Content uploaded by John Luciano Renne
Author content
All content in this area was uploaded by John Luciano Renne
Content may be subject to copyright.
49
Transportation Quarterly, Vol. 57, No. 3, Summer 2003 (49–77).
© 2003 Eno Transportation Foundation, Inc., Washington, DC.
This is the fourth in a series of articles
for Transportation Quarterly analyz-
ing urban travel trends and differences
in travel behavior among a range of socioe-
conomic groups.1We examine the 2001
National Household Travel Survey (NHTS),
which was released in January 2003. Our
focus is on interrelated variations in motor
vehicle ownership, mobility levels, means of
transportation (travel mode), trip distance,
time of day of travel, and purpose of travel
as these dimensions of travel behavior vary
by income group, ethnic and racial group,
sex, and age. We compare the results of the
2001 NHTS with those of its predecessor,
the Nationwide Personal Transportation
Survey (NPTS), in 1969, 1977, 1983, 1990,
and 1995.
The most salient trend in American travel
behavior over the past four decades has been
increased reliance on the private car for
urban travel, with corresponding declines in
public transit and walking. The journey-to-
work portion of the US Census, for exam-
ple, reports that the percentage of work trips
made by public transit fell from 12.6% in
1960 to only 4.7% in 2000 (see Table 1).
The share of walk trips fell from 10.3% to
only 2.9%. Conversely, the private car’s
share of work trips rose from 66.9% to
87.9%.2Similarly, the series of NPTS and
NHTS surveys, which also include nonwork
Socioeconomics of Urban Travel:
Evidence from the 2001 NHTS
The 2001 National Household Travel Survey (NHTS) confirms most of the same travel trends
and variations among socioeconomic groups documented by its predecessors, the Nation-
wide Personal Transportation Surveys (NPTS) of 1969, 1977, 1983, 1990, and 1995. The pri-
vate car continues to dominate urban travel among every segment of the American population,
including the poor, minorities, and the elderly. By comparison, public transport accounts for
less than 2% of all urban travel. Even the lowest-income households make only 5% of their
trips by transit. The most important difference in the 2001 NHTS is the doubling in modal
share of walk trips in cities, due to a much improved survey technique that captured previously
unreported walks.
While the private car dominates travel, there are important variations in auto ownership
and travel behavior by income, race, ethnicity, sex, and age. Overall, the poor, racial and
ethnic minorities, and the elderly have much lower mobility rates than the general popula-
tion. Moreover, the poor, blacks, and Hispanics are far more likely to use transit than other
groups. Indeed, minorities and low-income households account for 63% of the nation’s
transit riders. Different socioeconomic groups also have different rates of carpooling, taxi
use, bicycling, and walking. In addition, they travel different distances and at different times
of day. Many of these socioeconomic variations in travel behavior have important conse-
quences for public policy.
by John Pucher and John L. Renne
50
TRANSPORTATION QUARTERLY / SUMMER 2003
trips, show that Americans have been relying
increasingly on the car for all their travel
purposes, not just for the journey-to-work
(see Table 2). Thus, the auto’s share of daily,
local travel rose from 81.8% of trips in 1969
to 86.4% in 2001, while public transit’s
share fell from 3.2% to 1.6% over the same
period.3
Corresponding to that increased reliance
on the automobile, motor vehicle ownership
is now almost universal in the United States,
with 91.7% of American households owning
at least one motor vehicle in 2001, and
58.5% of households owning two or more
vehicles.4Indeed, the total number of motor
vehicles per household rose from 1.2 in 1969
to 1.9 in 2001, and the number of motor
vehicles per licensed driver rose from 0.7 to
1.1.5Yet further confirming this growing
auto availability, the total number of autos
and light trucks per 1,000 persons rose from
340 in 1960 to 766 in 2001, giving the USA
by far the highest rate of personal vehicle
ownership in the world, about 50% higher
than in most Western European countries.6
While these aggregate statistics confirm
the extreme auto dependence of American
cities, they mask important variations by
region of the country, by city size, and
among socioeconomic groups. There are
important differences in travel behavior by
income, age, sex, race, and ethnicity. Motor
vehicle ownership, mobility rates, means of
transport, trip distance, trip purpose, and
time of day of travel vary from one group to
another. Such differences can be crucial in
designing equitable transport policies at all
government levels.
For example, peak-hour congestion pric-
ing on roadways and off-peak discounts for
transit should take into account the income
differences of travelers by time of day. Simi-
larly, the regressivity of financing transporta-
tion through gasoline taxation, roadway
tolls, transit fares, and user charges of any
sort depends on the income distribution of
Table 1: Trends in Modal Split for the Journey-to-Work (1960 - 2000)
(percentage of work trips by means of transportation)
Mode of Transportation Census Year
1960 1970 1980 1990 2000
Total Auto 66.9 77.7 84.1 86.5 87.9
SOV na na 64.4 73.2 75.7
HOV na na 19.7 13.4 12.2
Public Transit 12.6 8.9 6.4 5.3 4.7
Walk 10.3 7.4 5.6 3.9 2.9
Bicycle na na 0.5 0.4 0.4
Work at Home 7.5 3.5 2.3 3.0 3.3
Other 2.6 2.5 1.1 0.9 0.8
All 100 100 100 100 100
Source: US Decennial Census, Supplemental Survey: Journey-to-Work, various census years, 1960 to 2000, as tabulated by Alan
Pisarski and reported in A. Pisarski, Commuting in America III. Washington, DC: Eno Transportation Foundation, forthcoming in
2003.
Note: Only the 1960 Census work trip survey included a category called “not reported,” which accounted for 4.3% of all 1960
responses. To make the 1960 distributions comparable with those of later years, which do not include an “unreported” category,
the 1960 reported modal shares were scaled up by a factor of 1.045 so that their total would equal approximately 100%.
travelers across different means of transport,
trip distances, locations, and times of day of
travel. On the benefit side, the equity impacts
of subsidy expenditures depend on variations
in socioeconomic characteristics of travelers
along those same dimensions of travel
behavior. The extent to which the poor ben-
efit from transit subsidies depends on the
degree to which they actually use the specif-
ic type of transit being subsidized. Disaggre-
gation of travel statistics also helps identify
groups suffering from low mobility and may
suggest the most effective approaches to rem-
edying their inadequate accessibility to trans-
port services.
The 2001 NHTS
The National Household Travel Survey was
conducted for the first time in 2001 and
replaces the Nationwide Personal Trans-
portation Survey for daily travel and the
American Travel Survey (ATS) for long-dis-
tance travel. Since this article deals exclu-
sively with urban travel, we focus on the
daily trip portion of the NHTS and compare
that part of the 2001 survey with the former
NPTS surveys of 1969, 1977, 1983, 1990,
and 1995. While the decennial Census pro-
vides information for the journey to work
(less than a fifth of all trips), the NPTS and
NHTS surveys are the only sources of com-
prehensive, nationwide data on trips for all
purposes. Similar to the NPTS surveys, the
NHTS reports a wide range of information
about the socioeconomic characteristics of
households, as well as their motor vehicle
ownership and many aspects of their travel.
For example, it reports the number of trips
per day and, for each trip, the means of trav-
el, day and time of travel, trip distance, and
trip purpose.
The 2001 NHTS was funded and coordi-
nated by the US Department of Transporta-
tion (Federal Highway Administration,
Bureau of Transportation Statistics, and the
National Highway Traffic Safety Adminis-
tration). Two private firms, however, actual-
SOCIOECONOMICS OF URBAN TRAVEL
51
Table 2: Trends in Modal Split for Daily Travel in the United States (1969-2001)
(percent of trips by transport mode, all trip purposes)
Mode of Transportation 1969 (1) 1977 1983 1990 1995 2001
Auto281.8 83.7 82.0 87.1 86.5 86.4
Transit 3.2 2.6 2.2 2.0 1.8 1.6
Walk2na 9.3 8.5 7.2 5.4 8.6
Bicycle na 0.7 0.8 0.7 0.9 0.9
Other35.0 3.7 6.5 3.0 5.4 2.5
Source: Federal Highway Administration, Nationwide Personal Transportation Surveys 1969, 1977, 1983, 1990, and 1995; and
National Household Travel Survey, 2001.
Note: Unlike all subsequent tables, these NPTS and NHTS modal split percentages are for daily, local travel in aggregate for the
entire USA, both urban and rural, as reported by the FHWA in its own NPTS and NHTS reports. Our own tabulations, from Table
3 onward, include only local trips in urban areas.
1. The 1969 NPTS did not sample walk and bike trips, thus artificially inflating the modal split shares of the motorized modes
compared to the NPTS surveys in later years. To ensure some degree of comparability, we adjusted downward the reported
motorized shares of trips in 1969 by 10%, using the percentage of walk and bike trips in 1977. That is why the column adds
to 90% and not 100%. Our adjustment is rough, but otherwise, the 1969 and later NPTS modal split distributions would be
completely incomparable.
2. The decrease in auto mode share from 1995 to 2001, and the corresponding increase in walk mode share during the same
period, are due to a change in sampling methodology that captures previously unreported walk trips.
3. The “other” categories includes mainly school bus trips, which account for roughly 2 -3% of all trips in each of the survey
years. It also includes taxicabs, ferries, airplanes, and helicopters.
52
TRANSPORTATION QUARTERLY / SUMMER 2003
ly conducted the survey through telephone
interviews: Westat (Rockville, MD) and Bat-
telle/Morpace (Farmington Hills, MI).
The 2001 NHTS incorporates several
important improvements in survey method-
ology, just as the 1995 NPTS had greatly
improved over earlier NPTS surveys. For
example, walk trips had been significantly
underreported in all earlier surveys. Thus, the
2001 NHTS included several special prompts
in the survey questionnaire to ensure that all
walk trips were reported. Moreover, because
earlier surveys had reported some question-
able trip lengths, multiple data collection
methods were used to achieve more accurate
trip distances. The 2001 survey also collected
more detailed information on trips made to
access transit services.
Of course, the NHTS suffers from all the
problems of telephone surveys. Most impor-
tantly, it undersamples low-income house-
holds without telephones. To correct that
problem, survey responses were weighted to
make the overall sample representative of the
population as a whole. Indeed, the weighting
of undersampled households in the 2001
NHTS was more extensive than in any pre-
vious survey. The NHTS does not, however,
take into account the increasing number of
households with only cellular phones that
cannot be reached by standard telephone
survey techniques.
The 2001 NHTS was conducted over the
14-month period from March 2001 to May
2002, thus ensuring coverage of every month
of the year. Unfortunately, that timing turned
out to be problematic due to the September
11, 2001 terrorist attacks on the World
Trade Center in New York City and the Pen-
tagon in Washington, DC. The attacks dis-
rupted transport services for months, espe-
cially curtailing long-distance travel. It is not
certain what impacts the attacks had on
urban travel, but it seems likely that both the
amount of travel and modal choice were
affected. That may have distorted the survey
results to some unknown extent.
As with the earlier NPTS surveys, the
NHTS only includes the civilian, noninstitu-
tionalized population of the United States. It
explicitly excludes motels, hotels, prisons,
military barracks, convents, monasteries,
and any living quarters with 10 or more
unrelated occupants. The NHTS included
college students, however, provided that dor-
mitory, fraternity or sorority rooms had tele-
phones and fewer than 10 occupants. The
2001 survey interviewed 25,721 households
nationwide, but we analyzed the responses
of only the 19,768 households living in
urban areas. We further restricted our analy-
sis to urban travel by eliminating all trips
over 75 miles. The resulting sample includ-
ed 173,974 urban trips (out of 248,517 total
trips for the entire NHTS sample). Our
analysis of the NHTS, therefore, varies from
other studies that examine the entire sample,
including nonurban households and trips.
Impact of Trip Purpose on Modal Choice
As already noted in Tables 1 and 2, public
transit has been serving a declining percent-
age of all trip purposes, but its share of work
trips has been consistently higher than for
nonwork trips. That is evident not only from
comparing the journey-to-work data from
the Census (Table 1) with the NPTS all-pur-
pose data (Table 2), but also from disaggre-
gating the NHTS data by trip purpose, as in
Table 3. It shows that transit served 3.7% of
all work trips in 2001, compared to 1.4% of
shopping trips, 1.0% of social and recre-
ational trips, and 2.2% of school and church
trips. The rail transit modes are especially
focused on the work trip.
Single occupant auto use (SOV) is the pre-
dominant choice for the work trip, account-
ing for 75.4% of all journeys to work.
Carpooling—via high occupancy vehicle
(HOV)—is much more prevalent, however,
for all other trip purposes, accounting for
over half of such trips. Family members are
often passengers on car trips for shopping,
53
SOCIOECONOMICS OF URBAN TRAVEL
recreation, church, and school, while they sel-
dom accompany each other to work.
Walking and bicycling are most used for
social and recreational trips and for trips to
school. Nonmotorized transportation is used
much less for work trips, probably due to the
longer length of work trips and the need to
minimize travel time. Likewise, few travelers
rely on walking or cycling for shopping,
probably because those modes are not well
suited to carrying packages. Moreover, most
shopping facilities are now located far from
residential neighborhoods, no longer within
walking or cycling distance for most house-
holds.
Regional Variations in Transit Use,
Walking, and Cycling
The nationwide aggregate statistics shown in
most tables in this article hide the enormous
variation in travel behavior from one region
of the country to another. As shown in Table
4, the most transit-oriented region, the Mid-
Atlantic, has a transit modal share that is 15
times higher than in the least transit-oriented
region, the East South Central (5.8% vs.
0.4%). The Pacific and New England
regions follow the Mid-Atlantic region in
order of their transit shares (2.2% and
1.8%, respectively).
Table 3: Variation in Modal Choice by Trip Purpose
(percentage of trips by means of transportation)
Mode of Transportation Trip Purpose
Work and Shopping and Social and School and
Work Related Services Recreation Church
Total Auto 92.1 91.5 84.1 72.9
SOV175.4 38.4 27.6 17.1
HOV216.8 53.2 56.6 55.9
Total Transit 3.7 1.4 1.0 2.2
Bus and Light Rail32.1 1.2 0.7 1.8
Metro/Subway/Heavy Rail41.1 0.1 0.3 0.4
Commuter Rail50.5 0.0 0.0 0.0
Total Nonmotorized 3.9 6.8 14.0 11.2
Walk 3.4 6.5 12.7 10.5
Bicycle 0.5 0.3 1.3 0.7
School Bus 0.1 0.0 0.2 13.6
Taxicab 0.1 0.1 0.1 0.1
Other 0.1 0.2 0.5 0.1
All 100 100 100 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
54
TRANSPORTATION QUARTERLY / SUMMER 2003
Regional variations in walking are also
striking, and strongly correlated with transit
modal share. Thus, the highest walk modal
share is also in the Mid-Atlantic states
(15.8%), followed by the Pacific region
(10.6%), and New England (10.3%). Con-
versely, the lowest shares of walk trips are in
the East South Central (6.0%) and the West
South Central (6.3%). The correlation
between transit use and walking is probably
due to the more walkable, compact urban
form in transit-oriented cities and the cru-
cial role of walking to access transit stops.
Bicycling has a somewhat different
regional pattern, with the highest level in the
Pacific (1.1%), but roughly the same levels in
the rest of the country (0.7% to 0.9%),
except for the East South Central, which has
a much lower level (0.4%). Thus, the East
South Central has the lowest levels of transit
use, walking, and cycling, and is the most
dependent on the auto for all travel.
Impact of Income on Travel Behavior
Just as with the 1995 NPTS, the 2001 NHTS
shows a striking increase in travel with
increased income levels. We have altered the
income categories in 2001 to account for
inflation and the shifting distribution of
households to higher income levels. Never-
theless, the impact of income on daily trip
frequency and mileage covered is virtually
the same for both surveys. Thus, households
with incomes less than $20,000 a year made
an average of 3.2 trips per person, per day
in 2001 compared to 4.8 trips per day for
households with incomes of $100,000 or
more (see Table 5). Not only do higher-
income households make more trips per day,
but they also make longer trips, covering
almost twice the total mileage per day of
low-income households (31.8 miles vs. 17.9
miles per person, per day).
The much lower mobility rates of the low-
income households might be interpreted as
a basic inequity in our urban transportation
system. Clearly, many low-income house-
holds are cut off from some destinations they
need to reach because they cannot afford the
automotive transportation needed to access
most parts of metropolitan areas. That is
especially serious in the case of inaccessible
job sites, since poverty is thus directly per-
petuated. Moreover, inability to reach med-
ical, educational, training, shopping, and
recreational facilities can also seriously
impair the quality of life of poor households.
Table 4: Regional Variations in Modal Shares for Transit, Walking, and Bicycling
(percentage of trips by transit)
East West East West
Mode of New Middle North North South South South
Transportation England Atlantic Central Central Atlantic Central Central Mountain Pacific
Total Transit 1.8 5.8 1.3 0.6 1.6 0.4 0.7 0.8 2.2
Bus and Light Rail 0.7 3.0 0.9 0.5 1.2 0.4 0.7 0.8 2.0
Metro/Subway/
Heavy Rail 0.9 2.3 0.2 0.0 0.3 0.0 0.0 0.0 0.1
Commuter Rail 0.3 0.5 0.2 0.0 0.1 0.0 0.0 0.0 0.1
Total Nonmotorized 11.0 16.7 9.5 7.3 8.5 6.4 7.1 9.5 11.7
Walk 10.3 15.8 8.6 6.6 7.6 6.0 6.3 8.7 10.6
Bicycle 0.7 0.8 0.9 0.7 0.9 0.4 0.8 0.8 1.1
Source: Calculated by the authors from the 2001 NHTS.
55
SOCIOECONOMICS OF URBAN TRAVEL
To some extent, however, the lower mobility
of low-income households reflects their high-
er rates of unemployment and retirement,
and thus fewer trips to work. Their shorter
trip lengths might also result from the con-
centration of the poor in central cities, where
things are closer together and do not require
such long trips as in the suburbs.
As expected, the rate of auto ownership
rises with increasing household income (see
Table 6 and Figure 1). While 26.5% of
households with incomes less than $20,000
have no motor vehicle at all, only 5.0% of
households in the next highest income cate-
gory ($20,000 to $39,999) have no motor
vehicle. Only 1.2% of households with
incomes over $75,000 have no motor vehi-
cle. Thus, by far the largest jump in auto
ownership comes at the low end of the
income scale. A car is obviously one of the
first purchases households make as soon as
they can, even if it strains their already lim-
ited budgets. Indeed, it is probably unique
to the United States that three-fourths of
even its poorest households own a car. That
reflects the extent to which the car has
become a virtual necessity for even the most
basic transportation needs in most Ameri-
can metropolitan areas.
Similarly, the rate of multiple car owner-
ship increases with income. Thus, the per-
centage of households with two or more cars
increases from 25.2% in the under $20,000
category to 50.9% in the $20,000 to
$39,999 category and 87.8% in the
$100,000 and over category. The percentage
of households with three or more cars
increases from 7.7% in the under $20,000
category to 15.3% in the $20,000 to
$39,999 category and 38.5% in the
$100,000 and over category. The sharp
increase in multiple car ownership with
increased income is fully expected, and is
also consistent with all earlier NPTS surveys.
Increased income obviously makes cars more
affordable. Moreover, there is a positive cor-
relation between income and household size
in the NHTS sample, so higher-income
households also have more cars because they
are larger. Nevertheless, even 7.7% of low-
income households reported owning three or
more cars, which seems a bit surprising. That
might reflect underreported incomes or sub-
stantial assets of retired households with low
current incomes.
Income is the primary determinant of
auto ownership, which, in turn, is the main
determinant of modal choice. As shown in
Table 7, the ownership of even one car dra-
matically transforms travel behavior. Thus,
transit use drops from 19.1% of trips by
households with no car to only 2.7% of trips
by households with one car. Equally striking,
walk trips fall from 41.1% of trips by house-
holds with no car to only 12.5% of trips by
households with one car. Bike trips fall from
Table 5: Daily Travel per Capita by Income Class
Household Income Trips per Day, per Person Miles Traveled per Day, per Person
Less than $20,000 3.2 17.9
$20,000 to $39,999 3.9 26.4
$40,000 to $74,999 4.2 30.2
$75,000 to $99,999 4.3 30.7
$100,000 and over 4.8 31.8
All 4.0 26.9
Source: Calculated from the 2001 NHTS by Mary Ann Keyes, Federal Highway Administration, US Department of Transportation.
Note: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
56
TRANSPORTATION QUARTERLY / SUMMER 2003
Table 6: Vehicle Ownership by Income Class
(percentage distribution within each income class)
Household Income
Vehicles Less than $20,000 to $40,000 to $75,000 to $100,000
Per Household $20,000 $39,999 $74,999 $99,999 and over All
026.5 5.0 2.3 0.9 1.5 8.3
148.3 44.1 26.8 13.1 10.7 33.2
217.5 35.6 45.6 50.6 49.3 37.4
3 or more 7.7 15.3 25.3 35.4 38.5 21.1
Total 100 100 100 100 100 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: The sample was limited to residents of urban areas. Vehicles include passenger cars, as well as station wagons, passen-
ger vans, sport-utility vehicles, pickup trucks, light trucks, motorcycles, mopeds, and recreational vehicles. This data include only
residents of urban areas and urban clusters.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Household Income
Less than
$20,000
$20,000 to
$39,999
$40,000 to
$74,999
$75,000 to
$99,999
$100,000
and over
All
3 or more cars
2 cars
1 car
no car
Figure 1: Vehicle Ownership by Income Class
(percentage of households in each income class)
Source: See Table 6.
57
2.4% to 0.7% of all trips. And taxi trips fall
from 1.0% to 0.2% of all trips. Subsequent
increases in auto ownership to two, three, or
more cars per household have relatively
minor additional impacts on travel behav-
ior, although they further decrease transit
use, walking, and cycling, as expected. Thus,
households with three or more cars make
only 0.5% of their trips by transit, 6.3% by
walking, 0.8% by bicycle, and 0.1% by taxi.
These patterns mirror those in the 1995
NPTS and roughly conform to expectations.
Both surveys find considerable auto use even
among households with no cars: 34.1% of
all trips in 2001 and 29.6% of all trips in
1995. Most of those auto trips are reported
as passengers in someone else’s car (for
HOV), but 5.2% were made as drivers in
2001 (vs. 5.7% in 1995).7That can only be
explained as the result of renting cars or bor-
rowing them from neighbors, friends, or rel-
atives who own cars.
The bad news for transit in Table 7 is that
most households abandon public transporta-
tion as soon as they own their first car. The
doubling of auto ownership per capita since
1960 is surely one of the most important rea-
sons for the steady decline in transit’s modal
share, as shown in Tables 1 and 2. The
already high and still rising level of auto
SOCIOECONOMICS OF URBAN TRAVEL
Table 7: Impact of Auto Ownership on Mode Choice
(percentage of trips by means of transportation)
Mode of Total Number of Vehicles in Household
Transportation 0 1 2 3 or more All
Total Auto 34.1 81.9 88.8 90.5 85.9
SOV15.2 36.8 36.6 42.5 37.3
HOV228.9 45.1 52.2 48.0 48.6
Total Transit 19.1 2.7 0.6 0.5 1.7
Bus and Light Rail314.1 1.9 0.4 0.3 1.2
Metro/Subway/Heavy Rail44.8 0.7 0.1 0.1 0.4
Commuter Rail50.2 0.2 0.1 0.1 0.1
Total Nonmotorized 43.5 13.2 8.8 7.1 10.4
Walk 41.1 12.5 7.8 6.3 9.5
Bicycle 2.4 0.7 0.9 0.8 0.9
School Bus 1.5 1.7 1.4 1.4 1.5
Taxicab 1.0 0.2 0.1 0.1 0.1
Other 0.9 0.3 0.4 0.3 0.4
All 100 100 100 100 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
58
TRANSPORTATION QUARTERLY / SUMMER 2003
ownership in the United States will remain a
strong deterrent to transit use in the coming
years.
Walking and cycling plummet with
increasing car ownership (from 43.5% to
7.1% of all trips), thus depriving people of
much needed exercise. With 64% of Ameri-
cans overweight in 2001, and 31% obese,
leading medical and public health journals
have explicitly advocated more walking and
cycling for daily travel as the most affordable,
feasible, and dependable way for Americans
to get the additional exercise they need.8Sim-
ilarly, the US Surgeon General specifically rec-
ommends more walking and cycling for prac-
tical, daily travel as an ideal approach to
raising physical activity levels.9The availabil-
ity of cars appears to present an almost irre-
sistible temptation to drive instead of walking
or cycling, even for short trips. Walking in
European cities has also declined over the
past few decades as auto ownership levels
have risen, and obesity levels are now rising
there as well, although they are only about a
third of American obesity rates.10
Unfortunately, the large increase in walk
trips registered by the 2001 NHTS is proba-
bly not due to actual increases in walking. As
already noted, there was a significant
improvement in the survey questionnaire to
capture the many walk trips not reported by
the earlier NPTS surveys. While the share of
trips by walking in 2001 seems realistic, the
jump from 5.5% in 1995 to 9.5% in 2001
(as seen in Table 2) is exaggerated, since pre-
vious surveys were so defective in their sam-
pling of walk trips. The slight decline in auto
modal share reported from 1995 to 2001 is
also artificial, since the new sampling proce-
dure for walk trips considerably raised the
number of total nonauto trips.
Table 8 shows the total impact of income
on choice of travel mode, thus reflecting both
its indirect impact via auto ownership and its
direct impact through the overall need to
travel and its correlation with employment.
It also reflects the tendency of higher-income
households to live in auto-dependent sub-
urbs, where cars are necessary to reach
almost all destinations. As expected, auto use
rises with income, but the only increase in
the auto’s share of travel is from the poorest
to the next higher income class (from 75.9%
to 87.3% of all trips). With subsequent
increases in income, there is virtually no
additional increase in auto modal split share.
Moreover, even the poorest households are
only slightly less likely than affluent house-
holds to make their trips as drivers instead of
as passengers in cars.
Just as Table 6 indicates that roughly
three-fourths of the poorest households own
at least one car, Table 8 shows that roughly
three-fourths of their trips are by car. Thus,
the automobile is the primary mode of trav-
el not only of the affluent but also of the
poor. Perhaps most surprising is that only
4.6% of the trips made by the lowest-income
households are by any form of public tran-
sit. Indeed, the poor use cars 17 times more
than transit for their urban trips (75.9% vs.
4.6%). Although the expense of owning,
insuring, and operating a car unquestionably
strains the limited budgets of poor house-
holds, they are left with virtually no alterna-
tive to the automobile. America’s polycentric,
sprawling metropolitan areas force almost all
households to own and use cars to reach
most destinations. In addition, transit sys-
tems often neglect the special travel needs of
low-income households. Indeed, several stud-
ies suggest that low-income neighborhoods
suffer from inferior service, excessively high
fares, overcrowding, and routes that do not
match their desired trip patterns.11
While transit use generally declines with
increased income, there are large and impor-
tant variations by type of transit. Bus usage,
in particular, plummets as incomes rise.
Thus, the poor are eight times as likely as the
affluent to take the bus (4.0% vs. 0.5% of
trips). In sharp contrast, the affluent are
three times more likely than the poor to take
suburban rail (0.3% vs. 0.1% of trips).
59
SOCIOECONOMICS OF URBAN TRAVEL
Bridging these two extremes, metro services
have a rather bipolar distribution of riders,
with usage concentrated most among the
poor and the affluent, but including many
riders in the middle-income classes as well.
The metro’s modal split share falls from
0.6% among the poor to 0.3% among the
middle class and then rises to 0.7% among
the most affluent.
These differences in rider incomes among
transit modes are due to many factors. Most
importantly, suburban rail tends to serve
long trips from high-income suburbs to well
paying jobs in the downtowns of major met-
ropolitan areas. Suburban rail can sometimes
outperform the automobile by offering
faster, more comfortable, more dependable,
and less stressful peak-hour travel, thus
attracting even affluent passengers. Bus trips
are generally shorter, slower, and less com-
fortable, and they focus more on local trips
within central cities. Since they also suffer
from an image of low-quality, lower-class
service, buses rarely compete with the auto-
mobile among affluent travelers. The excep-
tions are a few specific markets such as
express services to large downtowns.
Metro services appear to serve the broad-
Table 8: Modal Split by Income Class (percentage of trips by means of transportation)
Household Income
Mode of Less than $20,000 to $40,000 to $75,000 to $100,000
Transportation $20,000 $39,999 $74,999 $99,999 and over All
Total Auto 75.9 87.3 88.1 87.4 86.9 85.9
SOV130.0 37.9 39.2 38.6 37.9 37.3
HOV245.9 49.5 48.9 48.7 49.0 48.6
Total Transit 4.6 1.4 1.1 0.9 1.5 1.7
Bus and Light Rail34.0 1.0 0.7 0.5 0.5 1.2
Metro/Subway/
Heavy Rail40.6 0.3 0.3 0.3 0.7 0.4
Commuter Rail50.1 0.0 0.1 0.2 0.3 0.1
Total Nonmotorized 17.0 9.7 9.0 9.4 9.5 10.4
Walk 16.2 8.8 8.1 8.5 8.7 9.5
Bicycle 0.9 0.9 0.9 0.9 0.8 0.9
School Bus 1.9 1.3 1.4 1.5 1.4 1.5
Taxicab 0.2 0.1 0.1 0.2 0.3 0.1
Other 0.3 0.2 0.4 0.6 0.4 0.4
All 100 100 100 100 100 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
60
TRANSPORTATION QUARTERLY / SUMMER 2003
est spectrum of the population, partly
because they are an essential way to get
around for almost everyone in New York
City, which accounts for half of the nation’s
metro riders, and thus dominates all nation-
al statistics on metro usage.12 High metro
usage by affluent riders might be partially
attributed to high-income households in
exclusive or recently gentrified inner city
neighborhoods. In addition, many new
metro systems, such as those in Washington,
DC and San Francisco, provide services com-
parable to suburban rail, extending far out
to affluent suburban communities. Subsi-
dized free or low-cost parking provided at
outlying stations further encourages use by
relatively affluent commuters. At the other
end of the spectrum, metro use by the poor
can be attributed to the many low-income
households living in inner city neighbor-
hoods within the service area of most metro
systems. The income distribution of metro
riders is bimodal not only in New York City
but also in Boston, Washington, DC, Chica-
go, and virtually every other major city with
a metro system.
Recent studies indicate that neighbor-
hoods around some rail transit stations have
been gentrifying, attracting increasing num-
bers of affluent households. As a result,
Figure 2: Differences Among Income Classes in Modal Shares of Public Transport,
Walking, and Cycling (percent of trips by type of transport, all trip purposes)
Household Income
Less than
$20,000
$20,000 to
$39,999
$40,000 to
$74,999
$75,000 to
$99,999
$100,000
and over
All
0%
5%
10%
15%
20%
25%
Bicycle
Walk
Public tranportation
Source: See table 8.
61
property values near such stations have risen
significantly.13 Indeed, low-income house-
holds can no longer afford the rising hous-
ing costs near some rail stations, forcing
them to move to areas with less transit acces-
sibility. The gentrification of working class
neighborhoods has helped revitalize many
inner cities and older suburbs, while increas-
ing transit use among the affluent. Unfortu-
nately, it has reduced the accessibility of low-
income households to rail transit, and
appears to have lessened their use of both
metro and commuter rail.
For example, households in the highest
income group in the 2001 NHTS ($100,000
and over) made 1.0% of their trips by metro
and commuter rail, while the highest income
group in the 1995 NPTS ($80,000 and over)
made only 0.7% of their trips by rail transit.
By comparison, the lowest income group in
the 2001 NHTS (under $20,000) made only
0.7% of their trips by rail transit, consider-
ably less than the 1.2% rail transit share of
the lowest income group in the 1995 NPTS
(under $15,000).14 That suggests that metro
and commuter rail use has been increasing
among the affluent but declining among the
poor.15 Although the 1995 and 2001 income
categories are not exactly comparable, the
31.5% increase in per capita income in the
USA during those years make the income
brackets roughly equivalent.16
Government intervention may be neces-
sary to ensure the affordability of transit-
accessible housing for poor and working
class households. For example, Fannie Mae’s
location-efficient mortgage program, which
focuses on neighborhoods near transit stops,
might be further expanded and targeted
more toward low-income households.17
Large differences in transit rider incomes
are important for public policy purposes,
since rail transit almost always requires
much larger subsidies than bus transit. Thus,
a refocusing of subsidies on improving bus
services would probably benefit the poor
more than spending most future subsidies on
expensive new rail transit systems. Of
course, there are many other reasons for sub-
sidizing rail transit. For example, some stud-
ies suggest that rail systems are more effec-
tive than buses in achieving congestion and
pollution relief, energy savings, economic
development, and more compact land use.18
Moreover, transit systems must be viewed as
a synergistic whole, and even households
that usually ride buses benefit from the
greater connectivity, speed, and coverage
permitted by truly multimodal transit serv-
ices. Low-income households make a much
higher percentage of their trips by transit in
large cities with multimodal systems that
include rail.
Table 9 shows variation in transit’s modal
share by income class and size of metropoli-
tan area. Transit use increases sharply with
population size. Thus, for all income groups
in aggregate, transit modal share rises from
0.4% in areas with less than 250,000 popu-
lation to 3.4% in areas with a population of
3 million or more.
Each of the income groups shown in
Table 9 uses transit much more in large met-
ropolitan areas than in small metropolitan
areas. While only 0.1% of affluent house-
holds use transit in small metropolitan areas,
that modal share rises to 2.2% in the largest
metropolitan areas. The increase is due to the
greater availability of rail transit in large
cities, and the greater likelihood that affluent
households will use rail transit compared to
bus transit. The jump in transit use by the
poor is even greater, from 1.1% to 10.6%.
And the poor use transit more than the afflu-
ent in every population size category. Yet the
ratio of transit mode shares between the
poor and the affluent is highest in the small-
est metropolitan areas (11:1) and lowest in
the largest metropolitan areas (5:1), indicat-
ing that the poor account for a higher per-
centage of total transit riders in small cities
than in large cities. In short, most transit rid-
ers in small cities are bus riders and most of
them are poor. By comparison, transit riders
SOCIOECONOMICS OF URBAN TRAVEL
62
TRANSPORTATION QUARTERLY / SUMMER 2003
in the largest metropolitan areas use both
bus and rail transit and include a much high-
er proportion of affluent users.
Table 8 reveals some interesting impacts
of income on rates of walking, cycling, and
taxi use. Walking declines sharply with
increasing income, from 16.2% of all trips in
the poorest income category to about 9% in
all other income categories. The difference
is all at the lower end of the income scale and
is clearly due to lower auto ownership, as
discussed earlier. Bicycling, by comparison,
appears to be roughly the same at all
incomes levels, accounting for about 0.9%
of all trips across the income spectrum. Taxi
use is bimodal, with the highest usage among
the poor and the affluent. For the poor, taxis
provide the closest substitute for the cars
they are less likely to own. For the affluent,
taxis provide convenient access to airports
and train stations, and quick local trips with-
in downtown areas.
Although most of these income differ-
ences are consistent with those shown in the
1995 NPTS, there are some discrepancies.
For example, the 1995 NPTS showed a
marked decline in cycling with increased
income, while the 2001 NHTS shows no
drop at all. It is possible that bicycling
among higher income classes has increased
substantially since 1995, or it might simply
be due to differences in survey methods.
Likewise, the 1995 NPTS showed a much
higher level of taxi use among the poor
(0.5%) than found in the 2001 NHTS
(0.2%). It is unclear why taxi use among the
poor is less pronounced than in 1995.
Table 10 reflects basically the same sort of
information as Table 8 but presents the dis-
tribution of each mode’s users among the var-
ious income classes, and not the distribution
of each income class’s trips among the modes
(modal split). This information is especially
useful for calculating the equity impacts of
transportation finance. It shows more clearly
than Table 8, for example, that mainly the
poor use buses. Households earning less than
$20,000 account for 47.1% of bus riders but
only 19.7% of metro riders and 6.3% of sub-
urban rail riders. Conversely, households
earning $100,000 or more account for
41.6% of suburban rail riders and 27.2% of
metro riders, but only 6.8% of bus riders.19
Table 10 highlights the bimodal nature of
taxi use, with 22.3% of taxi passengers from
the lowest income class and 33.3% from the
highest income class. Pedestrians are some-
what more concentrated in the lower income
Table 9: Public Transit’s Market Share by Population Size and Household Income
(percentage of trips by transit)
Household Income
Metropolitan Less than $20,000 to $40,000 to $75,000 to $100,000
Area Population $20,000 $39,999 $74,999 $99,999 and over All
Less than 250,000 1.1 0.4 0.1 0.1 0.1 0.4
250,000 - 499,999 2.2 0.3 0.4 0.1 0.3 0.6
500,000 - 999,999 1.8 0.9 0.1 0.1 0.1 0.6
1,000,000 - 2,999,999 5.4 0.6 0.6 0.3 0.3 1.1
3 million or more 10.6 3.4 2.3 1.5 2.2 3.4
Nation 4.8 1.1 0.9 0.7 1.1 1.6
Source: Calculated by the authors from the 2001 NHTS.
Note: The metropolitan statistical area (MSA) totals in this table differ slightly from our other urban totals because MSAs by defi-
nition include entire counties, parts of which can be rural.
63
classes, but bicyclists are distributed evenly
across the entire income spectrum, roughly in
proportion to their share of the population.
As expected, high-income households
make longer trips than low-income house-
holds, as shown in Table 11. For all modes in
aggregate, the average trip length for low-
income households is 1.5 miles shorter than
for the highest-income households (5.6 miles
vs. 7.1 miles). Differences in car trip lengths
are not very large, however—only a mile
between the top and bottom income classes
(6.7 miles vs. 7.7 miles). That suggests that
any user charge or tax proportional to vehi-
cle miles traveled (such as roadway tolls or
the gasoline tax) would be regressive, since
SOCIOECONOMICS OF URBAN TRAVEL
Table 10: Income Distribution of Each Mode’s Users
(percentage composition by income class)
Household Income
Mode of Less than $20,000 to $40,000 to $75,000 to $100,000
Transportation $20,000 $39,999 $74,999 $99,999 and over All
Total Auto 12.3 25.0 32.5 14.4 15.8 100
SOV111.2 24.9 33.3 14.7 15.9 100
HOV213.2 25.0 31.8 14.2 15.8 100
Total Transit 37.8 19.8 21.0 7.4 14.1 100
Bus and Light Rail347.1 21.4 19.0 5.6 6.8 100
Metro/Subway/
Heavy Rail419.7 18.7 25.2 9.1 27.2 100
Commuter Rail56.3 7.0 26.1 19.1 41.6 100
Total Nonmotorized 22.7 22.8 27.4 12.8 14.3 100
Walk 23.6 22.6 26.9 12.6 14.2 100
Bicycle 13.5 24.1 32.8 15.0 14.6 100
School Bus 17.9 22.1 30.0 15.0 15.0 100
Taxicab 22.3 12.5 14.0 17.9 33.3 100
Other 12.3 16.5 30.5 23.2 17.4 100
All 13.9 24.6 31.7 14.2 15.7 100
Overall Sample Distribution
Households 22.7 27.8 27.9 10.3 11.3 100
Persons 17.5 25.2 30.1 13.1 14.0 100
Trips 13.9 24.6 31.7 14.2 15.7 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
64
TRANSPORTATION QUARTERLY / SUMMER 2003
the poor would pay only slightly less than
the affluent, and the payments would be a
much higher percentage of their incomes.20
To offset such regressivity, the tax revenues
would have to be distributed in a way that
explicitly benefits low-income households.
While the lengths of auto trips vary only
slightly by income, the differences are much
larger for transit. Low-income households
make transit trips that are only about half
as long as those by the most affluent transit
riders, but there is substantial variation by
type of transit. Metro trip lengths are only
slightly different among income classes, pos-
sibly due to the long subway trips made by
low-income residents of The Bronx, Brook-
lyn, and Queens to other parts of New York
City’s vast subway network. Income-based
differences in bus and commuter rail trip
lengths are much larger. The affluent make
bus trips that are almost twice as long as
those made by poor households (10.3 miles
vs. 5.9 miles), and they make commuter rail
trips that are four times longer (27.8 miles
Table 11: Average Trip Length by Mode and Income Class
(in miles)
Household Income
Mode of Less than $20,000 to $40,000 to $75,000 to $100,000
Transportation $20,000 $39,999 $74,999 $99,999 and over All
Total Auto 6.7 7.4 7.8 7.7 7.7 7.5
SOV16.4 7.0 7.9 8.4 7.9 7.6
HOV26.9 7.7 7.7 7.2 7.4 7.5
Total Transit 6.0 8.0 8.3 12.0 13.2 8.3
Bus and Light Rail35.9 7.7 7.0 6.7 10.3 6.8
Metro/Subway/
Heavy Rail47.2 8.3 8.0 14.7 8.7 8.7
Commuter Rail57.4 13.5 18.3 23.2 27.8 22.1
Total Nonmotorized 0.7 0.8 0.8 0.9 1.0 0.8
Walk 0.6 0.7 0.7 0.7 0.9 0.7
Bicycle 1.5 1.5 1.8 2.4 2.5 1.9
School Bus 5.0 5.2 5.6 5.2 5.2 5.3
Taxicab 4.1 5.2 7.5 6.2 5.6 5.6
Other 2.2 2.8 3.0 8.8 5.6 4.7
All 5.6 6.7 7.1 7.1 7.1 6.8
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
65
vs. 7.4 miles). That suggests that distance-
based fares would generally favor the poor,
since they make shorter trips. The exception
appears to be metro systems, but there are
some metro systems where distance-based
fares would also favor the poor. Indeed, the
systems in Washington, DC and San Fran-
cisco already have distance-based fare sys-
tems. By comparison, the flat-fare structure
in New York City not only encourages long
trips but also discourages short trips, since
riders pay the same whether they travel one
mile or thirty miles, and can transfer for free
between subway lines as well as between
subway and bus lines.
Finally, low-income households make
considerably shorter walk and bike trips
than high-income households. Their walk
trips are about two-thirds as long, and their
bike trips are three-fifths as long. The longer
trips of more affluent households might be
due to higher incidence of recreational walk-
ing and cycling for exercise or relaxation. It
might also be due to the more central loca-
tions of poor households, where more com-
pact, mixed-use neighborhoods facilitate
shorter trips.
The last of the income-based differences
we examine here is the variation in time of
day of travel. Somewhat similar to the trip
distance patterns in the previous table, there
are no major differences among income
classes in their time of day of car travel. The
lowest-income category accounts for 9.4%
of peak-hour car trips vs. 11.0% of off-peak
car trips (see Table 12). Thus, peak-hour
pricing of roadways might be quite regressive
indeed, either forcing the poor off the roads
during peak hours or extracting burdensome
fees from them out of their limited budgets.
Of course, the proceeds of congestion pricing
could be redistributed to offset its regressivi-
ty, but the initial pricing itself unquestion-
ably would be regressive. In London, for
example, revenues from the newly instituted
congestion pricing in the city center are used
for improvements to public transport. The
revenues might also be used to finance dis-
count transit passes for low-income riders or
special services targeted to serving low-
income neighborhoods.
Time-of-day differences in transit travel
are much larger. For all transit modes in
aggregate, the poor account for 24.9% of
peak-hour transit trips but for 39.4% of off-
peak trips. The differences are greatest for
the rail transit modes. Poor households
account for twice the percentage of off-peak
metro riders as peak-hour riders (18.1% vs.
8.9%) and four times the percentage of off-
peak commuter rail riders as peak-hour rid-
ers (11.7% vs. 3.1%). Thus, large off-peak
discounts on transit fares would greatly ben-
efit poor transit riders. Since rail transit
enjoys substantial extra capacity during the
off-peak hours, the marginal cost of any
additional riders then would be virtually
zero, justifying very low off-peak fares even
on efficiency grounds.
From an equity perspective, the preced-
ing variations in auto ownership, mobility,
and travel behavior among different income
groups are probably the most important.
Nevertheless, there are significant variations
by ethnic and racial group, by sex, and by
age group that must also be considered in the
development of transport policies.
Variation in Travel Behavior by Race and
Ethnicity
Because blacks and Hispanics have consid-
erably lower incomes than whites, the dif-
ferences in travel behavior among these three
groups also reflect differences among income
classes. One thing they have in common is
that they all rely overwhelmingly on the pri-
vate car to get around. Although whites
make the highest percentage of trips by car
(87.6%), the other three groups are not far
behind, with Asians and Hispanics at 83.1%
and blacks at 78.9% (see Table 13).
SOCIOECONOMICS OF URBAN TRAVEL
66
TRANSPORTATION QUARTERLY / SUMMER 2003
Table 12: Peak vs. Off-peak Travel by Income Class
(percentage distribution of each mode’s users by time of day and income)1
Household Income
Mode of Less than $20,000 to $40,000 to $75,000 to $100,000
Transportation $20,000 $39,999 $74,999 $99,999 and over All
Total Auto
Peak 9.4 22.2 33.8 15.9 18.8 100
Off-peak 11.0 24.0 33.1 14.7 17.1 100
Total Transit
Peak 24.9 20.1 22.2 12.8 20.0 100
Off-peak 39.4 21.0 18.9 5.4 15.2 100
Bus and Light Rail2
Peak 36.8 24.6 20.5 10.3 7.9 100
Off-peak 47.3 21.8 18.2 4.7 8.1 100
Metro/Subway/Heavy Rail3
Peak 8.9 15.3 28.1 11.3 36.5 100
Off-peak 18.1 22.2 21.8 6.0 31.9 100
Commuter Rail4
Peak 3.1 9.9 19.8 25.2 42.0 100
Off-peak 11.7 5.0 18.3 13.3 51.7 100
Taxicab
Peak 8.8 20.6 14.7 20.6 35.3 100
Off-peak 18.4 15.8 13.3 15.2 37.3 100
All Modes
Peak 10.5 22.1 33.2 15.7 18.4 100
Off-peak 12.0 23.7 32.6 14.6 17.1 100
All Modes & All Incomes
Peak 31.2
Off-peak 68.8
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. Peak period was defined as 6 to 9 a.m. and 4 to 7 p.m. on weekdays; off-peak included all other times.
2. Light rail also includes conventional streetcars.
3. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
4. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
67
The two nonmotorized modes show quite
different usage patterns. Bicycling is the
highest among whites and Hispanics (0.9%
of all trips). For whites, cycling is mostly for
recreation, while for Hispanics, it is to reach
the workplace. Walking is lower for whites
(8.6%) than for the other three groups, who
make 12%-13% of their trips by walking.
The largest differences among racial and
ethnic groups are in their use of transit.
Blacks are almost six times as likely as whites
to take their trips by transit in general (5.3%
vs. 0.9%), and they are eight times as likely
to take the bus (4.2% vs. 0.5%). They are
also more likely to take the metro (0.9% vs.
0.3%) and even commuter rail (0.2% vs.
0.1%). Hispanics use transit less than blacks
but still about three times more than whites
(2.4% vs. 0.9%). Their use of rail transit is
the same as blacks, but they rely on buses
four times more (2.0% vs. 0.5%). By com-
parison, Asians show just the reverse tenden-
cy, with the highest rail transit modal split
shares of any group but with bus usage less
than among blacks or Hispanics. That might
reflect the concentration of Asian immigrants
in the very largest American cities with
extensive rail transit systems.
It is clear from Table 13 that racial and
ethnic minorities rely far more on transit
than whites. Moreover, they account for a
large percentage of all transit users (not
shown in Table 13). Blacks and Hispanics
together comprise 54% of the country’s tran-
sit users: 62% of all bus riders, 35% of all
metro riders, and 29% of all commuter rail
riders.21 If one includes low-income house-
holds as well, the combination of blacks,
Hispanics, and low-income nonminority
households comprises an even higher per-
centage of transit riders: 63% overall, and
73% of bus riders, 44% of metro riders, and
31% of commuter rail riders.
Thus, improving transit services and fare
structures in American cities would generally
benefit minorities, as well as low-income
households. Nevertheless, blacks, Hispanics,
and poor households all rely primarily on
bus transit and far less on rail transit. Subsi-
dies spent on improving bus systems would
especially favor minorities, as well as low-
income households in general.
As documented extensively in the litera-
ture, most transit systems have tended to
take minority and low-income “captive rid-
ers” for granted and focused their fare and
service policies on attracting middle-class
and affluent riders out of their automobiles.22
In many cases, the result has been lower-
quality service for the poor and minorities
and superior service, at high public subsidy
cost, for the affluent. New and extended rail
transit systems, in particular, have been
aimed at luring affluent suburban motorists
out of their cars to reduce congestion, air
pollution, and energy use in American cities.
Some have argued that it would be both
more equitable and more efficient to target
limited subsidy dollars to inner city bus serv-
ices that are cheaper, more intensively used,
and require far less subsidy per passenger
served.23
The impacts of transit subsidies, service
distribution, and fare structure on minority
groups have had legal consequences, espe-
cially during the 1980s. Civil rights organi-
zations filed numerous administrative com-
plaints and law suits against transit systems
whose fare and service policies were seen as
discriminating against minority riders. They
claimed that such discrimination violates
Title VI of the Civil Rights Act, even if it is
not intentionally aimed at harming minori-
ties but has that effect. Recent court rulings
requiring proof of intent have virtually ended
legal challenges of this sort. Nevertheless, it
remains an important issue, especially since
minorities comprise such a high percentage
of transit riders.24
Variation in Travel Behavior by Sex
At least in terms of their travel behavior,
women and men are becoming more and
SOCIOECONOMICS OF URBAN TRAVEL
68
TRANSPORTATION QUARTERLY / SUMMER 2003
more alike. As shown in Table 14, there are
only minor differences by sex in choice of
travel mode. Men and women both rely on
the private car for about 86% of their urban
trips. The only difference here is that women
are more likely than men to carpool (51.5%
vs. 44.7%), perhaps because mothers often
chauffeur their children to school, sports
events, and friends’ houses. Transit use, taxi
use, and walking are only slightly different
among men and women. The only major dif-
ference in travel behavior is that women are
far less likely to cycle (0.5% vs. 1.2%). By
comparison, women cycle almost as much
as men in countries such as The Netherlands,
Denmark, and Germany, where cities have
invested heavily in cycling infrastructure and
a range of policies to make cycling safe.25
Table 13: Variation in Modal Choice by Race/Ethnicity
(percentage of trips by means of transportation)
Race/Ethnicity
Transportation Black Asian White Hispanic6
Total Auto 78.9 82.7 87.6 83.1
SOV135.7 33.5 40.1 27.5
HOV243.2 49.3 47.6 55.5
Total Transit 5.3 3.2 0.9 2.4
Bus and Light Rail34.2 1.8 0.5 2.0
Metro/Subway/Heavy Rail40.9 1.1 0.3 0.3
Commuter Rail50.2 0.3 0.1 0.1
Total Nonmotorized 13.2 12.3 9.6 12.6
Walk 12.6 11.7 8.6 11.8
Bicycle 0.6 0.5 0.9 0.9
School Bus 2.1 1.4 1.3 1.5
Taxicab 0.2 0.2 0.1 0.1
Other 0.2 0.1 0.4 0.3
All 100 100 100 100
Overall Sample Distribution7
Percent of Total Households 11.3 2.1 74.3 8.7
Percent of Total Trips 11.5 2.7 69.9 12.7
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
6. The Hispanic category was defined to be mutually exclusive of blacks and whites.
7. Rows do not add to 100% because some racial and ethnic categories are not shown.
Mode of
69
Variation in Travel Behavior by Age
Table 15 shows that mobility rates are lowest
for children and the elderly, both in terms of
trips per day and mileage covered. The age
group 25-64 has the highest mobility at 4.4
trips per day and 32.7 miles per day. That is
a third more trips per day than children and
the elderly, and almost twice the mileage per
day. Within the elderly grouping, however,
there are enormous variations in mobility
rates, much larger than the differences
between the elderly and nonelderly. Thus,
persons 85 years and older made only 1.9
trips per day, less than half the 3.9 trips per
day made by those 65 to 69 years old. Simi-
larly, persons 85 years and older covered
only about a third as many miles per day as
persons 65 to 69 years old.
While mobility rates clearly decline for the
elderly, their choice of travel mode is quite
similar to the rest of the adult population (see
Table 16). Just as other Americans, they are
overwhelmingly dependent on the car for
getting around town. Indeed, they rely on the
car for 89.1% of their trips, a higher per-
centage than for any other age group and
three percentage points higher than the pop-
ulation as a whole. That is not surprising
given the greater convenience, comfort, and
privacy of the auto compared to other
modes. What is perhaps surprising is that the
SOCIOECONOMICS OF URBAN TRAVEL
Table 14: Variation in Modal Choice by Sex (percentage of trips by means of transportation)
Sex
Transportation Male Female All
Total Auto 85.6 86.0 85.8
SOV140.8 34.5 37.6
HOV244.7 51.5 48.2
Total Transit 1.7 1.8 1.7
Bus and Light Rail31.1 1.3 1.2
Metro/Subway/Heavy Rail40.4 0.4 0.4
Commuter Rail50.2 0.1 0.1
Total Nonmotorized 10.6 10.5 10.5
Walk 9.3 9.9 9.6
Bicycle 1.2 0.5 0.9
School Bus 1.6 1.3 1.4
Taxicab 0.1 0.1 0.1
Other 0.5 0.3 0.4
All 100 100 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
Mode of
70
TRANSPORTATION QUARTERLY / SUMMER 2003
elderly make over half of their car trips as
drivers, while most other age groups (except
40 to 64) make more trips as passengers than
as drivers. Clearly, the elderly rely on the
mobility and independence that the automo-
bile enables them to preserve as they grow
older. The main concern is that many elderly
continue to drive in spite of serious deterio-
ration of their eyesight, hearing, and reflexes,
thus endangering themselves and others.
While elderly Germans and Dutch make
over half their trips by walking or cycling,
those nonmotorized modes account for only
9% of the trips of elderly Americans.26 Even
the Dutch elderly who are 75 or older make
a fourth of all their trips by bike. Germans
in this 75+ age group make 7% of their trips
by bike. By comparison, Americans who are
65 or older make only 0.4% of their trips by
bike.
In the United States, there are no feasible
alternatives to the private car for most trip
purposes in most cities. That forces the eld-
erly to drive, whether they want to or not.
Not only does the forced reliance on the pri-
vate car expose elderly Americans to consid-
erable traffic dangers, it deprives them of
valuable physical exercise they would get
from walking and cycling.
There are few differences between the
findings of the 1995 NPTS and the 2001
NHTS regarding the impact of age on travel
behavior. The mobility rate differences
among age groups are virtually identical. The
modal split share of walking almost doubles
for all age groups, but that is due to the
change in survey methodology. The 1995
NPTS and 2001 NHTS both confirm the
overwhelming reliance of the elderly on the
private car, as well as their high proportion
of car trips as drivers. The 2001 NHTS,
however, reports a decline in transit use by
the elderly (from 2.2% in 1995 to 1.3% of
all trips in 2001).
It is notable that the elderly are less likely
than the population as a whole to take tran-
sit (1.3% vs. 1.7% of trips). Most of the
transit trips the elderly make are by bus, with
the two rail transit modes together account-
ing for only 0.1% of all trips by elderly
Table 15: Impact of Age on Mobility Levels
Age Trips per Day, per Person Miles Traveled per Day, per Person
5 to 15 3.4 17.1
16 to 24 4.0 28.3
25 to 39 4.4 32.9
40 to 64 4.4 32.4
65+ 3.4 18.7
65 to 69 3.9 24.4
70 to 74 3.8 20.8
75 to 79 3.1 16.2
80 to 84 2.8 13.6
85+ 1.9 9.2
All 4.0 27.0
Source: Calculated from the 2001 NHTS by Mary Ann Keyes, Federal Highway Administration, US Department of Transportation.
Note: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
71
Americans. That might reflect the difficulty
of reaching rail transit stations, which tend
to be located farther away and require either
a long walk or a bus trip and transfer to
reach them. The elderly also have difficulty
negotiating the stairs in some rail stations,
many of which are still not accessible for per-
sons with disabilities. That is especially true
in old subway systems like New York City’s,
where less than 5% of stations are wheel-
chair accessible.27 At most stations, the rider
must negotiate two or three long flights of
stairs and long, circuitous passageways.
Older subway and commuter rail systems,
with over 80% of the country’s rail transit
passengers, have found it too expensive to
fully convert their stations.
In addition, most rail systems are radially
designed, with a focus on serving peak-hour
work trips between the suburbs and down-
town. That obviously is not the sort of trip
most elderly need to make. For shopping,
medical, or social trips during the off-peak,
bus services are usually a better option. That
might also help explain the lesser use of rail
transit by the elderly.
SOCIOECONOMICS OF URBAN TRAVEL
Table 16: Impact of Age on Modal Choice (percentage of trips by means of transportation)
Age
Transportation 5 to 15 16 to 24 25 to 39 40 to 64 65 & over All
Total Auto 70.7 85.3 87.4 89.8 89.1 85.8
SOV10.5 39.2 43.6 51.9 45.7 37.6
HOV270.2 46.1 43.8 38.0 43.4 48.2
Total Transit 1.1 2.9 2.1 1.5 1.3 1.7
Bus and Light Rail30.9 2.1 1.2 1.0 1.2 1.2
Metro/Subway/Heavy Rail40.1 0.6 0.7 0.3 0.1 0.4
Commuter Rail50.0 0.2 0.2 0.2 0.0 0.1
Total Nonmotorized 18.4 10.0 9.8 8.2 9.3 10.5
Walk 15.2 9.3 9.2 7.8 8.9 9.6
Bicycle 3.2 0.6 0.6 0.4 0.4 0.9
School Bus 8.9 1.2 0.0 0.0 0.1 1.4
Taxicab 0.1 0.1 0.2 0.1 0.1 0.1
Other 0.8 0.4 0.3 0.3 0.2 0.4
All 100 100 100 100 100 100
Source: Calculated by the authors from the 2001 NHTS.
Notes: In order to isolate urban travel, the sample was limited to residents of urban areas and trips of 75 miles or less.
1. SOV (single occupancy vehicle) includes vehicles with driver and no passengers.
2. HOV (high occupancy vehicle) includes vehicles with two or more occupants.
3. Light rail also includes conventional streetcars.
4. Metro/subway/heavy rail includes elevated rail and rail rapid transit.
5. Commuter rail includes suburban/regional rail systems and short-distance service provided by Amtrak.
Mode of
72
TRANSPORTATION QUARTERLY / SUMMER 2003
Conclusions and Policy Implications
The most obvious message from the 2001
NHTS is that the private car continues to
dominate urban travel among every segment
of the American population. Indeed, the car’s
percentage of total trips probably increased
from 1995 to 2001, even though the 2001
NHTS shows a slight decline from 1995. As
noted previously, the NHTS used a new sur-
vey methodology that almost doubled the
number of reported walk trips, which in turn
reduced the percentage of car trips. More
surprising, perhaps, is the increased share of
HOV trips compared to SOV (from 51.5%
of car trips in 1995 to 56.6% in 2001). The
increase might be due to the long-term
decline in the percentage share of work trips,
which have the lowest vehicle occupancy,
and a corresponding rise in the percentage
share of nonwork trips, which have the high-
est vehicle occupancies. Thus, the finding
does not necessarily contradict US Census
data that report a long-term decline in car-
pooling for the work trip. Rather, it may
reflect the declining relative importance of
the journey to work, which in 2001 account-
ed for less than a fifth of all trips.28
Public transit’s share of urban trips con-
tinued to decline between 1995 and 2001,
from 2.2% to 1.7%, but the reported decline
is exaggerated due to the increased sampling
of walk trips.29 Since total unlinked transit
trips—as reported by transit systems—
actually rose by over a fifth between 1995
and 2001, such a sharp decline in transit’s
market share seems unlikely.30 Some of the
reported increase in unlinked transit trips,
however, was artificial, resulting from addi-
tional transfers caused by the redesign of
route networks with timed-transfer hubs.
Moreover, the US Census shows a consider-
able decline in transit’s market share of the
work trip from 1990 to 2000 (from 5.3% to
4.7%). That also lends some credibility to
the declining transit share of total trips
(from 2.2% to 1.7%) reported by the 2001
NHTS.
Nonmotorized transportation’s share of
urban trips (not shown in Table 2, which
includes both urban and rural trips)
increased from 6.8% to 10.4% between
1995 and 2001. Bicycling’s share remained
stable at 0.9%, while the walking share rose
from 5.9% to 9.5% due to the survey
changes noted earlier. Taxi use declined from
0.18% to 0.13% of all urban trips.31
Clearly, the 1995 NPTS and 2001 NHTS
are not directly comparable. As noted earli-
er in our description of the NHTS survey,
several significant changes in methodology
were made that affected the results. Thus, all
the differences between 1995 and 2001 sta-
tistics must be viewed with caution. Never-
theless, the 1995 NPTS and 2001 NHTS
show almost identical patterns of differences
in travel behavior among different socioeco-
nomic groupings. For example, both surveys
confirm that only a small percentage of the
urban poor use any form of transit (6.8% in
1995 vs. 4.6% in 2001) and instead rely on
the auto for the vast majority of their trips
(75.9% in both 1995 and 2001). Both sur-
veys confirm the income disparities among
transit riders, with bus riders the poorest and
commuter rail riders the most affluent. Both
show that poor transit riders are more likely
to ride during the off-peak and to make
shorter trips than affluent riders. Differences
in travel behavior by ethnic and racial group,
sex, and age are also virtually the same in
2001 as in 1995.
The overall policy implications of this
socioeconomic analysis of the 2001 NHTS
are roughly the same as those proposed by
one of the authors in his analysis of the 1995
NPTS.32 The disadvantaged in our society,
especially the poor, minorities, and the eld-
erly, depend crucially on the private car to
get around the cities they live in. They use
public transit for only a tiny percentage of
their overall trips. Thus, public transit can-
73
not be the main strategy for improving the
mobility of these groups. Automobiles are
obviously a necessity for disadvantaged
groups for reaching most employment, edu-
cational, medical, shopping, social, and
recreational destinations. Even those who
cannot really afford cars or who have physi-
cal or mental disabilities are forced to rely on
the car.
Nevertheless, public transit plays a critical
role in assuring the mobility of disadvan-
taged groups in the largest, densest cities. In
metropolitan areas with populations of 3
million or more, public transit serves 9.7%
of the trips of blacks, 10.6% of the trips of
the poor, and 28.7% of the trips of house-
holds without cars.33 It is essential that gov-
ernment housing policies be coordinated
with transportation in order to ensure the
continued accessibility of disadvantaged
groups to transit. As noted earlier, low-
income households are currently being dis-
placed through the gentrification of neigh-
borhoods around rail stations. Furthermore,
government agencies have been decentraliz-
ing public housing for the poor and building
it at lower densities, often located in neigh-
borhoods with little if any transit service.
Both housing and transportation policies
should be coordinated to facilitate the acces-
sibility of low-income households to transit.
Walking is probably the most ignored
mode of transport, both in general as well as
in reference to its importance among the dis-
advantaged. As shown in Tables 8 and 13,
walking accounts for 16.2% of the trips by
the poor, 12.6% of trips by blacks, and
11.8% of the trips of Hispanics. Yet in the
United States, facilities for pedestrians are
often inconvenient or nonexistent, leading to
fatality rates per mile traveled 36 times high-
er than for occupants of cars and light
trucks.34 The lack of pedestrian safety espe-
cially affects minorities and the poor. For
example, blacks account for 20% of all
pedestrian deaths, almost twice their 12%
share of the total population.35
In The Netherlands and Germany, pedes-
trian fatalities per mile walked are only a
tenth as high as in the United States.36 Euro-
pean countries have invested heavily in
extensive auto-free pedestrian zones; pedes-
trian-activated crossing signals; pedestrian
refuge islands for crossing wide streets; wide,
well-lit sidewalks on both sides of all streets;
and traffic calming of most residential neigh-
borhoods. Moreover, German and Dutch
pedestrians benefit from comprehensive
restrictions on motor vehicle use, rigorous
traffic education of motorists, and strict
enforcement of traffic regulations protecting
pedestrians. Such measures are essential for
improving pedestrian safety in the USA as
well.
While over $75 billion a year is spent on
federally-assisted roadway projects, less than
$1 billion a year is spent on pedestrian and
bicycling projects.37 Only 0.7% of federal
transportation funds are spent on improving
the pedestrian environment and making it
safer to walk. Moreover, “no state spends
more than 2.7% of their federal transporta-
tion funds on sidewalks, crosswalks, traffic
calming, speed humps, multiuse paths, or
safety programs for cyclists and pedestri-
ans.”38 Given the importance of walking in
our overall urban transportation system, it
is regrettable that all levels of government in
the United States have so woefully neglected
the needs of pedestrians.
The improved survey methodology in the
2001 NHTS reveals the crucial importance
of walking for getting around cities, especial-
ly for the poor, minorities, and those without
cars. Of course, there are many reasons to
encourage more walking among all groups—
to reduce roadway congestion, air pollution,
noise, parking needs, energy use, and above
all, to provide more daily physical exercise
for everyone. Walking is especially important
for the poor and minorities. Not only is it the
most affordable of all transport modes, but it
is also the most feasible in the inner city
neighborhoods where many poor and minor-
SOCIOECONOMICS OF URBAN TRAVEL
74
TRANSPORTATION QUARTERLY / SUMMER 2003
ity households are concentrated and where
so many things are within walking distance.
Moreover, walking is the most important
access mode for reaching transit stops. Since
the poor and minorities depend on transit so
much more than other socioeconomic
groups, walking is crucial for that reason as
well. For all these reasons, it is essential that
federal, state, and local government agencies
focus more on improving the safety, conven-
ience, and feasibility of walking in our cities.
Endnotes
1. John Pucher, Chris Hendrickson, and Sue McNeil. “Socioeconomic Characteristics of Transit Riders:
Some Recent Evidence.” Traffic Quarterly 35(3) (1981): 461-483; John Pucher, and Fred Williams.
“Socioeconomic Characteristics of Urban Travelers: Evidence from the 1990 NPTS.” Transportation
Quarterly 46(4) (1992): 561-582; John Pucher, Tim Evans, and Jeff Wenger. “Socioeconomics of Urban
Travel: Evidence from the 1995 NPTS.” Transportation Quarterly 52(3) (1998): 15-33.
2. The 1960 Census figures, unlike all later census years, included an “unreported” category that
accounted for 4.3% of all survey responses. To make the 1960 modal split distributions comparable with
later census years, the authors scaled up all reported modal shares by a factor of 1.045 so that the
modal shares add up to approximately 100%.
3. It is important to note that these NPTS and NHTS modal split distributions in Table 2 differ from
those in subsequent tables because they include all local, daily travel in the USA, including both rural and
urban areas. These distributions were supplied directly by the Federal Highway Administration of
USDOT. Long-term trend data were available only on this nationwide basis. Our own cross-tabulations
of the 2001 NHTS, shown in subsequent tables, include only urban areas, except for Table 9, which
includes some rural portions of counties in metropolitan statistical areas.
4. Calculated by the authors from the 2001 NHTS. For full details, see Table 6 of this article.
5. Federal Highway Administration and Bureau of Transportation Statistics. Inklings: Preliminary
Results from the 2001 NHTS. Washington, DC: US Department of Transportation, 2003.
6. Federal Highway Administration. Highway Statistics. Washington, DC: US Department of Trans-
portation, various years; and International Road Federation. World Road Statistics 2002. Washington,
DC: International Road Federation, 2002.
7. John Pucher, Tim Evans, and Jeff Wenger. “Socioeconomics of Urban Travel: Evidence from the
1995 NPTS.”
8. Katherine M. Flegal, Margaret D. Carroll, Cynthia L. Ogden, and Clifford L. Johnson. “Prevalence
and Trends in Obesity Among Adults, 1999-2000.” Journal of the American Medical Association
288(14) (2002): 1723-1727; Carlos Dora. “A Different Route to Health: Implications of Transport
Policies.” British Medical Journal 318 (1999): 1686-1689; Jeffrey P. Koplan, and William H. Dietz.
“Caloric Imbalance and Public Health Policy.” Journal of the American Medical Association 282 (1999):
1579-1581; Douglas Carnall. “Cycling and Health Promotion.” British Medical Journal 320 (2000):
888; Simon P. Wolff, and C.J. Gilham. “Public Health Versus Public Policy? An Appraisal of British
Urban Transport Policy.” Public Health 105 (1991): 217-228; Mayer Hillman. “Health Promotion:
The Potential of Non-motorized Transport,” in Tony Fletcher, and Anthony J. McMichael (eds). Health
at the Crossroads: Transport Policy and Urban Health. London: Wiley and Sons, 1997.
9. US Department of Health and Human Services. “Physical Activity and Health: A Report of the Sur-
geon General.” Atlanta, GA: Centers for Disease Control and Prevention, 1996; and US Department of
75
Health and Human Services. Healthy People 2010: Understanding and Improving Health. 2nd ed. Wash-
ington, DC: US Government Printing Office, November 2000.
10. John Pucher, and Christian Lefevre. The Urban Transport Crisis in Europe and North America.
London: Macmillan Press, 1996; World Health Organization. Obesity in Europe: The Case for Action.
London: International Obesity Taskforce of the World Health Organization, September 2002. Accessible
at: http://www.iotf.org/media/globalprev.htm.
11. John Pucher.“Discrimination in Mass Transit.” Journal of the American Planning Association 48(3)
(1982): 315-326; Mark Garrett, and Brian Taylor. “Reconsidering Social Equity in Public Transit.”
Berkeley Planning Journal 13 (1999): 6-27; R. Bullard, and G. Johnson, eds. Just Transportation. Stony
Creek, CT: New Society Publications, 1997.
12. John Pucher. “Renaissance of Public Transport in the USA?” Transportation Quarterly 56(1) (2002):
33-50.
13. R. Cervero, and M. Duncan. “Benefits of Proximity to Rail on Housing Markets: Experiences in
Santa Clara County.” Journal of Public Transportation 5(1) (2002): 1-18; R. Cervero, and M. Duncan.
“Transit’s Value-Added Effects: Light and Commuter Rail Services and Commercial Land Values.”
Transportation Research Record 1805 (2002): 8-15; J. Lin. “Gentrification and Transit in Northwest
Chicago.” Journal of the Transportation Research Forum 56(4) (2002): 175-191; G. Knaap, C. Ding,
and L. Hopkins. “Do Plans Matter? The Effects of Light Rail Plans on Land Values in Station Areas.”
Journal of Planning Education and Research 21(1) (fall 2001): 32-39.
14. John Pucher, Tim Evans, and Jeff Wenger. “Socioeconomics of Urban Travel: Evidence from the 1995
NPTS.”
15. This might also be true of light rail transit, but there were so few light rail observations in both the
1995 NPTS and the 2001 NHTS that it was impossible to separate out light rail for detailed socioeco-
nomic analysis of its riders.
16. Thus, if the 31.1% growth rate is applied to the $15,000 income level in 1995, it would yield
$19,700, quite close to the $20,000 cutoff we used for 2001. Applying 31.1% to the upper income cat-
egory of $80,000 in 1995 yields $104,800, somewhat higher than the $100,000 category cutoff we
used for 2001. The 31.1% growth in per capita income from 1995 to 2001 is derived from US Bureau
of the Census, 2002 Statistical Abstract of the United States, Table 2, on population trends and Table
640, on personal income trends.
17. For details on location efficient mortgages, see http://www.locationefficiency.com.
18. Vukan Vuchic. Transportation for Livable Cities. New Bruswick, NJ, CUPR Press, 1999; and Peter
Newman, and Jeffrey Kenworthy. Sustainability and Cities, Washington, DC, Island Press, 1999.
19. We tried to disaggregate metros into old systems (such as in New York City, Boston, and Chicago)
and new systems (such as in Washington, DC, San Francisco, and Atlanta), since the two types have quite
different designs and rider characteristics. We also tried to disaggregate light rail systems into old street-
car systems (such as in Boston and San Francisco) and new LRT systems (such as in St. Louis, Sacra-
mento, Portland, OR, and San Jose, CA). Unfortunately, there were not enough sample observations to
permit such further disaggregation. Indeed, it was not even possible to produce a separate category for
light rail and streetcar combined, since they only generated 38 total trip observations (0.02% of all trips).
Thus, LRT/streetcar had to be lumped in with bus services, as in previous census and NPTS studies.
20. Taxes and user charges are regressive when payments as a percentage of income fall with increasing
household income.
21. Calculated by the authors from the 2001 NHTS.
SOCIOECONOMICS OF URBAN TRAVEL
76
TRANSPORTATION QUARTERLY / SUMMER 2003
22. John Pucher. “Discrimination in Mass Transit;” Mark Garrett, and Brian Taylor. “Reconsidering
Social Justice in Public Transit”; Robert D. Bullard, and Glenn S. Johnson. Just Transportation.
23. See note 22 above.
24. See note 22 above.
25. John Pucher, and Lewis Dijkstra. “Making Walking and Cycling Safer: Lessons from Europe.” Trans-
portation Quarterly 54(3) (2000): 25-50.
26. See note 25 above.
27. Information provided by the New York City Transit Authority.
28. Federal Highway Administration and Bureau of Transportation Statistics. Inklings: Preliminary
Results from the 2001 NHTS. Washington, DC: US Department of Transportation, 2003.
29. The 1.7% transit modal share cited here for 2001 is for urban travel only, compared to a 1.6%
transit modal share for both urban and rural travel combined, as shown in Table 2. Likewise, the 2.2%
transit modal share cited here for 1995 is for urban travel only, as reported in John Pucher, Tim Evans,
and Jeff Wenger. “Socioeconomics of Urban Travel: Evidence from the 1995 NPTS,” Exhibit 3. By
comparison, the 1.8% transit share shown in Table 2 includes both urban and rural travel.
30. John Pucher. “Renaissance of Public Transport in the USA?”
31. See Table 8 and John Pucher, Tim Evans, and Jeff Wenger. “Socioeconomics of Urban Travel: Evi-
dence from the 1995 NPTS,” Exhibit 3.
32. See note 31 above.
33. Calculated by the authors from the 2001 NHTS.
34. See note 25 above.
35. Surface Transportation Policy Project. Mean Streets 2000. Washington, DC: Surface Transportation
Policy Project, 2001.
36. See note 25 above.
37. Federal Highway Administration. Highway Statistics 2000. Washington, DC: US Department of
Transportation, 2002; US Rep. James Oberstar. Opening remarks at Railvolution Conference, Wash-
ington, DC, October 4, 2002; Surface Transportation Policy Project. Mean Streets 2000. Washington,
DC: Surface Transportation Policy Project, 2001.
38. Surface Transportation Policy Project, 2001, p. 5.
Acknowledgments
The authors would like to thank Martin Wachs, Alan Pisarski, Steven Polzin, W. Bruce Allen, Susan
Liss, Bryant Gross, Nancy McGuckin, and Mary Ann Keyes for their advice and assistance in analyzing
the 2001 NHTS. We take full responsibility, however, for any remaining errors and for all opinions
expressed in this article.
77
John Pucher is a professor in the Bloustein School of Planning and Public Policy at Rutgers
University (New Brunswick, New Jersey). Since earning a Ph.D. at the Massachusetts Institute
of Technology in 1978, Pucher has conducted research on a wide range of topics in transport
economics and finance, including numerous projects he has directed for the US Department of
Transportation, the Canadian government, and various European ministries of transport. In
1996 Macmillan Press (UK) published The Urban Transport Crisis in Europe and North
America, which summarizes Pucher’s comparative research on transport systems, travel behav-
ior, and public policies. Currently, his research focuses on walking and bicycling, and in par-
ticular, how American cities could learn from European policies to improve the safety, con-
venience, and feasibility of these nonmotorized modes in the United States. Pucher is
co-principal investigator of a project for the Robert Wood Johnson Foundation that exam-
ines the need for Americans to increase their walking and cycling for daily transportation as
the best way to ensure adequate levels of physical exercise and enhance overall public health.
He is also working on a pedestrian/bicycle bill of rights that would improve walking and
cycling conditions in American cities by reforming existing traffic statutes, which currently
favor the motorized modes.
John Luciano Renne is a Ph.D. candidate and lecturer at the Bloustein School of Planning and
Public Policy at Rutgers University. He teaches a graduate course in Sustainable Urban Devel-
opment and is also a project manager at the Voorhees Transportation Policy Institute, where
he is currently evaluating the New Jersey Transit Villages Initiative. Renne is an FHWA Eisen-
hower Transportation Fellow and Eno Transportation Foundation Fellow. He recently pub-
lished “Facilitating the Financing and Development of ‘Smart Growth’” in the “Ideas in
Motion” section of the spring 2002 edition of Transportation Quarterly.
SOCIOECONOMICS OF URBAN TRAVEL
