in2 poundforce per square inch 6.89 kilopascals kPa
APPROXIMATE CONVERSIONS FROM SI UNITS
Symbol When You Know Multiply By To Find Symbol
LENGTH
mm millimeters 0.039 inches in
m meters 3.28 feet ft
m meters 1.09 yards yd
km kilometers 0.621 miles mi
AREA
mm2 square millimeters 0.0016 square inches in2
m2 square meters 10.764 square feet ft2
m2 square meters 1.195 square yards yd2
ha hectares 2.47 acres ac
km2 square kilometers 0.386 square miles mi2
VOLUME
mL milliliters 0.034 fluid ounces fl oz
L liters 0.264 gallons gal
m3 cubic meters 35.314 cubic feet ft3
m3 cubic meters 1.307 cubic yards yd3
MASS
g grams 0.035 ounces oz
kg kilograms 2.202 pounds lb
Mg (or "t") megagrams (or "metric ton") 1.103 short tons (2000 lb) T
TEMPERATURE (exact degrees)
oC Celsius 1.8C+32 Fahrenheit oF
ILLUMINATION
lx lux 0.0929 foot-candles fc
cd/m2 candela/m2 0.2919 foot-Lamberts fl
FORCE and PRESSURE or STRESS
N newtons 0.225 poundforce lbf
kPa kilopascals 0.145 poundforce per square inch lbf/in2
*SI is the symbol for th International System of Units. Appropriate rounding should be made to e comply with Section 4 of ASTM E380.
(Revised March 2003)
TABLE OF CONTENTS
Executive Summary...........................................................................................................1
Chapter 1. Introduction.....................................................................................................7
Purpose of the Study........................................................................................................7
Research Approach and Report Organization .................................................................8
Chapter 2. Literature Review..................................................................................... 8
Chapter 3. Survey of Officials ................................................................................... 9
Chapter 4. Analysis of Corridors ............................................................................... 9
Chapter 5. Land Use Impacts on Design ................................................................. 10
Chapter 6. Summary, Conclusions, and Recommendations .................................... 10
Appendices............................................................................................................... 10
Chapter 2. Literature Review .........................................................................................11
Introduction ...................................................................................................................11
Brief Historical Perspective and Trends........................................................................12
Key Trends............................................................................................................... 13
How Does Land Use Impact Travel? ............................................................................17
Most Important Variables and Types of Travel Most Affected ....................................22
The Role of Transit........................................................................................................25
The Paradox of Self-Selection.......................................................................................27
Density, Congestion, Access, and Mobility ..................................................................30
Market Forces and Equity Issues...................................................................................33
Travel Model Capabilities .............................................................................................39
Conclusions ...................................................................................................................42
Chapter 3. Survey of Officials.........................................................................................47
Overview .......................................................................................................................47
Survey Design and Administration ...............................................................................47
Analysis of Survey Responses ......................................................................................50
Summary of Survey Findings........................................................................................53
Method of Involvement or Type of Influence Over Land Use and
Development Decisions ........................................................................................... 53
Importance of Key Factors When Making Development Decisions ....................... 57
Involvement in Transportation Impacts of Development Proposals or
Land Use Plans ........................................................................................................ 61
Experience with Mixed-Use Development Concepts .............................................. 75
Anticipated Effect of Mixed-Use Development on Travel...................................... 77
Community Acceptance of Mixed-Use Concepts.................................................... 81
Availability and Usefulness of Information on Mixed-Use Development .............. 82
Views on Appropriate Development Types............................................................. 84
Identification of Congestion Trouble Spots............................................................. 90
Chapter 4. Analysis of Corridors ...................................................................................93
Introduction ...................................................................................................................93
Approach .......................................................................................................................94
Characteristics of Study Areas ......................................................................................96
Scottsdale ................................................................................................................. 96
Bell Road ................................................................................................................. 97
Central Avenue ........................................................................................................ 97
Tempe ...................................................................................................................... 98
Existing Road Capacity and Traffic Conditions..........................................................108
Transportation Network Characteristics......................................................................111
Scottsdale ............................................................................................................... 112
Bell Road ............................................................................................................... 112
Central Avenue ...................................................................................................... 112
Tempe .................................................................................................................... 113
Traffic Conditions .......................................................................................................113
Scottsdale ............................................................................................................... 118
Bell Road ............................................................................................................... 127
Central Avenue ...................................................................................................... 127
Tempe .................................................................................................................... 128
Composition of Traffic on Key Facilities....................................................................129
Scottsdale ............................................................................................................... 130
Bell Road ............................................................................................................... 130
Central Avenue ...................................................................................................... 130
Tempe .................................................................................................................... 131
Transit Use ..................................................................................................................133
Directionality of Travel and Internal Trip Capture .....................................................135
Scottsdale ............................................................................................................... 136
Bell Road ............................................................................................................... 138
Central Avenue ...................................................................................................... 138
Tempe .................................................................................................................... 138
Average Trip Lengths..................................................................................................139
Walk and Bike Use......................................................................................................141
Summary .....................................................................................................................146
Land Use ................................................................................................................ 146
Road System and Traffic Conditions..................................................................... 147
Traffic Sources....................................................................................................... 148
Transit .................................................................................................................... 149
Internal Trip Capture.............................................................................................. 149
Average Trip Lengths ............................................................................................ 150
Walkability............................................................................................................. 150
Chapter 5. Land Use Impacts on Travel......................................................................151
Introduction .................................................................................................................151
Travel Database...........................................................................................................153
Land Use Database and Variable Creation..................................................................155
Density ................................................................................................................... 155
Diversity................................................................................................................. 156
Design .................................................................................................................... 157
Destinations............................................................................................................ 159
Data Analysis ..............................................................................................................160
Socioeconomic Characteristics of Jurisdictional Study Areas............................... 160
Land Use Characteristics.............................................................................................164
Density ................................................................................................................... 164
Diversity................................................................................................................. 168
Land Use Mix (Entropy)........................................................................................ 172
Design .................................................................................................................... 172
Regional Accessibility ........................................................................................... 177
Basic Travel Relationships..........................................................................................182
Average Trip Lengths ............................................................................................ 182
VMT Relationships......................................................................................................188
Clear Relevance of 4Ds Land Use Relationships........................................................191
Regression Analysis ....................................................................................................194
Model Validation ................................................................................................... 197
Sensitivity Analysis of Land Use and Travel Behavior Relationships.................. 199
Conclusions .................................................................................................................203
Chapter 6. Summary, Conclusions, and Recommendations......................................205
Overall Summary ........................................................................................................205
Literature Review................................................................................................... 207
Survey of Officials................................................................................................. 208
Travel Behavior Analysis Results.......................................................................... 209
Traffic Analysis Results......................................................................................... 213
Conclusions .................................................................................................................214
Recommendations .......................................................................................................216
References ...................................................................................................................221
Appendix A. Summary of Open-Ended Question Responses from
Survey of Officials ...................................................................................227
Appendix B. Congested Corridors as Identified by Survey of Officials ...................249
Appendix C. Selection of Study Corridors ..................................................................255
Purpose ........................................................................................................................255
Process for Selecting Corridors...................................................................................255
Phoenix........................................................................................................................258
Expressways........................................................................................................... 258
Arterials.................................................................................................................. 258
Tucson .........................................................................................................................260
Expressways........................................................................................................... 260
Arterials.................................................................................................................. 260
Appendix D. Traffic Counts from 2006/2007 MAG Regional
Traffic Volume Study..............................................................................264
LIST OF FIGURES
Figure 1. Population and VMT Trends ............................................................................. 14
Figure 2. Household Characteristics: 1969 through 2001 ................................................ 15
Figure 3. Change in Household Travel Characteristics: 1969 through 2001.................... 16
Figure 4. Source of Growth in Annual Household VMT ................................................. 17
Figure 5. Daily Per Capita VMT vs. Residential Density in Baltimore Region............... 20
Figure 6. Accessibility Benefits from Compact, Mixed Land Use................................... 33
Figure 7A. Ways in Which Elected Officials Become Involved in
Land Use Decisions ........................................................................................ 54
Figure 7B. Tools Available to Planning and Zoning Officials to Influence Type,
Scale, Timing, or Impact of Development Projects........................................ 55
Figure 7C. Role of Local Planners in Land Use and Development Decisions ................. 56
Figure 7D. Role of State and Regional Officials in Land Use and
Development Decisions ..................................................................................57
Figure 8. Frequency of Elected Officials Personally Reviewing Transportation
Impacts/Needs of a Development Proposal (Question 4) .................................. 1
Figure 9. Elected Officials Asked Planning Staff to Consider Transportation
Impacts (Question 5) .......................................................................................... 1
Figure 10. Coordination between Elected Officials and Other Organizations on
Development Plans or Transportation Impacts (Question 6).......................... 63
Figure 11. Extent and Manner in which Planning and Zoning Officials Consider
Transportation Impacts ................................................................................... 64
Figure 12. Level at which Planning and Zoning Officials Consider Transportation
Impacts............................................................................................................ 66
Figure 13. Planning and Zoning Officials’ Awareness of Transportation
Modeling Tools for Impact Assessment ........................................................... 1
Figure 14. Participation of Local Planners in Land Use/Development
Review Process ............................................................................................... 67
Figure 15. Geographic Scale at which Local Planners Reviewed Traffic
Impact Analysis ..............................................................................................69
Figure 16. Tools or Procedures Local Planners Used to Evaluate
Transportation Impacts ...................................................................................70
Figure 17. Development Traffic Impact Data Obtained from Local Planners’
Agency or Jurisdiction .................................................................................... 71
Figure 18. Participation of State and Regional Officials in Local Land
Use Decisions.................................................................................................. 73
Figure 19. State and Regional Officials’ Organizations Obtained Traffic Impact Data..... 1
Figure 20. State and Regional Officials’ Organizations Coordinate with
Other Agencies about Land Use ..................................................................... 74
Figure 21. Extent of Direct Experience with Mixed-Use Projects ................................... 75
Figure 22. Officials Who had Received Applications for Mixed-Use Projects................ 76
Figure 23. Officials Who Encouraged Applications for Mixed-Use Projects .................. 77
Figure 24. Expected Effect of Mixed-Use Development on Traffic Congestion ............. 78
Figure 25. Expected Effect on Transit Use....................................................................... 79
Figure 26. Expected Effect on Pedestrian and Bicycle Travel ......................................... 80
Figure 27. Community Support for Mixed-Use Development ......................................... 81
Figure 28. Sufficient Information to Make Informed Judgments about
Traffic Impacts................................................................................................ 82
Figure 29. Value of Additional Information on Impacts .................................................. 83
Figure 30. Perceived Value of Additional Information .................................................... 84
Figure 31. Most Appropriate Future Development Type for My Jurisdiction ................. 85
Figure 32. Most Appropriate Future Development Types for Region.............................. 87
Figure 33. Development that Should Happen in Region vs. What is Likely to Happen .. 89
Figure 34. Scottsdale Study Area...................................................................................... 96
Figure 35. West Bell Road Study Area............................................................................. 97
Figure 36. North Central Avenue Study Area .................................................................. 98
Figure 37. Tempe Study Area........................................................................................... 99
Figure 38. Scottsdale Road Study Area Development Characteristics........................... 104
Figure 39. Bell Road Study Area Development Characteristics..................................... 105
Figure 40. Central Avenue Study Area Development Characteristics ........................... 106
Figure 41. Tempe Study Area Development Characteristics.......................................... 107
Figure 42. Scottsdale Study Area: Major Roadways and Number of Lanes .................. 109
Figure 43. Bell Road Study Area: Major Roadways and Number of Lanes................... 109
Figure 44. Central Avenue Study Area: Major Roadways and Number of Lanes ......... 110
Figure 45. Tempe Study Area: Major Roadways and Number of Lanes........................ 110
Figure 46A. 2008 Midday V/C Ratios on Scottsdale Road............................................ 119
Figure 46B. 2008 Midday Vehicular Volumes on Scottsdale Road............................... 119
Figure 47A. 2008 Midday V/C Ratios on Bell Road...................................................... 120
Figure 47B. 2008 Midday Vehicular Volumes on Bell Road......................................... 120
Figure 48A. 2008 Midday V/C Ratios on Central Avenue............................................. 121
Figure 48B. 2008 Midday Vehicular Volumes on Central Avenue............................... 121
Figure 49A. 2008 Midday V/C Ratios inTempe............................................................. 122
Figure 49B. Midday Vehicular Volumes in Tempe........................................................ 122
Figure 50A. 2008 PM Peak Period V/C Ratios on Scottsdale Road .............................. 123
Figure 50B. 2008 PM Peak Period Vehicular Volumes on Scottsdale Road ................. 123
Figure 51A. 2008 PM Peak Period V/C Ratios on Bell Road ........................................ 124
Figure 51B. 2008 PM Peak Period Vehicular Volumes on Bell Road ........................... 124
Figure 52A. 2008 PM Peak Period V/C Ratios on Central Avenue ............................... 125
Figure 52B. 2008 PM Peak Period Vehicular Volumes on Central Avenue .................. 125
Figure 53A. 2008 PM Peak Period V/C Ratios in Tempe .............................................. 126
Figure 53B. PM Peak Period Vehicular Volumes in Tempe .......................................... 126
Figure 54. Comparison of Study Area Characteristics ................................................... 162
Figure 55. Study Area Net Residential Density.............................................................. 166
Figure 56. Residential Density—Percent by Group ....................................................... 169
Figure 57. Distribution of Land Uses by Area................................................................ 170
Figure 58. Distribution of Land Uses Net of Open Space .............................................. 171
Figure 59. Entropy: Index of Land Use Mix and Balance.............................................. 173
Figure 60. Entropy Index: Percent by Group.................................................................. 174
Figure 61. Walk Opportunities ....................................................................................... 175
Figure 62. Walk Opportunities—Percent by Group ....................................................... 176
Figure 63. Regional Transit Accessibility ...................................................................... 178
Figure 64. Regional Transit vs. Auto Accessibility........................................................ 180
Figure 65. Ratio of Job Accessibility by Transit vs. Auto.............................................. 181
Figure 66. Average Trip Length (in miles)—Home-Based Work Trips ........................ 183
Figure 67. Average Trip Length (in miles)—Home-Based Shopping Trips .................. 184
Figure 68. Average Trip Length (in miles)—Home-Based Other Trips ........................ 185
Figure 69. Average Trip Length (in miles)—Nonhome-Based Trips............................. 186
Figure 70. Nonhome-Based Other and Home-Based Shopping Trip
Lengths in Relation to Entropy ..................................................................... 187
Figure 71. Nonhome-Based Other and Home-Based Shopping Trip
Lengths in Relation to Walk Opportunities .................................................. 187
Figure 72. Daily Per Capita VMT (Weekday)................................................................ 188
Figure 73. Total Daily VMT per Capita—Percent by Mileage Group ........................... 190
Figure 74. Daily HBW VMT per Capita—Percent by Mileage Group .......................... 190
Figure 75. Daily Nonwork VMT per Capita—Percent by Mileage Group .................... 191
Figure 76. Daily VMT per Capita vs. Net Residential Density ...................................... 192
Figure 77. Daily VMT per Capita vs. Land Use Mix (Entropy)..................................... 192
Figure 78. Daily per Capita VMT vs. Weighted Retail/Service
Opportunities within One-Quarter Mile........................................................ 193
Figure 79. Daily VMT per Capita vs. Regional Transit Accessibility to Jobs ............... 193
LIST OF TABLES
Table 1. Population and Transportation Statistics of 23 Largest Metropolitan Areas...... 28
Table 2. Effect of Higher Density on Vehicle Travel....................................................... 31
Table 3. Description of Survey Sample ............................................................................ 49
Table 4. Importance of Key Factors When Making Development Decisions .................. 58
Table 5. Factors that Respondents Ranked Important or Very Important ........................ 60
Table 6. Participation of Local Planners in Development Review or Traffic Impact
Evaluations, by Job Category ............................................................................ 68
Table 7. Appropriate Land Use for My Community ........................................................ 86
Table 8. Appropriate Land Use for Region ...................................................................... 88
Table 9. Time Periods of Most Severe Congestion .......................................................... 91
Table 10. Source of Congestion in Named Corridor ........................................................ 91
Table 11. State and Regional Officials—Source of Congestion in Named Corridor ....... 92
Table 12. Characteristics of the Four Study Areas ......................................................... 101
Table 13. Comparison of MAG Model Assigned Link Volumes with
Actual Traffic Counts..................................................................................... 115
Table 14A. 2008 V/C Ratios on Selected Links Based on MAG Model Forecasts ....... 117
Table 14B. V/C Ratios on Selected Links Adjusted to Reflect Actual Counts .............. 117
Table 15A. Composition of Travel on Selected Links—Mid-Day Period ..................... 132
Table 15B. Composition of Travel on Selected Links—PM Peak Period...................... 132
Table 16. Comparative Transit Mode Share by Study Area........................................... 133
Table 17. Trip Orientation and Internal Capture Rates by Trip Purpose........................ 137
Table 18. Average Trip Lengths by Trip Purpose (in miles).......................................... 140
Table 19A. Daily Household (HH) Trips by Purpose and Mode ................................... 143
Table 19B. Daily Household (HH) Trip Rates by Purpose and Mode ........................... 144
Table 19C. Percent of Daily Household (HH) Trips by Purpose (Walk or Bike) .......... 145
Table 20. Comparative Characteristics and Performance of the Four Study Areas ....... 147
Table 21. MAG Study Areas from 2001 Travel Survey................................................. 154
Table 22. List of Opportunity Value Weights by SIC Code........................................... 161
Table 23. Socioeconomic Characteristics of Study Areas (MAG Jurisdictions)............ 162
Table 24. Additional Socioeconomic Characteristics..................................................... 163
Table 25. Sociodemographic Categorization of Study Areas......................................... 164
Table 26. Comparison of Net vs. Gross Residential Density ......................................... 165
Table 27. Residential Density by Area Classification .................................................... 167
Table 28. Models of Household Vehicle Ownership and VMT for
Phoenix/MAG Region (2001 HTS)................................................................ 194
Table 29. Estimates of Point Elasticities............................................................................. 1
Table 30. Comparison of Model vs. Survey Estimates of Household
Vehicle Ownership and VMT ........................................................................ 198
Table 31. Examination of VMT Sensitivity to Land Use Variables
Using 4Ds Models .......................................................................................... 200
Table 32. Characteristics of Phoenix Corridors.............................................................. 262
Table 33. Characteristics of Tucson Corridors ............................................................... 263
1
EXECUTIVE SUMMARY
The purpose of this project was to analyze and interpret the relationship between higher-density
development and traffic congestion. Governments have expressed increased
interest in the possible benefits of compact, mixed land use—referred to in many circles
as smart growth—to reduce auto dependency and use. If true, this finding could be of
significance in planning solutions to a host of transportation system investment,
performance, and impact issues.
Before considering any type of formal policy position in relation to land use, the Arizona
Department of Transportation (ADOT) is expanding its understanding of the relationships
between land use development patterns and transportation. Among ADOT’s key
questions are:
 Does higher-density development reduce auto use, to what extent, and in response
to what factors?
 Does higher-density development also generate higher levels of traffic congestion
simply due to the higher concentration of activity?
 Do Arizonans know about smart growth, and what are their perceptions of its
impacts and desirability?
The research study that is summarized in this report was commissioned to address these
specific issues. It involved a national-scale review of research and evidence on
transportation and land use relationships; detailed local analysis of these relationships
using data from metropolitan Phoenix; and a survey of officials in Arizona’s metropolitan
areas about their perceptions of land use/transportation, how higher-density development
is viewed, and whether there would be receptiveness for compact, mixed-use approaches
regionally and in their own area.
The findings of this study confirm the benefits of better land use. In its assessment of a
prodigious volume of research on this topic, the project’s literature review was able to
highlight the following findings:
 Density and Vehicle Miles Traveled Using residential density as a primary
indicator of concentrated land use, a variety of studies have shown that
households in higher-density (i.e., more urban) settings tend to own fewer
vehicles, drive less, walk and take transit more often, and generate one-half to
one-third of the daily vehicle miles traveled (VMT) of their suburban
counterparts.
 Beyond Density: Research has found that the effects of land use on travel
behavior are rooted in factors beyond simple density. Also important are related
factors such as mix of uses, auto- vs. pedestrian-oriented design, and regional
accessibility enhanced by multiple travel choices (especially transit). These
characteristics of density, diversity, design, and destinations are commonly
referred to as the 4Ds.
 Travel Purpose: Work travel, which is associated with peak period congestion,
generally garners most of the attention in transportation planning and policy
deliberations. Indeed, where compact land use is focused around high-quality
2
regional transit—at both the origin and destination of a journey—commuters will
use transit in large numbers because of its convenience. However, the travel
market that may be most influenced by compact mixed-land use is nonwork
travel, which accounts for as much as 80 percent of routine household travel and
has been the fastest-growing segment since the 1980s. This relationship/trend can
be directly linked to land use, recognizing that in conventional suburban areas
almost all household needs—shopping, transporting children, personal business,
social, and recreation—require private vehicle travel. Areas where residents live
in older, mixed-use communities with nearby services and restaurants show a
much greater concentration of travel to local destinations—including walking,
biking, or short car trips—despite a daily commute that may well be a long
distance solo-driver trip.
 Market Forces: Critics of smart growth approaches to land use maintain that it is
a planner’s notion that does not reflect market realities. However, real estate
industry experts assert that the reason more compact, mixed-use development has
not occurred has to do with restrictive local zoning codes or traffic level of service
standards, and not because of market demand, which is gauged as twice as high as
current build rates. This is borne out by visual preference surveys that show a
general preference for older (pre-World War II) suburban development patterns,
which are more compact and walkable, and foster more social interaction between
residents.
To ascertain the validity of these research findings in the Arizona environment,
researchers performed a number of detailed studies using local data and both existing
planning tools as well as some new ones developed specially for analysis of the role of
land use. These analyses were focused on the Phoenix metropolitan area and were
performed using data, modeling tools, and staff support of the Maricopa Association of
Governments (MAG).
Conventional four-step transportation planning models are unable to account for
important differences in land use as represented by the 4Ds. The influence of land use is
most relevant at the level of the individual traveler and what they can walk to within ¼ to
½ mile at either origin or destination. Four-step models operate at a traffic analysis zone
(TAZ) level of aggregation, which is generally much too coarse for discerning land use
differences. For this reason, this genre of models also does not deal directly with
nonmotorized trips such as biking or walking, which are a critical element in compact
mixed use designs.
Using data from MAG’s 2001 regional household travel survey supplemented with
information from its travel model and geographic information systems (GIS) databases,
researchers developed a set of regression models to quantify the relationships between
travel behavior, traveler demographic characteristics, and measures of the 4Ds. These
models show the effect of the 4Ds of land use on both household vehicle ownership and
on household VMT. Residential density, land use mix, walk opportunities, and regional
transit accessibility to jobs were the variables used to represent the 4Ds. Negative signs
on the coefficients for these variables indicate that as each of the land use variables
3
increases, vehicle ownership and VMT rates decline proportionately. Vehicle ownership
is an important determinant of travel in the VMT models, so when it is reduced in
relation to better land use, its effect is compounded by also acting to reduce VMT.
The MAG region was divided into 17 jurisdictional areas and the household travel survey
database used to explore differences in travel in relation to these key land use factors.
Higher-density and more mixed-use areas such as South Scottsdale, Tempe, and East
Phoenix were found to behave significantly differently from lower- density/less mixed-use
areas like Glendale, Gilbert, and North Scottsdale. Residential density for the more
compact areas ranged from 6.14 to 6.94 households per acre vs. 2.86 to 3.61 households
for the lower-density group. These higher-density areas also had better mix (0.53 vs. 0.30
value on a 0 to 1.0 entropy index scale); more retail and service opportunities within
walking distance (42.4 vs. 15.4); and considerably more jobs accessible by transit (59,000
vs. 27,000). The implications of these differences may be seen in various travel measures,
including:
 Vehicle Ownership: 1.55 vs. 1.92.
 Average Trip Lengths: 7.4 vs. 10.7 miles for home-based work trips; 2.7 vs. 4.3
for home-based shopping trips; 4.4 vs. 5.2 for home-based other trips; and 4.6 vs.
5.3 for nonhome-based trips.
 Per Capita VMT: 10.5 miles per day vs. 15.4 miles per day.
The 4Ds models were subsequently used to investigate the potential impact of improved
land use characteristics in each of the 17 areas. To do this, average residential densities
were raised to 10 households per acre (vs. on the order of two to four in most places),
land use mix was brought to the ideal entropy index value of 1.0, the number of walk
opportunities was increased to 100 in all places, and regional transit accessibility was
raised to somewhere between the current minimum and maximum for the respective area.
This resulted in estimates of VMT reduction of 20 percent to 45 percent, with an average
overall of 25 percent.
Having reasonably demonstrated that areas in Phoenix with higher density generated less
vehicle travel per capita than lower-density areas, the second hypothesis investigated was
whether a higher concentration of activity would also lead to localized traffic congestion.
A sample of four urban corridors was selected for detailed study—also in the Phoenix
area—based on information from local and regional officials that these areas were
perceived to have major traffic congestion issues: Scottsdale Road between Thomas Road
and Chaparral Road in the older, southern part of Scottsdale; Central Avenue north of
downtown Phoenix, between McDowell Road and Camelback Road; the Mill Avenue
and Apache Boulevard corridor through Tempe; and West Bell Road in the northwest
part of the region, connecting the central valley with the newer communities of Peoria,
Glendale, and Surprise.
The objective was to examine the interplay between the intense development patterns in
these areas and the condition of traffic on the street and road network. Researchers
performed the following assessments:
4
 Density: Three of the areas—south Scottsdale, Central Avenue, and Tempe—
exhibited some of the highest development densities in the region while Bell Road
served an area of intense activity spread over a large area of moderate to low
density.
 Composition: The Scottsdale, Central Avenue, and Tempe areas also had a high
level of mix in their land uses, while Bell Road was heavily residential. The
overall jobs-to-housing ratio in the Bell Road area was only 0.49 compared to
1.42 in Scottsdale, 2.30 in Tempe, and 5.60 in the Central Avenue corridor. Retail
jobs per household (a measure of access to local services) was not quite as
skewed, but Bell Road’s ratio of 0.31 was still only about half of the 0.56 to 0.65
level found in the other three areas.
 Road Network: Each area is served by the one mile arterial “super grid,” with no
area having a freeway closer than two miles from its center. However, a major
distinction occurs in the secondary road system, with Central Avenue and
Scottsdale having a rich network of secondary streets on one-eighth mile spacing.
Tempe’s secondary grid is not quite as fine but is still much better than the Bell
Road corridor, which has little secondary road system beyond subdivision
networks.
 Transit: Central Avenue and Tempe are well-served by the regional bus system
and are also connected by the region’s inaugural light rail line (not operational at
the time of the analysis). Scottsdale is moderately served by transit, while Bell
Road has only park-and-ride bus service. Transit accounts for about 6 percent of
all internal trips in Scottsdale, and between 3 percent and 6 percent of external
trips. In the Central Avenue corridor, about 8 percent of internal trips and about 7
percent of external trips are made by transit, while in Tempe about 3 percent of
internal trips and 5 percent to 10 percent of external trips are by transit. In
contrast, less than 1 percent of all trips in the Bell Road corridor involve transit.
 Traffic Congestion: Interestingly, traffic congestion levels were much lower in
the Scottsdale Road and Central Avenue corridors than in the Bell Road corridor.
Volume-to-capacity (V/C) ratios in both Scottsdale and Central Avenue were in
the 0.8 to 0.9 range in the PM peak period compared to 1.6 to almost 2.0 along
Bell Road. Tempe fell predictably in between given its less-articulated secondary
road network, with V/C ratios in the neighborhood of 1.0. Tempe also employs
traffic-calming strategies on its secondary road network to discourage cut-through
traffic, which pushes traffic onto major arterials.
 Through Traffic: Traffic volumes in each of the four areas are affected by
through travel (no trip end within the defined area). Central Avenue, Bell Road,
and Apache Boulevard in Tempe all had rates of through travel that accounted for
about half of peak period traffic volumes. Without this through traffic movement,
Central Avenue and Tempe would be relatively uncongested, though Bell Road
would still be congested from its internal volume. Scottsdale’s rate of through
traffic on the measured links was much less—about 22 percent to 28 percent—
probably due to the design of the local grid, which encourages through travelers to
use peripheral streets.
5
These findings tended to corroborate responses elicited from participants in the project’s
survey of officials. For this survey, which was conducted early in the project, researchers
distributed 423 questionnaires and received 134 responses from a diverse list of elected
officials, planning and zoning officials, transportation planners, and members of other
relevant disciplines in the Phoenix, Tucson, and Flagstaff metropolitan areas. Some of the
key discoveries in this investigation are given below:
 Traffic Congestion Concerns: While important, traffic congestion was rated as
less a factor in project review than were issues of compatibility with adopted
plans and impact on surrounding neighborhoods and businesses.
 Familiarity: Most officials were familiar with mixed-use concepts, had been
involved in the review of these concepts, and had even encouraged submission of
such projects.
 Transportation Impacts: The overwhelming majority of officials responding
believed that compact, mixed-use development would increase transit use and
nonmotorized travel, though only about one-third felt unequivocally that it would
lead to less traffic congestion. (Most were unsure.)
 Desirability: The great majority believed that the region would benefit from more
mixed-use centers and corridors, focusing employment in centers and corridors,
and building more mixed-use communities. About 80 percent believed that their
own community would support compact, mixed-use development.
Residential/retail and office/retail mixed use were the most highly rated
combinations.
These findings suggest an opportunity to advance the dialogue on and support for
compact, mixed-use development in Arizona’s metropolitan areas. Among the initiatives
that might be considered are the following:
 Education: There is a need to better inform the public, the business community,
and officials about the nature and benefits of compact, mixed use. Themes
developed in this project can serve as educational messages.
 Better Analysis Tools: Local planners and planning commissions are still using
traditional traffic engineering approaches to assess the impact of development
projects. By looking only at traffic congestion levels on adjacent links, ignoring
through travel, and failing to account for the efficiencies of mixed-use
development on lower vehicle trip rates and VMT, progressive projects are likely
to be rejected or unreasonably downsized. The metropolitan planning
organizations should take steps to add 4D enhancements to their existing tools.
 Visioning and Plan Overhauls: Existing long-range or comprehensive plans
may be silent or devoid of a position on compact, mixed-use development.
Regional or local targeted visioning exercises can raise visibility and
understanding of the issues, leading to greater acceptance and support in updated
plans.
 Incentives: Adoption of compact, mixed-use development approaches can be
encouraged in various ways. Grant monies and/or technical assistance can be
offered to support studies or demonstration projects. Several states prioritize state
program or grant funding based on demonstrated steps by a jurisdiction to
embrace and incorporate key elements in their plans, codes, or procedures.
6
 Supportive Infrastructure: A key incentive in its own right, local land use
choices can be influenced by the manner in which transportation resources are
distributed. Priorities can be placed on investments that will most contribute to
concentrated land use policies such as transit investments, local street and
sidewalk infrastructure, or rehab/upgrade of facilities in older developed areas.
7
CHAPTER 1. INTRODUCTION
PURPOSE OF THE STUDY
The purpose of this research project was to examine the relationship between land use
and traffic congestion, recognizing that the way land is used affects the volume and
character of traffic on the street and highway network. Similarly, adding new roads or
expanding existing roads has an impact on the way abutting land is used.
There is considerable controversy over whether increasing the density of development
(i.e., a higher number of persons or employees per square mile of land) would reduce or
increase traffic congestion. Some researchers argue that compact, mixed-use
development is inherently more efficient and sustainable, using less land and reducing
private vehicle use rates by bringing people and activities closer together, and also
providing densities that are capable of supporting walking and effective transit services.
Other researchers say that conventional patterns of low-density development with
different land uses (residential, commercial, industrial, institutional), separated from each
other and reachable only by car, are much more in character with Americans’ preference.
Further, they argue, increasing density will only lead to more traffic congestion and loss
of personal mobility.
Better data on the relationship between land uses and traffic congestion could help lead to
better decisions—decisions that could help reduce traffic congestion, improve air quality,
enable safer travel, and lower roadway infrastructure costs.
Given the abundant supply of available land in the Southwest and that most growth has
occurred during the era of the automobile, it is no surprise that development patterns in
Arizona have been expansive rather than concentrated. In such an environment, the
consideration of higher density, more urban growth concepts may seem out of place.
However, growing evidence from research on the nature of compact, mixed-use
development—particularly if it is also focused around efficient regional transit service—
suggests that it generates considerably less vehicle use and VMT than contemporary low-density
development. The question is whether these benefits are sufficiently high that
higher-density development should be seriously considered in sustaining travel mobility
over the long term or whether any such concentration of development would also
generate proportionately higher levels of congestion in the adjacent area that would be
unacceptable and negate any positive results.
In terms of public acceptance, the nature of this issue is sufficiently volatile that any
attempt to promote compact land use policies could raise questions and possible
resistance. To better understand these relationships and define an appropriate policy or
position on land use, the Arizona Transportation Research Center (ATRC) commissioned
this study to perform the following investigations:
8
 Review the key literature on the relationship between land use density and
intensity (i.e., land uses like sports arenas or shopping malls that tend to draw
large volumes of traffic) and traffic congestion.
 Survey/interview a sample of officials from metropolitan Tucson and Phoenix
regarding their land use decision-making and the level of attention paid to traffic
congestion.
 Evaluate a set of urban corridors (i.e., major roadway thoroughfares, including
connecting roadways traversing an urban region) by using samples from
metropolitan Phoenix and Tucson. Upon selection, use appropriate data to
examine the nature of the relationship between land use and traffic congestion in
the corridors.
 Identify methods by which future land use decisions could be made to better
contribute to mitigating traffic congestion.
RESEARCH APPROACH AND REPORT ORGANIZATION
These objectives became the basis of the research plan, each through a major task activity
resulting in a task report summarizing the approach and key findings. The task reports
were reviewed by members of the Technical Advisory Committee (TAC), with comments
and suggestions reflected in the final documents. These earlier products have now been
integrated into a complete report, with each major activity report representing a chapter in
the final report document. The following describes the program of research, which
corresponds to the organization of this report.
Chapter 2. Literature Review
Early in the project, investigators conducted an extensive review of the literature about
land use related research studies to provide a solid basis for shaping the research
hypotheses to be tested and providing empirical evidence of the effects of land use on
transportation. A synthesis study on this subject performed for the Transit Cooperative
Research Program (TCRP) of the Transportation Research Board (TRB) served as a
starting point (Kuzmyak et al. 2003). Chapter 15, “Land Use and Site Design,” is one of
19 individually published volumes that make up TCRP Report 95, Traveler Response to
Transportation System Changes Handbook. The literature reviewed in this earlier project
dated back to the 1970s, and was then updated for the current study through a series of
computer-assisted searches of reports, monographs, and journal and newspaper articles.
Researchers identified and catalogued more than 100 sources, and incorporated findings
from about 70 sources into the new review. These findings were organized according to
the following topics:
 Trends: Examines trends in population growth, demographic changes,
development patterns, and transportation investment as they relate to rates of
growth in vehicle ownership, use, and VMT.
 Key Factors: Evaluates land use attributes such as density, mix of uses,
connectivity, and accessibility that appear to affect travel behavior.
 Travel Markets: Explores how land use impacts different purposes of travel,
particularly work vs. nonwork travel.
9
 Transit: Explores the role and importance of transit investments in guiding
development patterns and impacting travel choices.
 Self-Selection: Addresses the issue of whether the differences in travel observed
in compact, mixed-use areas are due to design or to the characteristics of people
who choose to live there.
 Density vs. Congestion: Explores contemporary arguments and evidence about
the link between higher density development and traffic congestion.
 Market Forces and Equity: Explores concerns about whether compact mixed
land use is affordable and whether planning for it defies market forces, and social
trends in location preferences that may be challenged.
 Planning Capabilities: Addresses the extent to which planning for compact,
mixed-use development is limited by the structure and capabilities of planning
tools (models) and supporting data.
Chapter 3. Survey of Officials
Report findings from a detailed survey targeted four types of officials in relation to their
role in land use decision-making: local elected officials (mayors, commissioners,
councils); planning and zoning officials; professional land use and transportation
planners; and other state and regional officials. The survey elicited information about:
 Methods of becoming involved in the land use planning and development process.
 Key factors influencing development decisions.
 Methods for assessing transportation impacts.
 Experience with and attitudes toward compact, mixed-use development.
 Expected transportation effects of compact, mixed-use development.
 Adequacy of information for assessing mixed-use development proposals.
 Most appropriate development options for a specific region and in the official’s
own jurisdiction.
Researchers also used the survey to identify the most congested highway corridors for
consideration in the detailed analysis of land use impacts on transportation and traffic
congestion.
Chapter 4. Analysis of Corridors
This chapter describes the results of the core activity in this study, namely investigation
of the nature and degree of the relationship between development patterns and adjacent
traffic congestion. After considerable review of potential examples, investigators selected
four corridor areas from the Phoenix area: Scottsdale Road, North Central Avenue, West
Bell Road, and the Mill Avenue/Apache Boulevard portion of Tempe. Next, they
assembled detailed information to examine the relationship between land development
characteristics and travel patterns and traffic congestion on adjacent facilities. Key steps
or elements of this analysis that are detailed in the chapter are:
10
 Detailed profiling of land use characteristics in the study areas, including density,
sociodemographic characteristics, jobs-housing balance, and general land use
mix.
 Transportation (street/highway) network characteristics, capacity, and traffic
conditions.
 Traffic composition, in particular the proportion of volume on selected links that
have an origin or destination within the study area vs. the share that is “through”
traffic.
 Transit network, coverage, service level, and use.
 Internal capture (retention) of resident travel within study area, by travel purpose.
 Average trip lengths across study areas in relation to development characteristics.
 Walkability and nonmotorized travel.
Chapter 5. Land Use Impacts on Design
Having demonstrated in Chapter 4 that high density was not clearly correlated with
higher congestion, the subsequent investigation was to explore options for the types of
policies that could be more conducive to reducing traffic congestion. Researchers
developed a methodology for identifying the effectiveness of key land use design and
transportation system measures in reducing VMT production. This work was performed
using data obtained from the MAG travel model database, land use database, and 2001
regional household travel survey. Using these data, researchers created a set of regression
models that predict the effects of different land use and transportation system options on
household auto ownership and VMT production rates. They used these modeling tools to
estimate the effects of higher density, better mix, better design, and greater regional
transit accessibility on VMT production for 17 distinct jurisdictions in the Phoenix
region.
Chapter 6. Summary, Conclusions, and Recommendations
The final chapter offers an overall summary of the project’s objectives, analytic approach
toward addressing those objectives, key findings, and recommendations for advancing the
concepts of compact, mixed-use development toward greater acceptance in planning and
implementation.
Appendices
Information in the appendices includes:
 A summary of open-ended responses to key questions in the survey of officials.
 A description of the candidate corridors recommended for study and the process
used to select the final sample for this analysis.
 Compilation of the MAG traffic counts used to validate the link congestion
analysis conclusions in Chapter 4.
11
CHAPTER 2. LITERATURE REVIEW
INTRODUCTION
To begin this investigation, researchers conducted a comprehensive search and review of
the relevant literature dealing with the transportation/land use connection. This review
encompassed a full range of research, empirical, and policy studies, tracking
developments in this topic since the late 1980s. Researchers used the findings from
Chapter 15, “Land Use and Site Design,” of the Traveler Response to Transportation
System Changes Handbook (Kuzmyak et al. 2003) as a starting point for this study. As a
result of this existing work, which dates back to the early 1970s, researchers could focus
the current review more heavily on work that has been done since 2002.
During this recent six-year period, the research and discussion has, if anything,
intensified. Many of the early debates—dealing mainly with whether land use does in fact
matter—have been substantially resolved, while new and more complex questions have
emerged, such as what land use factors actually contribute to travel behavior, and
whether residential self-selection plays a central role in the observed differences in travel.
The positive aspect is that the increased attention to the topic has brought more and better
data and methods, which have permitted more incisive analysis.
More than 100 sources were identified for this study, and almost 70 of those reviewed
were regarded as sufficiently relevant to be formally documented and included in this
synthesis. (See the bibliography of this report for a summary of the key points or findings
of these sources.)
Following a brief review of the history and trends in land use and transportation, the
remainder of this chapter summarizes the findings from this review, organized in relation
to the key research or policy questions that spawned the particular study or article. Those
questions are as follows:
 Does land use impact travel, to what extent, and through what mechanisms?
 What types of travel are most affected?
 How important is the role of transit?
 Is there a self-selection bias that confounds the travel effect attributed to land use?
 What is the market for acceptance of alternative land use approaches, and what
factors impede its propagation? Are there equity implications attached to planning
approaches that manage land use?
 Does density cause more congestion, and what are its effects on mobility and
accessibility?
 How well do current travel models account for key land use relationships?
12
BRIEF HISTORICAL PERSPECTIVE AND TRENDS
In the late 1980s, important trends were converging to draw high-level attention to the
issues of urban sprawl, traffic congestion, and associated fiscal and environmental
concerns. Whereas the predominant growth pattern following World War II was
suburban, the mid-1980s marked an important tipping point in this trend when the
majority of employment—not just households—came to be located outside the central
cities of metropolitan areas. A variety of factors fueled this growth, primarily the excess
capacity in suburban portions of the interstate highway system and an era of “cheap
money” from the savings and loan industry. The former made huge portions of
undeveloped land accessible to the metropolitan region, while the latter encouraged large-scale
real estate speculation. Seemingly overnight, major office parks and massive
regional shopping malls sprouted up around highway interchanges, seizing upon the high
accessibility, lower space cost, and the instant advertising associated with those locations.
In Arizona this trend was further encouraged by the attraction of sales tax revenues from
these new projects.
The readily available capital led to massive overbuilding in many areas, empty structures,
land flips, and the subsequent demise of the savings and loan industry. However, the
exodus of employment from the central business district was accelerated, along with
well-established residential trends, most of it without the construction of new
noninterstate transportation capacity—particularly for cross-suburban movements. As a
result, suburban-to-suburban traffic flows rapidly grew to consume available road
capacity, suddenly transforming many previously tranquil, quasi-rural areas to traffic
levels previously seen only in core urban areas. The problem was of sufficient gravity
that the U.S. Department of Transportation (USDOT) labeled it the “suburban mobility
crisis” and directed considerable resources to the development of initiatives aimed at
stimulating local solutions to these problems through demand management and public-private
partnerships. Effective solutions were not found, since the land patterns and
densities were described as dense enough to cause major traffic congestion, but not to
support transit or other transportation alternatives. Because new road capacity was
limited by available space and funding, most of these areas continue to experience
pronounced levels of traffic congestion, limiting additional growth that has been forced to
move further outward because of growth caps or traffic ordinances.
Superimposed on these critical trends of the late 1980s, the nation’s transportation
infrastructure faced a fiscal crisis. Years of deferred maintenance were suddenly made
evident in broken pavements, the collapse of several major bridges, and a determination
that many others were structurally or functionally “deficient” in their ability to carry the
traffic volumes and vehicle weights that were being increasingly imposed on them
(because of an increase in large trucks, including tandem semitrailers). Critical
maintenance, repair, and replacement had been ignored because of insufficient planning
and a failure to allocate resources at the federal and state level. The critical nature of this
discovery led to the passage of the revolutionary Intermodal Surface Transportation
Efficiency Act of 1991, or ISTEA, which engineered changes to the federal gas tax and
transportation trust fund to avail billions of additional dollars to state transportation
13
programs to avert a collapse of the nation’s transportation system. With these new funds,
however, came strict new conditions, forcing states and metropolitan areas to make
preservation their top priority, followed by safety, management (measures to improve
efficiency), and capacity expansions. Consequently, 85 percent of federal road dollars
have been consumed by preservation, with generally less than 5 percent available for new
capacity. As a result, relatively little money has been available to address congestion
problems through new roads—a problem that not only hasn’t gone away, but has become
more compelling as deteriorating conditions are again making news and the yield from
both federal and state gas taxes (fixed cents per gallon) has declined in the face of rapidly
rising fuel prices.
Key Trends
A recent study for USDOT explores some of the key trends in travel that have been
occurring in the United States over the past several decades (Polzin 2006). In his
assessment, Polzin concludes that changes in the trends that have been driving the growth
in VMT may in fact be slowing down, though the nonlinear relationship between VMT
and congestion is such that slower VMT growth may not portend lower rates of
congestion growth. This is because of the non-linear relationship between volume,
capacity, and speeds, where additional increments of volume beyond a given V/C ratio
result in rapid declines in speed, and even reductions in speed and throughput beyond a
critical V/C level.
Using aggregate data from the Federal Highway Administration (FHWA) Highway
Statistics database and household travel data taken from the National Household Travel
Survey (NHTS), researchers plotted total population, total VMT, and total household
VMT from 1977 through 2001 to coincide for comparison purposes with the dates when
the NHTS was conducted (Figure 1). While the population has only grown by 30 percent
during this period, total VMT has grown by more than 90 percent, and the VMT
generated by households (i.e., travel for personal as opposed to commercial purposes) has
grown by 151 percent. The fact that household VMT has grown five times faster than the
rate of population provides strong corroboration with people’s perception that traffic
congestion has gotten rapidly worse. Statistics on highway lane mileage from Highway
Statistics further shows that road capacity has grown no faster than the rate of population
during the same period.
14
Figure 1. Population and VMT Trends.
Demographers such as Alan Pisarski have attempted to explain the factors underlying this
growth. Key factors cited include increases in household income, workforce participation
and driver’s license acquisition by women, increased auto ownership, and other changes
in household composition (such as size and life cycle). Discussion about these trends in
forums on traffic congestion staged throughout the 1990s generally maintained that they
would soon reach saturation levels, leading to a subsequent tailing off in the rate of
growth in VMT and congestion. Pisarski later acknowledged that in his 1996 Commuting
in America II, he was wrong about single-occupant vehicle (SOV) use having peaked,
and that it has continued to rise. However, he attributed this error in projection to
growing affluence and travelers’ association of driving with maximizing their ever-increasing
value of time (Pisarski 2005).
15
Conspicuously missing from these assessments, however, is any accounting for the role
of land use. Throughout this period, American society was rapidly suburbanizing,
resulting in greater reliance on the automobile for household travel, fewer opportunities
to walk or take transit, fewer occupants per auto trip, and longer trip lengths for trips of
all purposes. Among the key demographic changes that were occurring in households
during this period was a decline in average household size of 18.2 percent as baby boom
children began to leave home. These trends, shown in Figure 2, as well as those in
Figures 3 and 4, are taken from the 2001 NHTS Summary of Travel Trends (Hu and
Reuscher 2004).
Figure 2. Household Characteristics: 1969 through 2001.
During this period, more women entered the workforce as workers per household
increased by 11.6 percent, and more acquired drivers’ licenses, increasing drivers per
household by 7.3 percent. However, the most graphic change beyond the decline in
household size was auto ownership, which increased 62.9 percent, from 1.16 per
household to 1.89. This means that the number of vehicles actually surpassed the number
of licensed drivers in the average household. With those additional vehicles, increasingly
necessitated by suburban residential and work locations, household vehicle travel
accelerated after 1983. As illustrated in Figure 3, despite the decline in household size,
the number of daily vehicle trips almost doubled—from 3.83 per day in 1969 to 5.95 in
2001, and daily household VMT accordingly increased from 34.01 to 58.05 miles—an
increase of more than 70 percent. On a per-person basis, given the simultaneous
reduction in household size, per capita household VMT more than doubled, from less
3.16
2.58
1.65 1.69
1.78
1.77
1.21 1.23 1.21 1.27 1.33 1.35
1.59
1.89
2.56 2.63
2.83
2.69
1.72 1.75
1.68 1.77
1.16
0
0.5
1
1.5
2
2.5
3
3.5
1969 1977 1983 1990 1995 2001
Persons per Household
Drivers per Household
Workers per Household
Vehicles per Household
16
Figure 3. Change in Household Travel Characteristics: 1969 through 2001.
than 11 miles per person per day to more than 22. This increase occurred not because of
longer vehicle trips lengths per se—which increased moderately from 8.5 to about 10
miles—but mainly because more vehicle trips were made.
Moreover, the primary increase in VMT has come not from more commute travel, which
has been mitigated somewhat by employment moving toward the suburban population,
but from the increased reliance on the automobile for nonwork travel. Figure 4 illustrates
this trend clearly. While annual household VMT for travel to and from work has
increased moderately—from about 4200 to about 5700 miles per year (even with more
workers per household)—VMT associated with travel for nonwork purposes has jumped
from 8240 per year to 15,463, an increase of 87.7 percent. The biggest increases were in
conjunction with travel for shopping, which increased by 130 percent, and personal and
family business, which increased by 112 percent. The reasons for this comparative
increase in discretionary travel may have something to do with increases in real income,
but (as will be shown later) is much more directly tied to the broad scale shifts in location
that have occurred, to places where it has become necessary to drive for virtually all
family-related business, from shopping to transporting children. Nonwork travel now
makes up 73 percent of daily household VMT, up from 66 percent in 1969, composing
the great majority of vehicle travel activity. Contrary to intuition, this nonwork auto
travel also comprises more than half of the travel on the roads during peak travel hours.
2.58
34.01 32.97 32.16
49.76
57.25 58.05
4.07
5.69 6.36 5.95
3.16 2.83 2.69 2.56 2.63
3.83 3.95
0
10
20
30
40
50
60
70
1969 1977 1983 1990 1995 2001
Persons per Household
Daily HH VMT
Daily HH Vehicle Trips
Average Vehicle Trip Length
Average VMT per Person
17
Figure 4. Source of Growth in Annual Household VMT.
These are the primary trends that help explain the changes in travel patterns and the
growth in traffic congestion and delay being experienced in most of America’s
metropolitan areas.
HOW DOES LAND USE IMPACT TRAVEL?
Intuition suggests that traffic volumes on a busy arterial roadway are associated with the
type and density of activity built along that roadway. For example, the entrance to a
shopping mall is frequently teeming with cars, and suburban commercial strips also give
visual evidence of customers maneuvering in and out of parking lots, or disrupting flow
by trying to cross several lanes of traffic to make a turn at a busy intersection. These are
inescapable facts of urban life. But what about bumper-to-bumper traffic on freeways?
Ten mile backups on a regional beltway? Or agonizing signal-to-signal delay moving
along a major or minor arterial, even where there is no obvious nearby commercial or
higher density residential development?
The fact is that travel patterns are complex, with travel being a “derived demand”
necessitated by the needs of households and customers to reach daily activities. In the
transportation planner’s parlance, the nature of the subsequent travel demand is best seen
as a regional “trip table” of productions and attractions, or demands for travel matched
against multiple locations where those demands can be fulfilled. Many factors are
considered when deciding where to go to satisfy a particular trip purpose, although travel
11,739
21,187
5,724
8,201
15,463
20,895
18,161
12,423 12,036
6,492
4,853
3,815 3,538 4,183
8,240 8,221
13,308
14,403
0
5,000
10,000
15,000
20,000
25,000
1969 1977 1983 1990 1995 2001
Annual VMT
To/From Work
All Non-Work
18
time and cost are frequently prominent in these decisions. The transportation network is
the mechanism by which this diverse pattern of origin demand and destination supply is
connected. Trips of many purposes and geographic orientation are superimposed upon the
network at any given time such that at any given location, the travel stream may be
composed of trips with many different purposes from many different locations
throughout the region. Typically, trips being made on the highest functional class of
highways (freeways and interstates) are the longer trips on the system, while those on
arterial and connector roadways have proportionately higher shares of local travel.
However, these relationships quickly dissolve if congestion clogs one group of facilities
more than another or the road system lacks sufficient connectivity between particular
points. As a result, travelers in suburban areas frequently use the freeways to make local
trips, traveling only between one or two exits, or longer distance travelers use local roads
to avoid congestion on higher class highways or as a shortcut. Hence, traffic volumes and
congestion on a given roadway segment are seldom well explained by immediately
adjacent development, but have multiple contributing causes.
Preferences with regard to housing type location and affordability, employment, schools,
and other factors cause households to make varied decisions about where to live, where
to work, where to shop or recreate, and how to travel. Subsequently, often-unforeseen
changes in household structure, employment status or location, or economic conditions
may call into question location choices that made great sense at an earlier time. Growth
and new development have a way of changing the conditions that were once prized and
expected, such as when a trouble-free, 30 minute commute gradually becomes a
frustrating 60 to 90 minute ordeal, or a favored shopping district is suddenly impossible
to access at particular times.
While no policy can guarantee a consistent level of access or travel speed to a given
location, an important hedge against losing the amenity associated with a particular
location is to ensure that it has travel options. If a long commute becomes too unbearable
to drive, the commuter with options lives within easy access of an efficient transit service
or can take advantage of a priority high occupancy vehicle (HOV) lane. For residents
fearing difficult access to traffic-choked shopping centers, location in a community that
offers a variety of services and amenities locally—either within walking distance or a
short car trip on local streets—can be a periodic or regular substitute to longer trips in
heavy traffic. Unfortunately, because of the uniqueness and scarcity of these communities
near transit with good local land use, the cost of living in these areas is frequently too
high for persons of limited economic means.
Few residential communities offer the option of good transit or HOV connectivity to the
surrounding region, or local access to key domestic needs and amenities. And too few
destination areas offer freedom from driving once there by virtue of compatible activities
and pedestrian access. Suburban development patterns stress uniformity (believing that it
is essential to keep different land uses separated), create housing developments that
ensure socioeconomic homogeneity, and shape street patterns to keep out through traffic
and make it easy to drive to. However, these designs also reduce travel options and place
residents under the influence of broader external development trends.
19
So, does land use have an association with travel behavior, and if so, what is the manner
of that association? At the most primary level of whether land use matters, a number of
academics took an early position that any link between land use and travel behavior was
inconclusive. Included in this group are such noteworthy spokespersons as Genevieve
Giuliano, Peter Gordon, Harry Richardson, Randy Crane, Marlon Boarnet, Pat
Mokhtarian, Xinyu Cao, and Susan Handy. Most if not all of these researchers are based
in California, and data from which their early conclusions were drawn (also principally
from Southern California) showed no clear association between characteristics of land
use as they measured them and travel behavior. Giuliano and Crane (Shay and Khattak
2005) maintained that the relationship between urban design and travel behavior was
complex and not completely understood. Giuliano argued that because urban areas in the
United States are already so accessible (by automobile, mainly), settlement patterns so
well-established, and maintenance of privacy so important, transportation plays an ever-decreasing
role in the locational decisions of households and businesses. Crane and
Boarnet, in series of studies, determined that “the relationship between neotraditional
design and travel behavior is made more complex by the difficulty of isolating the
various design elements that may have a causal relationship to travel behavior –
moreover, some traits, such as visual appeal, sense of safety or community, are hard to
define and the synergy of design is hard to measure” (Shay and Khattak 2005). To the
extent that any differences in behavior were observed, most of these researchers
suggested that the predominant effect being observed was one of “self-selection”—
namely, that persons who were inclined to walk, take transit, or be comfortable living in
closer contact with others were more likely to choose to reside in places that offered
those opportunities. As discussed earlier, a key issue is also that these areas are currently
only affordable to people of means. The issue of self-selection is explored in a later
section of this report.
Such a uniform dismissal of a relationship between land use patterns and travel behavior
contradicts a large and growing body of empirical evidence from other studies. A much
larger contingent of researchers has taken the position and conducted the research to
affirm that in fact, land use does have an impact on travel behavior. Key among this
group are Robert Cervero, Kara Kockelman, Kevin Krizek, Reid Ewing, and Larry Frank
(Kuzmyak et al. 2003). A variety of empirical studies performed in the mid- to late-1990s
revealed that households residing in more traditional urban neighborhoods with a mixture
of uses, walkable streets, and good access to transit tended to own fewer vehicles, make
fewer vehicle trips, and generate less VMT than their counterparts living in suburban
subdivisions.
A 1995 study by Cervero and Radisch (1996) of two demographically similar
neighborhoods in the East Bay area of San Francisco found significantly lower rates of
auto use in the traditionally designed neighborhood for commuting (63 percent vs. 79
percent) and for nonwork travel (85 percent vs. 96 percent). A significant difference was
found in the degree to which residents in the traditional community walked or used
bicycles, amounting to 52 percent of all trips under two miles in the traditional
community vs. 17 percent in the comparison community. In a 1997 study of trip rates and
20
VMT in Seattle neighborhoods, Rutherford, McCormack, and Wilkinson (2001) found
that average daily travel mileage by all modes was considerably less (17 to 22 miles per
person per day) in three traditional neighborhoods than in demographically matched inner
and outer suburban neighborhoods (30 and 39 miles, respectively). Walk shares were also
much higher (18 percent vs. about 2 percent to 3 percent) in the traditional neighborhoods
(Kuzmyak et al. 2003).
Working with data from Baltimore’s 2001 regional household travel survey, Kuzmyak
developed measures of household and per capita VMT for 32 different neighborhood
clusters in the survey sample, and compared their performance with residential density as
a measure of urban vs. suburban land use. As seen in Figure 5, a remarkably strong
relationship was found between per capita VMT generation in these neighborhood
clusters and residential density, with households in the more urban locations generating
between 10 and 20 VMT per person per day compared with rates of 30 to 50 miles per
day in the more typical suburban areas. What is interesting about this relationship is not
only the tendency for VMT rates to increase with lower densities, but to do so at a
nonlinear rate, reflected in the logarithmic curve fitted to the data with an R2 value of
0.727. This relationship was surprisingly independent of traveler affluence; as a
regression of household VMT vs. household income failed to show a statistically
meaningful relationship. Very similar relationships were discovered in data from the
Washington, D.C. region (Kuzmyak, Baber, and Savory 2006).
Figure 5. Daily Per Capita VMT vs. Residential Density in Baltimore Region.
21
Early studies such as these simply compared communities on a simple measure like
residential density, or on a binary basis, as to whether they did or did not resemble
traditional vs. conventional settings. Subsequent research has become much more focused
on identifying the specific characteristics of land use that impact travel, while more
directly controlling for socioeconomic and other key differences.
What these later studies have revealed is that land use is about much more than simply
density. By reducing distances between households and activities, compact development
improves accessibility by all modes of travel. Walking becomes more feasible, but also
vehicle trip lengths are shortened by the existence of more local opportunities. While
density is a strong surrogate for proximity, the kinds of land uses that are mixed and the
character of the mix also matter. Also, the ease with which travelers can part with their
cars and walk to and among these various activities as well as to reach transit service is
an important determinant. This set of relationships has been dubbed the “3Ds” of local
land use—density, diversity, and design—with density reflecting intensity of
development (people or jobs), diversity representing both the degree and balance of mix,
and design representing the layout of the area in relation to ease of pedestrian access.
While it is difficult to attribute the coining of the 3Ds concept to any given researcher,
Cervero and Ewing are widely associated with its use and quantification. In a paper titled
“Travel and the Built Environment—A Synthesis,” , the authors presented elasticities of
demand for vehicle trips and VMT related to each of the 3Ds, which they abstracted from
14 different studies (Ewing and Cervero 2001). These elasticities were subsequently
adopted into the U.S. Environmental Protection Agency’s (EPA’s) Smart Growth Index
Model for use in local planning activities and for areas seeking emissions credit for land
use actions (Kuzmyak et al. 2003).
Most land use researchers, including Cervero and Ewing, also recognized that the effects
of land use on travel behavior were not just occurring at the neighborhood level, but at
the regional level as well. In essence, the link between land use and travel behavior is tied
to the concept of accessibility—the nearness of activities and the breadth of choices
represented in those opportunities. The 3Ds do a good job of reflecting local accessibility
differences, but also important is the degree of access to opportunities available outside
the neighborhood, or regional accessibility. Picturing two communities that are otherwise
similar with regard to local mix and design, clearly the community that offered more
activities and opportunities to its residents outside its boundaries within a given travel
time window would have a much different profile in terms of travel opportunities and
subsequent travel choices. This greater regional accessibility is a function of both the
physical proximity of external opportunities and the ability of the transportation system to
reach them. Hence, areas with good highway access and connectivity would be expected
to have good regional accessibility, but the same area served by good regional transit
service would be expected to have even greater regional accessibility when viewed from
a multimodal perspective. Regional accessibility is now often referred to as the fourth D
of land use, representing destinations. Ewing and Cervero also derived an elasticity for
regional accessibility to employ in the Smart Growth Index Model. They found that if
regional accessibility were doubled for a given household, VMT would decline by about
22
20 percent. And if density, diversity, design, and regional accessibility were all doubled,
the combined impact would be a 35 percent reduction in household VMT (Ewing 2005).
MOST IMPORTANT VARIABLES AND TYPES OF TRAVEL MOST
AFFECTED
A number of studies have focused on measuring the ways in which land use
characteristics influence travel (Bose and Fricker 2003; Boarnet and Sarmiento 2003;
Barnes 2003; Khattak and Rodriguez 2005; Krizek 2003a, b; Krizek and Johnson 2006;
Paez 2006; Soltani and Allan 2006; Hess et al. 1999; Hess, Vernez-Moudon, and
Logsdon 2001; Shay and Khattak 2005; Ewing 2005; Frank, Kavage, and Litman, 2006;
Zhang 2006; Lam and Niemeier 2005; Urban Land Institute 2005; Cervero 2006; Targa
and Clifton 2004; Marshall and Grady 2005; Rodriguez, Khattak, and Evenson 2006; and
Yi 2006).
Perhaps the landmark study in measuring the relationships between transportation and
land use was performed by Kara Kockelman. Using data from the San Francisco Bay
Area, she attempted to explain differences in travel as a function of three key factors:
socioeconomic characteristics, regional accessibility, and local accessibility. Kockelman
used data from the regional household travel survey as the source for trip rates by mode,
VMT, and auto ownership. She measured regional accessibility as the cumulative number
of jobs reachable in all other TAZs, discounted by the travel time to reach them. And she
tested a variety of 3Ds measures for local accessibility, including population and
employment densities, and measures of diversity for land use mix and entropy for mix
balance derived using GIS tools and fine-scale land use data. Regression analysis was
used to determine the degree of statistical relationship between these factors and
household VMT, trips by auto and walking, and auto ownership. Her analysis showed
that household VMT increased with household size, income, and auto ownership, but
declined with improvements in regional accessibility (elasticity of -0.31) and in the 3D
variables of mix (dissimilarity, elasticity of -0.10) and balance (entropy, elasticity of
-0.10). Density did not prove to be an important explanatory factor. Another important
finding was the role of land use on auto ownership, with increases in regional
accessibility, and local dissimilarity and entropy all acting to reduce household auto
ownership (Kockelman 1996).
Using a similar approach, Kuzmyak developed a set of household VMT and auto
ownership models from Baltimore data. This research found the same direction of
influence from the regional accessibility and local 3Ds variables, but was strengthened by
the addition of a new variable to measure walkability (design). Kockelman did not use
such a measure in her models, fearing problems of subjectivity with a pedestrian
environment friendliness (PEF) type index of walkability. Using GIS tools as well as a
database locating employment by size, type, and specific geography, Kuzmyak created a
walk opportunities index that summarizes the opportunities lying within a one-quarter
mile buffer of a household. Each opportunity is identified, given a value based on the
Standard Industrial Code (SIC) identity and size, and its value is discounted by the walk
time required to reach it on the respective street grid. The index is quite similar in
23
behavior to the measure of regional accessibility, only for walking. The Baltimore models
proved to be statistically more robust (much higher R2 values), while the elasticities for
the land use variables were of similar magnitude. The Baltimore research also
demonstrated the specialized role played by local land use. The local 3Ds variables were
very important in the models predicting auto ownership and nonwork VMT (and total
VMT), but were not significant in the home-based work models. What this implies is that
households apparently make decisions about how many vehicles to own and how to
travel for nonwork trip purposes based on the 3Ds characteristics of their neighborhoods
(as well as regional accessibility, of course), but these characteristics do not seem to be
important in work-related travel. The influence of local land use is felt indirectly, through
auto ownership, and again through multimodal regional accessibility for work-related
travel (Kuzmyak, Baber, and Savory. 2006).
Supporting this important finding is research performed by Solimar Research for the
South Bay Cities area of Los Angeles. Researchers used travel surveys to study travel
behavior in four mixed-use neighborhoods in the southwestern portion of Los Angeles
County: the older portions of Redondo Beach, Torrance, Ingleside, and El Segundo.
These areas included a mix of socioeconomic levels, but also a reasonable offering of
shopping and services within walking distance of residents. The survey found that
residents of these areas made about three-fourths of their grocery and other shopping trips
and about half of their restaurant trips to the local center. The percentage of these trips
made by walking as opposed to driving ranged from 31 percent to 72 percent, depending
on the trip type, the particular area, and the distance of the household from the center of
town. Residents also made many walk trips to the centers simply for pleasure and
exercise. This travel behavior stood in stark contrast to almost exclusive reliance on
driving for work trips, given that most workers were employed at noncentral locations
which were not served by transit and where free employer parking was almost universal
(Solimar 2005).
Similar results were found in a 1998 study by R. L. Steiner of six traditional shopping
districts in the Oakland-Berkeley area of San Francisco. The districts, which had a variety
of mix and scale of business activity, were all in middle class neighborhoods of
moderately high density (13 to 21 persons per acre), and had Main Street type
characteristics with good pedestrian access. Surveys found that a significant percentage
of customers at each site got there by walking. Weekday shares were 20 percent to 38
percent walk and 41 percent to 79 percent auto, with much higher walking rates (24
percent to 65 percent) among residents living within one mile of the district (Kuzmyak et
al. 2003, p. 15-52). So, while the primary benefits of mixed, compact land use may be on
nonwork travel, with a secondary effect on work travel through the influence of auto
ownership and regional accessibility, should we lessen expectations for land use to
influence work travel? Not necessarily.
While the suburban exodus of jobs has moved work closer to employees in many
instances, many other factors influence commuters’ travel choices. First, commuting to a
suburban job almost guarantees use of an auto. Not only are most of these areas too
scattered to be reached in any other way, but once there, the commuter is likely to be
24
dependent on a private vehicle for any other need. Trips for food and personal business
like banking, filling a prescription, or attending a meeting generally require use of car. So
the 3Ds of land use that are so important in lessening car dependency at the residence
also come into play at the destination. The degree to which employment destinations have
walkable densities, mix of uses, pedestrian facilities, managed parking, and ideally transit
access has a major impact on commuting mode choice decisions. In his paper “Built
Environments and Mode Choice: Toward a Normative Framework,” Cervero found that
accounting for density, mix, sidewalk coverage, and regional accessibility in home-based
work mode choice models added major explanatory power in predicting the likelihood
that commuters would opt for alternatives modes (Cervero 2002).
Another important land use factor in work travel behavior is jobs-housing balance.
Demographers and trends specialists like Pisarski point out that with multiworker
households, it becomes very difficult to optimize residential location to ensure an
efficient commute. Seldom would it be expected that both wage earners would work in
the same general location and, hence, share the benefits of a planned commute advantage.
Moreover, given the frequency with which either job or residence locations change, an
“ideal” location often abruptly shifts on one or both ends, rendering the original location
planning moot. While not a complete solution to this practical dilemma, an important
planning consideration is the balance in the location of jobs and housing.
In many areas, local jurisdictions have tried to direct their employment growth to
particular areas, often distinct from current or proposed housing. The result is long
commutes over imperfect road networks, often involving long, circuitous paths that add
miles to the actual distance. Added to this is a frequent imbalance in functional
jobs/housing balance, where the jobs are not particularly well-matched to the
characteristics of the resident workers. Conversely, appropriately skilled workers for the
given jobs cannot find housing nearby that they can afford. Each of these imbalances
exacerbates the efficient connection of worker to job and contributes to trip length and
traffic volumes.
Table 15-14 in Kuzmyak et al. presents findings from a number of studies of the effect of
jobs/housing balance and commute travel behavior. Review of this information
concluded that even with good matching of resident and workplace job skills, jobs-housing
balance is at best an indicator of the potential for matchups that would internalize
commute travel in small areas. However, as area size grows, jobs-housing balance
becomes more of a force in enabling productive matchups. Results of studies by Frank
and Pivo, Ewing, Cervero, and others suggest shorter commute trip lengths by 7 percent
to 30 percent in balanced areas. The strength of this relationship must be tempered,
however, by the context in which the measurement is made, since characteristics like
density, centrality in the urban region, and transit access have an important bearing on the
ultimate benefit of balanced jobs and housing (Kuzmyak et al. 2003).
A fairly recent study by Cervero and Duncan attempted to determine whether jobs-housing
balance or retail-housing mixing produced the greater impact on vehicle travel.
Using data from the San Francisco Bay Area, they examined the degree to which job
25
accessibility is associated with reduced work travel and how closely retail and service
accessibility to residences is correlated with miles and hours traveling to shopping
destinations. They found that higher accessibility to occupationally matched jobs reduced
VMT and vehicle hours traveled (VHT) for work trips, particularly when such job
matches were plentiful within four miles of home. They found elasticities for work tours
to be considerably higher than those for shopping tours (0.329 vs. 0.137) such that even
while the share of daily VMT devoted to shopping and services was higher than for
commuting (42.8 percent vs. 36.7 percent), the higher elasticity meant that access to jobs
reduced VMT 72.5 percent more than access to shops and services (Cervero and Duncan
2006).
THE ROLE OF TRANSIT
Intensified, compact, mixed land use schemes are often associated with proposals for
major investments in rail transit systems. The resultant development, termed transit
oriented development (TOD) serves the dual objective of providing a logical location for
intensified development while also encouraging greater ridership levels on the transit
system. Advocates argue that the transit focus is essential to concentrating development
patterns in a way that is impossible with auto-shaped, low-density sprawl. Critics argue
that the massive resources diverted to a rail transit system are misspent, given the few
people likely to use the systems and the opportunity missed in diverting those resources
from highway projects that would benefit the most people.
Given the described importance of multimodal regional accessibility in shaping auto
ownership and longer-distance travel decisions, the strategic role that can be played by
high-quality regional transit is evident. If that transit service is independent of the surface
roadway network, as with rail or even bus rapid transit, its ability to provide a
consistently high level of accessibility to regional destinations over time amidst growing
road congestion has great value in preserving mobility. Perhaps the most strategic value
of such a system, however, is the excuse it provides to create development nodes around
station areas. These nodes then contain the characteristics of higher density, mixed use,
and walkability that breeds lower auto ownership, more internal trips, less VMT, and
more walking. While compact, mixed-use developments can be located virtually
anywhere, they are given additional stimulus when located near a transit node because of
the additional dimension of regional accessibility they provide the respective community.
Linked in a system, they also provide an ensemble of varied destinations that residents
can easily access if they can’t find what they want in their own neighborhood. TOD
specialists like G.B. Arrington suggest that rates of household vehicle trip generation in
TODs may be as much as 50 percent less than those in comparable conventional
developments (Arrington 2007).
At the same time, there is no denying the expense and skepticism associated with new
transit systems, whose primary purpose is to shape future land use. In mature
metropolitan areas like Washington, D.C., San Francisco, and Philadelphia, the basic
transit system already exists, its use patterns are well-established, and time has allowed
the importance of the transit stations to be translated into higher land values and demand
for higher intensity development. At that point, the major challenge is to guide that
26
development so it occurs in the most productive and sustainable fashion. In newer areas
that haven’t grown up in the presence of regional transit, like Los Angeles or Portland,
Oregon, the formula for success may not seem as easily replicated.
An article in the Los Angeles Times levied strong criticism at the logic of investing
billions of public dollars on transit and TOD projects in Los Angeles between 2001 and
2005, and yet residents are still driving. Reporters examined driving habits at four
housing complexes built at or near transit stations along both the Red and Gold lines and
found that only a fraction of the residents shunned cars and used transit, particularly
during the morning rush hour. They discovered that many of the drove to workplaces
because either their place of employment was not near a station, it was not easy to get
about without a vehicle at the destination, they had free parking, or it simply took longer
or cost more to take transit. However, the reporters also concluded that the transit
system’s failure was due to the false assumptions that most traffic was generated by
commuting and that most people worked downtown, neither of which is true. In fact,
most of the construction in the TODs to date has been for housing rather than
employment or mixed use, meaning that thousands of people are now clustered near
transit stations that they only occasionally use and still have few local travel options
(Bernstein and Vara-Orta 2007).
Many critics of TOD point to Portland, Oregon, as an expensive, failed experiment. One
such prominent critic is Randal O’Toole, founder of the American Dream Coalition.
Despite Portland’s unusual commitment to planning around transit, at the expense of
improving roads and allowing more freedom for development at the periphery of the
region (beyond the Urban Growth Boundary), O’Toole argues that the plan has not been
a success. While Portland’s planners claim that its residents love transit and use it
frequently, O’Toole points out that the region lost many transit riders in the 1980s when
the high cost of construction forced cuts in bus service, dropping transit share from 2.6
percent in 1980 to 1.8 percent in 1990. Over the next 12 years, while ridership slowly
climbed back to 2.3 percent of travel, he projects that the situation will again deteriorate
as additional service cuts are made (O’Toole 2007).
While O’Toole’s arguments draw attention, extent of transit ridership in Portland relative
to its size is an interesting consideration. As seen in Table 1, with a regional population in
2005 of about 2.175 million people, Portland ranked 23rd in size among major U.S.
metropolitan areas. However, its residents logged more than 104 million annual transit
trips, which qualified for 10th highest among U.S. metro areas (which is 100 million
annual riders more than it had in 1979). Considered in relation to its population size, this
meant that Portlanders averaged 48.1 transit trips per person per year in 2005, which was
7th highest in the country, placing them behind only New York (146.8); Chicago (51.6);
Philadelphia (57.4); Washington, D.C. (78); and Boston (80.1), and only slightly behind
San Francisco (51.9). Its transit use rates are multiples above places like San Diego (9.9),
Dallas (11.9), Houston (16.7), Miami (19.3), and Tampa/St. Petersburg (4.3), while on a
par with older Eastern transit cities like Baltimore (38.7), and Pittsburgh (29.2. Similarly,
its VMT per capita rate—a measure of its auto dependency and the demand its residents
place on the highway system—is 23.6, which is quite favorable in comparison to places
27
like Dallas (31.1), Houston (36.9), and Atlanta (33.8). This efficiency in VMT generation
shows up in congestion delay, as residents of Dallas, Houston, and Atlanta experience
considerably more hours of congestion delay than residents of Portland (figures are
presented in a later section).
When Portland first decided to develop its future growth patterns around its light rail
transit (LRT) system, it had more than its share of critics. One such critic was the town of
Gresham, located between downtown Portland and the airport. While Gresham chose to
distance itself from the transit system during its construction and early operating period,
by the late 1990s it annexed land to incorporate the MAX line within its jurisdictional
borders.
THE PARADOX OF SELF­SELECTION
An important set of arguments challenging the rationale for advocating compact,
traditional land use policies suggests that while persons living in such areas may in fact
drive less and walk more, the reason for this difference in behavior lies more with
individuals and their attitudes toward these opportunities than the areas themselves. They
argue that such individuals may be predisposed to such behavior and seek out
communities in which they can indulge these priorities. Forcing persons without these
predispositions into neighborhoods that favor walking and transit use may not yield the
desired result that these individuals will drive less. This point of view may be largely
attributed to a 1994 study by Kitamura et al., which analyzed travel behavior differences
among five diverse neighborhoods in the San Francisco Bay Area. Attitude surveys,
combined with travel diaries, were used to conclude that residents’ attitudes toward their
neighborhoods and their travel patterns were highly correlated and, in fact, that the
attitudes showed greater statistical significance than the neighborhood characteristics
(Kuzmyak et al. 2003).
A number of researchers have since taken interest in this perspective, including Susan
Handy, Patricia Mokhtarian (part of the 1994 Kitamura team), Xinyu Cao, and Kevin
Krizek—all academic professionals with extensive research backgrounds (Handy 2006;
Handy, Cao, and Mokhtarian 2005; Schwanen and Mokhtarian 2005; and Krizek 2003a).
In addition to the findings of the Kitamura study, an early study by Newman and
Kenworthy (1989) on the correlation between density and gasoline consumption (i.e.,
VMT per capita) for a sample of international cities came under criticism for failure to
account for major underlying factors such as transit availability and income in ascribing
major benefits to higher density. This set of events has made the land use research field,
and academic researchers in particular, extremely zealous about following acceptable
28
Table 1. Population and Transportation Statistics of 23 Largest Metropolitan Areas.
Size
Rank Urban Area
Population1
(thousands)
Annual
Transit
Ridership2
(rank, in
millions)
Annual
Transit
Trips per
Capita
Daily
VMT per
Capita3
Delay per
Peak
Traveler4
(hr/yr)
1 New York, NY 18,815 2,759.8 (1) 146.8 15.5 46
2 Los Angeles, CA 12,875 451.5 (3) 35.1 22.7 72
3 Chicago, IL 9,525 492.3 (2) 51.7 20.5 46
4 Dallas/Ft Worth, TX 6,145 73.3 (15) 11.9 31.1 58
5 Philadelphia, PA 5,827 334.5 (6) 57.4 18.9 38
6 Houston, TX 5,628 94.6 (13) 16.7 36.9 56
7 Miami, FL 5,413 104.7 (9) 19.3 19.2 50
8 Washington, D.C. 5,306 414.1 (4) 78.0 22.9 60
9 Atlanta, GA 5,278 142.4 (8) 26.9 33.8 60
10 Boston, MA 4,482 394.9 (5) 80.1 20.3 46
11 Detroit, MI 4,467 35.6 (25) 8.0 24.1 54
12 San Francisco, CA 4,203 218.2 (7) 51.9 22.4 60
13 Phoenix, AZ 4,179 45.7 (22) 10.9 27.3 48
14 Riverside/S. Bern,
CA
4,081 NA NA 24.5 49
15 Seattle, WA 3,309 98.6 (12) 29.8 25.8 45
16 Minneapolis -St.
Paul, MN
3,208 69.7 (16) 21.7 24.5 43
17 San Diego, CA 2,974 29.3 (27) 9.9 23.7 57
18 St. Louis, MO 2,803 46.4 (21) 16.6 28.7 33
19 Tampa/St. Pete, FL 2,723 11.7 (36) 4.3 22.8 45
20 Baltimore, MD 2,668 103.4 (11) 38.7 21.4 44
21 Denver, CO 2,464 86.3 (14) 35.0 22.1 50
22 Pittsburgh, PA 2,355 70.0 (17) 29.2 22.7 16
23 Portland, OR 2,175 104.5 (10) 48.1 23.6 38
1 Table of U.S. Metropolitan Statistical Areas, U.S. Census Bureau, July 2007.
2 2005 Annual Transit Ridership by Metropolitan Urban Area, Federal Transit Administration,
U.S. Department of Transportation.
3 Our Nation’s Highways, Federal Highway Administration, U.S. Department of Transportation,
2005.
4 Annual Mobility Report, Texas Transportation Institute, 2005.
29
scientific principles regarding causality. Four criteria are cited by Handy as necessary to
prove causality in a relationship:
 Association: The cause and effect are statistically connected.
 Time order: The cause precedes the effect in time.
 Nonspuriousness: No third factor creates an accidental or spurious relationship
between the variables.
 Causal mechanism: The mechanism by which the cause influences the effect is
known.
It has been argued by the self-selection proponents that the first criterion has been
essentially demonstrated, namely that residents of neighborhoods with higher levels of
density, land use mix, transit accessibility and pedestrian friendliness walk more and
drive less than residents of places with lower levels of these characteristics. However,
they point out, most of these studies have reached their conclusions from cross-sectional
data, and while they have controlled for sociodemographic differences among
communities and travelers, they have not accounted for the effects of attitudes toward
travel. Hence, the time order and nonspuriousness criteria have not been addressed,
leaving open the possibility of self-selection in which individuals who would rather not
drive choose to live in neighborhoods that are conducive to driving less (Handy, Cao, and
Mokhtarian 2005). Schwanen and Mokhtarian (2005) used such a cross-sectional
approach, but with a different methodology that incorporated attitudes and found that
neighborhood type did impact travel behavior, even after attitudes were accounted for,
and Cao and Mokhtarian (2005) found that characteristics of the built environment
influenced walking behavior even after accounting for a preference toward walkable
neighborhoods.
In a 2003 APA Journal article, Krizek reported on research to try to address this issue by
taking two important steps: improving the measures of urban form themselves to better
reflect the characteristics of neighborhood accessibility, and opting for a longitudinal as
opposed to cross-sectional approach to studying travel behavior changes in relation to
land use. He employed data from the multiyear Puget Sound regional panel survey in
which he examined changes in the structure, neighborhood, and travel behavior of 430
households that had changed residential location within the region during consecutive
two-year survey intervals. He found that households do, in fact, change their travel
behavior when they are exposed to different urban forms following a move. Models
revealed that in the presence of improved neighborhood accessibility, households
increased their number of daily trip tours (journeys to and from home), but the number of
trips per tour decreased as did both total personal miles of travel and VMT. In other
words, they made more trips, but the trips were shorter, single-purposed, and less likely
to involve auto use (Krizek 2003a).
Handy, Cao, and Mokhtarian obtained similar corroboration of a causal effect from land
use by also applying a quasi-longitudinal approach to data from eight northern California
communities. Four pairs of traditional vs. conventional suburban, demographically
matched neighborhoods were selected from the Sacramento, Modesto, Santa Rosa, and
30
San Jose areas. Roughly 500 residents from each neighborhood were surveyed on their
travel behavior, vehicle ownership, neighborhood characteristics and preferences, and
travel attitudes. Vehicle miles per respondent were found to be 18 percent higher among
residents of the suburban neighborhoods. To sort out the effects of neighborhood
characteristics from attitudes and preferences, a set of multivariate models was estimated
using vehicle miles driven as the dependent variable. In these simple models, when
attitudes were accounted for, no significant effect of built environment was determined.1
However, as a stronger test of causality, a longitudinal methodology was applied to
measure changes among residents who had recently moved. These models revealed
significant associations between changes in travel behavior in response to changes in the
built environment even when attitudes had been accounted for, providing support for a
causal relationship (Handy, Cao, and Mokhtarian 2005).
While no one is prepared to yet proclaim complete satisfaction with the premise that
changing land use will lead to a fully corresponding change in travel behavior (i.e.,
toward more walking and less driving), increasing statistical evidence is being furnished.
An interesting parallel question is whether travelers have enough experience with higher-density,
mixed-use, transit and pedestrian-serviceable land forms to be able to form an
experience-based set of preferences that withstand the test of time and alternative land
use offerings, which are currently in short supply.
DENSITY, CONGESTION, ACCESS, AND MOBILITY
A legitimate concern among critics of compact, mixed-use development patterns is the
effect of higher density on traffic levels and congestion. In a 2003 article, Wendell Cox
argued that one of the principal reasons that smart growth or compact city strategies
cannot reach its objective of reducing traffic congestion (or its rate of growth) is because
of the strong positive relationship between higher population density and higher traffic
volumes. He claimed that as population densities rise, vehicle use also rises and cited
research sponsored by the FHWA (Ross and Dunning 1997) that shows traffic volumes
rising at least 80 percent of the rate of the corresponding increase in population density.
Moreover, he suggested, as more vehicle miles occur in a confined geographical location,
traffic slows down and is subject to more stop-and-go operation, leading to increased
time spent in traffic and higher air pollution emissions since most vehicle tailpipe
pollutants are emitted at higher rates at lower speeds. To illustrate his hypothesis, Cox
fitted data from the Texas Transportation Institute’s (TTI’s) 2000 database to a linear
regression, resulting in a formula that predicts vehicle miles per square mile in relation to
population density, as displayed in Table 2 (Cox 2003).
1It should also be noted, however, that the measures used to capture local and regional land use
and accessibility are extremely important, and have generally not been rigorously applied in
studies such as these.
31
Table 2. Effect of Higher Density on Vehicle Travel.
Population/
Square Mile
Vehicle
Miles/
Square Mile
Compared to
Density of
1,000 per
Square Mile
Vehicle
Miles per
Person
Compared to
Density of 1,000 per
Square Mile
5000 93,069 2.90 18.6 0.58
4000 77,835 2.42 19.5 0.61
3000 62,601 1.95 20.8 0.65
2000 47,367 1.47 28.7 0.89
1000 32,133 1.00 32.1 1.00
Source: Cox, W., “How Higher Density Makes Traffic Worse,” 2003.
Cox’s model predicts that as population density per square mile increases, so does vehicle
travel intensity in terms of vehicle miles per square mile. Indeed, at a density of 5000
persons per square mile, the number of vehicle miles per square mile would be 93,069,
which is a multiple of 2.9 times the rate of 32,133 vehicle miles per square mile at a
population density of 1000. However, increasing population density by a factor of five
only increases VMT density by 2.9, implying an inherent efficiency in the higher
population density. The two columns on the right have been calculated from the first two
columns to highlight how the average person in the higher-density environment (5000
persons/mi2) generates only 18.6 VMT, vs. 32.1 in the low-density environment, a VMT
per capita savings of about 43 percent. This represents a considerable savings in highway
construction and maintenance requirements to taxpayers, but also translates into less
travel delay for users.
To be sure, if 2.9 times the number of vehicle trips were squeezed onto the same highway
network, congestion levels would probably rise rapidly, given the nonlinear nature of
traffic flow as volumes approach design capacity. However, there is no accounting in
Cox’s analysis as to what the actual traffic congestion levels would be since he did not
account for the corresponding highway capacity. Nor did he account for the presence of
transit in diverting some of these trips or for higher rates of walking that would allow
people to reach desired activities independent of the number of cars on the roads. The
relationship between metropolitan population density and annual hours of delay per peak
hour traveler for several metropolitan areas—New York, NY (4313 persons/mi2 and 46
hours delay); Portland, OR (2853 persons /mi2, 38 hours); Dallas (2188 persons /mi2, 58
hours); Atlanta (1694 persons /mi2, 60 hours); Houston (1618 persons /mi2, 56 hours);
and Phoenix (2028 persons /mi2, 48 hours)—suggest that numerous mitigating factors
beyond simply density contribute to predictions of traffic congestion, delay, and the
quality of the travel experience.
Density itself, without attention to mix, balance, and connectivity, could very well create
nightmarish traffic. Hence, construction of high-density employment in one location,
high-density commercial activity along an arterial highway, and multifamily housing
development with no services nearby is probably a recipe for traffic disaster. However, if
the uses are mixed, if distances are compact, and if connections are pedestrian-friendly,
32
the literature shows such developments will internalize a higher proportion of their trips,
relieving the road network of large amounts of VMT. Still, these areas with purposely
designed compact, higher-intensity development may well be locations of higher traffic.
If they are nodes on a system of arterial highways, the higher rates of activity in these
places most likely will slow down traffic. The question, however, is whether that is
uniformly a bad thing. If the transportation objective is to move as many cars as fast as
possible, regardless of trip length or orientation, then the slowdown is judged a bad thing.
However, if the objective is to allow as many people as possible to access as many
activities as meet their needs in as little time as possible, then the occasional loss of free-flow
traffic conditions may be a productive compromise. For those travelers making
longer-distance trips that would be inconvenienced by higher congestion in the activity
nodes, provision of good regional transit service can begin to offset this inconvenience by
providing choices. This vision is rooted in the concept of accessibility, which is
increasingly viewed as a more desirable and more achievable goal than mobility as
defined by private auto travel. From the standpoint of economics, a much higher level of
social welfare is achieved when more people are able to maximize their activity needs at
lower cost (time and monetary).
For example, in Figure 6, Household A resides in an area that is more compact and
pedestrian-oriented with various nearby services, while Household B resides in a more
typical residential subdivision where there are no nearby services. Household B has
access to two large supermarkets within 10 to 15 minutes drive from home. Household A
has access to one of these major supermarkets (a slightly longer drive), but also has
access to a smaller supermarket within three-quarters of a mile of home to which
residents can walk in 15 minutes or drive within five minutes. In addition, a small
neighborhood grocery store, a bakery, and a 16-hour convenience store are within easy
walking distance of home (one-eighth to one-quarter mile). Household A has greater
accessibility as well as more choices and amenity, than Household B. Moreover,
Household A has indemnity against traffic congestion delay. Household A may wish to
do its major shopping at the large supermarket, but at busy travel times (or over time as
traffic levels rise) that destination may be much less attractive than the smaller or more
specialized options within the neighborhood.
In 2007, researchers at the University of Minnesota’s Center for Transportation Studies
demonstrated why accessibility may be the most appropriate lens for viewing the
performance of the transportation system. The report begins by suggesting a different
way of looking at the annual congestion indices published by TTI, arguing that while
congestion is a serious issue, counting cars and clocking speeds fails to tell the whole
truth about land use and transportation relationships. Using data from the Twin Cities,
they note that while the Twin Cities is not at the top of the national list for traffic
congestion, traffic is getting worse and delays doubled during the 1990s. However,
during this same time, the number of workers and the number of jobs reachable within
30, 45, and 60 minutes increased in almost all of the TAZs studied. This increase was
attributed to jobs moving closer to workers and vice versa such that commuting times
went up by no more than 5 minutes. In accord with this finding, researchers observed that
33
Figure 6. Accessibility Benefits from Compact, Mixed Land Use.
an explosion of townhouses and condos in urban centers over the past decade has brought
many new residents into activity-rich TAZs and dramatically increased the number of
destinations that are easily accessible. This occurs, they point out, despite the fact that the
density of people and activities ensures