Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 51
TCRP H-12 Final Report
3.0 EMPIRICAL FINDINGS AND IMPLICATIONS
FOR MODELING
3.! Introduction
This chapter reports Me results of a review of the empirical literature on the transportation
land-use interaction, at both We "macro" level of urban fonn (density of development, overall urban
structure, etc.) arid the "micro" level of loch area design. The objectives of this review are:
to identify our current state of knowledge concerning the transportation - land-use
interaction, particularly with respect to the impacts which various land-use and
transportation policies are likely to have on the system,
to identify gaps in our understanding (i.e., areas in which eve are unsure of the
direction/magn~tude of policy impacts), as a guide to fiercer research. arid
to discuss We implications of these findings for the development and application of
integrated transportation - land-use models.
The recent series of reports by Parsons Brinkerhoff. Quade arid Douglas Inc. (PBQD) tI995,
1996a, 1996b, 1996c] provided art excellent starting point for this study, as well as a comprehensive
review of the literature up to 1994. Thus. although studies up to ~ 994 are also reviewed in this
report, greater emphasis is placed on the more recent studies which have been published since ~ 994.
The review is divided into two main parts. The first reviews the empirical literature on the
impact of urban form on Gavel behavior. The second then considers the problem from the other
Ejection by examining the evidence concerning the impact of transportation fin particular. transit)
on urban form. Deviled results from the review are presented in Appendix A. Section 3.2 of this
chapter surnmanzes these results with a focus on their implications for the design and use of
integrated urban models.
3.2 Implications for Integrated Transportation - Land-Use Models
3.2.1 Impacts of Urban Form on Transit
Table 3. ~ attempts to summarize the empincal evidence conceding urban form impacts on
Ravel which is discussed In greater detail in Appendix A. To bnug some structure to this summary,
the table is divided into four main categories: residential density impacts, employment density
impacts, accessibility impacts, arid neighborhood design impacts. The first three of these categories
can be thought of as "macro" properties of urban form, while the fourth is intended to capture more
- 27
OCR for page 52
TCRP H-17 Final Report
Table 3.! Summary of Empirical Evidence Concermng
Urban Form Impacts on Travel
(a) Residential Density Impacts
Transit demand function of: she
of downtown, distance from downtown,
residential density
Transit use increases sharply when residential
density goes from 7-16 dwelling un~ts/acre
1973 NETS data -- density does not explain
· . . . .
much variance In transit usage, soc~o-econom~cs
more important
.
[PusDkarev & Zup an, 1977, 1980; pg. 177]
[Smith, 1984, pg. 177]
[PMM, 1975. pg. 177]
1990 NETS data -- threshold for relationship [Levison & Kumar, 1994, pg. 177]
between density & mode choice is 10,000 persons/sq.mi.
1991 FHWA Highway Statistics -- urban regions
with higher densities generate higher transit
trips per capita and less VMT per capita, trends
hold when analyzed at the zonal rather than
regional level; total trip rates do not vary
substantially with density
Households in higher density neighborhoods have
lower auto ownership levels (after controlling
for income); people with fewer cars use transit
more and generate less VMT; net density effects,
however, are generally small In practical terms
Density has negligible impact on travel behavior
(except with respect to auto ownership) once
accessibility is accounted for
Higher densities, in combination with
neighborhood design, reduce trip rates
& encourage non-auto travel
tDunphy & Fisher, 1996, pg. ~ 78]
[Shimek, 1996b, pg. 179]
~Kockelman, 1997, pg. iS0]
[Cervero & Kockelman, 1997, pg. ISi]
- 28
OCR for page 53
TCRP H-12 Final Report
Table 3.l, cont'd
Higher residential densities, combined win
greater concentration of employment In the
CBD & inner suburbs, plus socio-economic differences,
contribute to higher transit mode splits In
Toronto relative to Boston
1985 American Housing Survey Data -- density
has a significant Impact on work trip
auto and transit mode shares; this Impact is
greater than the land-use mix impact
Increased density, combined win mixed land-uses,
reduced auto ownership levels and commuting
distances
Density greatly influences commuter mode choice,
transit Hip rates and rapid rail boardings. much
more so man urban design or land-use mix
Rail access walk mode shares affected by neighborhood
density and land-use mix; the distances walked,
however, are not; rail access transit mode share
primarily affected by transit service levels and
not density/land-use mix per se
Catchment areas for rail stations vary with
density of development
Density affects auto ownership, which, In turn,
affects transit ndership
Density does not directly affect commuting
VMT/worker
Residential density near commuter rail stations
not significant in explaining choice of walk
access mode, once socio-economic factors properly
[Schimek, 1996a, pg. 1811
ECervero, 1996, pg. IS4]
[Cervero, 1996, pg. 184]
[PBQD, 1996a, pg. 186]
EPBQD, 1996c, pg. 187]
[PBQD, 1996c, pg. 187]
[Messenger & Ewing, 1996, pg. 189]
[Miller & Ibrah~m, 1998, pg. 189]
[Loutzenheiser, 1997, pg. 192]
- 29
OCR for page 54
TCRP H-12 Final Report
Table 3~l, cont'd
(b) Employment Density Impacts
Increased concentration of employment in
CBD & inner suburbs contributes to higher
transit modal shares
Employment density thresholds exist for
modal shifts from SOV to transit/walking:
20-50 employees per acre -- moderate shifts;
> 75 employees per acre -- significant shifts
Station area employment density influences
commuter raid boardings
Suburban employment centre density affects
work trip mode split
The larger the size of a suburban employment
center, the more potential exists for
ride-shar~ng (ride sharing increases by 3.5%
for every 5000 jobs)
(c) Accessibility Impacts
Accessibility an unportant determinant of
VMT and mode choice
A stronger relationship between mode choice
and urban form exists when both ends of the
trip are considered
Increases in both "local" and "regional 't
accessibility results in shorter shopping Hip
distances, but no reduction in shopping trip
frequency
.
Residences and commercial areas must be close and barriers
[Schimek, 1996a, pg. ISI]
[Frank & Pivo, 1994 pg. ISI]
[PBQD, 1996a, pg. IS6]
[Cervero, 1989, pg. IS6]
[Cervero, 1989, pg. IS6]
[Kockelman, 1997, pg. IS0]
[Frank & Pivo, 1994, pg. 181]
tHandy, 1993, pg. IS2]
tHandy, 1995, pg. 183]
- 30
OCR for page 55
TCRP H-12 Final Report
Table 3.l, cont'd
Increased employment In the Toronto central area
offset by Increased residential development, resulting
in little net growth In peak-period connoting into
the central area
Commuting cost (and hence jobs-hous~ng balance)
has little Impact on residential location choices
in Los Angeles region
Jobs-housing balance, In combination with auto
ownership level and transit service level, affects
transit mode split
VMT/worker increases with distance from the COD
and/or other major employment centres; jobs-hous~ng
balance has little impact on VMT/worker
Good regional accessibility to activities is
more important Can localized density or use Iriix
In reducing vehicular travel; iobs-hous~n~ balance has little
, .~ , I.
Impact on vehicular travel, but accessibility of residences
to a range of land-uses does reduce vehicular travel
Distance from home to store Important in Me
choice of destination for walk shopping trips
(~1) Neighborhood Design Impacts
Comparison of new, mixed land-use comrnun~ties
with "semi-planned" traditional suburbs -- no
significant difference in VMT or transit usage,
except for recreational trips
Pre-war, traditional comrnun~ties versus
standard, post-war suburbs: transit use & walking
significantly higher In traditional communities;
total trip rates and auto-driver trip rates
[Nowlan & Stewart, 1991, pg. 188]
[Giuliano & Small, 1993, pg. lSS
[Messenger and Ewing, 1996, pg. ~ 89]
[Miller & Ibrahim, 1998, pg. 189]
[Ewing, 1995, pg. 190]
tHandy, 1996b, pp. 191]
[Burby, et al., 1974, pg. 178]
[Friedman, et al., 1994, pg. 178]
- 31
OCR for page 56
TCRP H-12 Final Report
Table 3.l, cont'd
Land-use mixing, land-use balance and
accessibility important in detennining VMT,
auto ownership and mode choice
[Kockelman, 1997, pg. 180]
Land-use diversity & pedestrian-oriented tCervero & Kockeknan, 1997, pg. iS1]
designs reduce trip rates & encourage non-auto travel
Travel behavior depends on the nature and
range of travel choices available within a
neighborhood, rather than on the "neighborhood
design" per se (e.g., "new" versus "old")
M~xed-use development Is more important than
density in affecting non-motor~zed work trip
mode shares
Neighborhood design variables have little
impact on work trip mode choice, although a
"traditional neighborhood" dummy variable
showed some statistical significance
Land-use mix affects work trip mode choice
In suburban office centers
Neighborhood type not statistically significant
in explaining travel behavior once socio-economic
factors are accounted for
Individual motivations and Imitations central to
the decision to walk; urban fonn then encourages or
discourages walking, given the motivation to do so
[Handy, 1995, pg. iS3]
[Cervero, 1996, pg. 184]
[PBQD~ 1996a' pg. 186]
[Cervero, 1989' pg. 186]
[McNally & Kulkarni7 1997' pg. 188]
[Handy' 1996b' pg. 191]
'micro" considerations of neighborhood design. It is recognized however that the distinctions
~ , ,
between these categories tespeclally the last two) can' at times' be fairly arbitrary. For each paper
reviewed in Appendix A, a very brief summary of key findings is presented, alters with the paper
citation. Since these summaries may often not do justice to the nuances of the papers' findings, the
OCR for page 57
TCRP H-12 Final Report
number of the page where each paper is first discussed in Appendix A is also shown, so that the
reader Carl refer back to the more detailed discussion of Me paper if so desired.
Despite Me diversity of results presented in this long table several generalizations, "working
hypotheses", etc. can be drawn from this review. These are itemized and discussed briefly below.
1. Residential density. The empirical evidence concerning Me role which residential density plays
in the determination of transit usage (as well as the use of non-motorized modes of travel such as
walk and bicycle) is very mixed. The debate concerning the role of residential density in determining
urban transportation "efficiency" or "sustainability" is one which has been ongoing within the urban
and transportation planning communities for some time. The "pro density" argument is often
associated with the work ofKenworthy end Newman [1989], among others, in which density seems
to be the single most important factor explaining macro differences in transportation energy use, etc.,
among cities.
Many other studies however, including several cited in this review, find the evidence much
less clear. In particular, the role of density as a direct explanatory variable with respect to transit
usage, auto VMT, etc. typically declines significantly within the statistical analyses once "other
factors" such as socio-economic characteristics of the trip-makers, accessibility by mode to
destinations, auto availability, etc. are accounted for te.g., PMM, 1975; Kockelman, 1997; PBQD,
~ 996c, Messenger and Ewing, ~ 996; Loutzenheiser, ~ 997; Miller and Ibrahim, ~ 9981. This issue is
pursued in more detail in the following discussion of these "other factors".
2. Transit supply. Relatively few studies explicitly include measures of transit supply in their
analyses of urban form impacts, presumably due to data limitations. As noted above, however, when
such variables are included in the analysis, they often are found to play a significant role in
explaining modal choices, VMT levels, etc., and to reduce the explanatory power of density
measures within Me analysis tPBQD, ~ 996c]. In order to understand this result, one must note that
a classic demand-supply relationship exists between factors such as residential density and transit
service levels. That is, the better the transit service, the more people will use it, the more people
potentially available to use a given service, the higher the level of service which can be cost-
effectively provided.
Figure 3.~(a) illustrates this relationship between potential transit demand arid the supply of
transit service. The demand curves Do, etc., represent the potential demand in different corridors,
which is presumably determined by factors such as their densities, accessibilities by transit to desired
destinations, population socio-economic attributes, etc. The "supply curve", S. represents a simple
"scheduler's rule"; i.e., a simple model of the transit agency's scheduling process as a function of
expected demand. The points of intersection between the supply curve S and the various demand
curves, D, represent the equilibrium ridership levels that result from the transit agency's decisions
on how much service to supply in each comdor arid travelers' responses to these service decisions.
OCR for page 58
TCRP H-12! Final Report
Figure 3.!
Transit Demand-Supply Relationships
._
-
. _
o
;^
-
.
ct
a
._
u,
sit
.~
o
-
._
cot
/
-
~/
/
/
D4
D3
D2
-- D1
Number of Daily Bus Runs
(a) Transit Ridership and Service Supply - Demand Relationships
(aclapted from Gonzales, 1980)
Combor Residential Density
(b) Transit Ridership Density Relationships Implied by Figure 3.l (a)
- 34
OCR for page 59
TCRP H-17 Final Report
Assuming Mat average condor density is one of the determinants of the demand curves, the
equilibnum demand levels can be replotted versus average corridor density, as illustrated in Figure
3.~(b). While this figure is completely hypothetical. it does represent a plausible relationship
~ , . _ _ ~ ~ , , ,: , , ~
between ndership and density. In particular. it is consistent with the "threshold effects" which are
reported with some consistency in the literature te.g.. Pushkarev and Zuparl, 1977, 198O, Smith,
1984; PMM, 1973], in which transit usage is found to increase dramatically once certain density
thresholds are exceeded.
At least two important points relevant to the current discussion can be drawn Mom Figure 3. ~ .
First, and most fundamental note that the points shown in Figures 3. ~ (a) and (b) do not trace out
the demand curve for transit as a function of either Posit service level or residential density. Rather,
the pouts in Figure 3. ~ (a) trace out the transit operator's supply curve. while the specific shape of
the curve described by the points in Figure 3. ~ (b) depends upon the equilibrated supply responses
ofthe transit operator, along with all the other factors relevant to Me determination oftransit demand
(auto ownership levels, population socio-economic attributes, etc.). Identification of the transit
demand function (and, thereby, determination ofthe "true" relationship between ridership and service
levels, density, etc.) requires explicit consideration of both the demand arid supply processes ~ d
Me inclusion of the relevant variables associated with each of these processes -- within some form
of"simultar~eous equations" analysis.
In the absence of such a comprehensive, simultaneous analysis the results are inevitably
subject to ecological fallacies. aggregation errors, etc. of unknown but potentially major proportions,
rendering the models armor policy implications drawn Mom such partial analyses extremely suspect.
Again, to take the hypothetical but realistic case of Figure 3 . I: Fissure 3. ~ (a) does not provide us with
a basis for predicting hove ndership in any of the four corridors being modeled (let alone any other
corridors not included in the base data) will respond to a change in transit se} vice level. Similarly,
Figure 3. ~ (b) does not provide a reliable basis for predicting how- density- changes within a corridor
might affect transit demand, since the details of this curve are based upon the combination of a wide
range of factors, which cannot be expected to extrapolate readily to other situations.
A second. much more minor point to draw Dom Figure 3. ~ (a) relates to the relative statistical
significance of variables such as transit se - -ice level and density in explaining observed travel
behavior. In such a situation it is not surprising that if one regressed transit demand (or any other
equivalent measure, such as auto VMT) versus transit service variables and density (among other
factors). density would not appear to be overly significant given that the relationship being traced
by the data points is between eldership arid service level, with density entering the relationship orgy
indirectly (via the supply-demand relationship discussed above).
3. Auto ownership. One quite consistent finding in the literature reviewed is that households in
higher density neighborhoods tend to own fewer vehicles, and that households owning fewer cars
then tend to use transit more Ad generate less VMT tShimek, 1996; Kockelman, 1997; Cervero,
OCR for page 60
TCRP H-12 Final Report
1996; Messenger and Ewing' 1996].
The role of auto ownership within the overall transportation - land-use interaction is often
overlooked. This may reflect the assumption that auto ownership in many portions of the United
States is so high arid so pervasive that it ceases to be an ~nteresi~ng explanatory (or policy) cleanable.
It may also result Tom the view Mat auto ownership is just one more socio-economic descriptor of
trip-makers. determined largely exogenously to the travel decision-making process.
In our view, these assumptions significantly underestimate the extent to which auto
ownership decisions are integral to the transportation - land-use interaction. In particular, auto
ownership is a critical "intermediate link" between household location choices (where to live, where
to work) and their subsequent activity/trave} decisions. Households who choose to live and/or work
in low density suburban areas wall of necessity (if not also preference) be "auto oriented". tend to
have a high auto ownership levels and make most if not all DIPS of any significant distance bY auto.
~,
TY ~1 1 1 ~· 1 ~1 · 1
., · , 1
Households Who live anchor work In denser, trans~t-onenteo communities (where the transit-
orientation arises from the trar~sit/lar~d-use Interaction discussed above), may opt to own fever cars
(east., only one instead of two or more). Once a household decides to own, say, one less car, it by
necessity is committed to driving less and using other modes of travel (transit, warm more, if it is
going to maintain a comparable level of activity. relative to a household which owns that "extra" car.
Thus as in the transit service case previously discussed, a proper specification of the urban form
travel demand interaction requires including auto ownership as an endogenous component of the
system.
· · · , · · ~ · . , ~ , .
In add~i~om it is worm noting that marry policy issues of current relevance (hoar carbon taxes
to vehicle technology options) have direct or indirect Contacts on vehicle ownership decision-makinc.
A. . . . .. .. , ~ . . ..
, ,
l his Adam Implies the need tor ~nciuct~ng auto ownership as an endogenous mode] component, so
that the mode! can adequately respond to these types of policies.
4. Socio-economics. It is not in the least surprising that socio-economic factors (over and above
auto ownership' discussed above) such as income, age, gender. occupation, etc. have a sigmficant
impact on travel behavior tPMM, 1975; Schimek' 1996b. Loutzenheiser, 19973. It is well known
that people's travel needs and capabilities vary dramatically by- such factors. Indeed, the primary
reason why most current travel demand modeling methods (e.~., random utility models; activity-
based models) are developed at the disaggregate level of the individual tup-maker is so that they can
properly capture the diversity of behavioral responses which occur among different types of people.
Two points need to be made concerning the role of socio-economic factors within integrated
urban models. First, as with Me other factors discussed above it is the interaction between socio-
econorriics and urban fonn which is central to the urlderstanding and modeling of people's locational
and activity/~ave} decision-making. Different people will respond to different density levels/urban
designs in different ways. It is, therefore, not a question of "which is more important", density or
- 36
OCR for page 61
TCRP H-17 Final Resort
socio-economics. In explaining behavior. Rawer, it is a question of understanding how behavioral
responses to chances in density, etc. will vary by socio-economic characteristics.
Second, given the importance of socio-econom~c factors, it is imperative that they be
explicitly represented within our modeling. systems, and that our models be sufficiently
disaggreaated to properly capture their effects
.. ., . . ~. , .
_
- ^^ . This implies the need to include within model
systems explicit representations of demographic arid economic process. A strong case can be made
that one of the reasons why- many advanced disaggregate modeling methods have not yet achieved
wide-spread adoption within operational planning contexts is the inability to predict credibly the
detailed socio-economic attributes which they require.
S. Employment Density. Employment density impacts have not been investigated to the same
extent as the residential density case. The reported findings, however, are quite consistent: increased
employment concentrations do have significant impacts on transit usage, walking (where feasible)
and ride-sharing [Schimek, 1996a, Frank and Pivo, 1994; PBQD. 1996a; Cervero, 1989]. These
results tend to hold for central business districts (CBDs), suburban employment centers, and
employment centers located near commuter rail stations. This strong, clear result relative to the
more ambiguous residential density case is likely due to the more direct relationship which almost
certainly exists between employment density and transit service supply (i.e., such centers are readily
Rentable loci for transit services', as well as the "levels of service" for other modes as well (e.a.-
~ · ~ ~ . ~ · . - · .~ ~
higher employment densities Increase the chances of ride-share "matches"; higher density areas,
particularly In CBDs' tend to have higher parking costs and/or walk times from parking).
~an,
Perhaps the single most important Implication for integrated urban modeling is to reinforce
the importance of the employment/activity center end of the trip. There is a strong tendency in both
theory and practice to focus on the residential side of the land-use problem. The spatial distribution
of employment (and, more generally, out-of-home activities, both work and non-work related),
however, may well be a much stronger "driver" of travel behavior and transportation supply options.
It is also arguable that this aspect of land-use may be more susceptible to successful planning control.
6. Accessibility. For current purposes, the tenn accessibility is used here simply to mean a variety
of measures of "how well connected" a given location is with activities of a given type (work
opportunities, shopping destinations, etc.), usually in terms of how much of a given activity is
located how close to the location in question. Thus, for example, one can speak of the accessibility
of a residential zone to employment opportunities. A specific variant on the accessibility theme is
the question of "jobs-housing balance" which is concerned with the extent to which job opportunities
exist within some fairly localized region for a set of workers within a given residential area.
While the literature is somewhat mixed with respect to this issue -- particularly with respect
to the efficacy of jobs-housing balance on VMT, etc., generally the discussion is about how
important accessibility is relative to other factors' not whether it is important. Conceptually
~ 7
_ ~ I _
OCR for page 62
TCRP H-12 Final Report
accessibility is central to transportation planning -- it is. very s~mply~ what we are in the business of
providing. The accessibility of people to workplaces and other activities, the accessibility of jobs
to workers, of stores to their market must be of some relevance to the activity/travel process if this
process is the least bit rational.
Indeed, one of the reasons why the literature on the impact of factors such as residential
. .. .
.
density or neighborhood design is so mixed is that it tends to ignore the critical question of
connectivity: it is of little use having a dense neighborhood which does not have good access to
relevant activity destinations; transit requires travel corridors of reasonable density, consisting of
both "production" and "attraction" points to be most effective, waking requires close proximity
between origins and destinations, etc. This is not to sail Mats in areas possessing very high degrees
of connectivity, etc. accessibility may be so universally high that it may not seem to matter; this
simply represents one extreme point on the continuum of possible situations.
Linking origins with destinations to create flows is the fundamental task of travel demand
models. Understanding how urban form combines with the transportation system to provide
accessibility to activities and how people respond to accessibility in both their location and
activity/travel choices is central to this travel demand forecasting process. To address this issue
comprehensively clearly requires an integrated approach to the modeling of location choice and
travel behavior.
7. l\ieighborhoo(1 design. Particularly in the last fear years. marry researchers have focussed on
more "micro-level" questions concerning the role which local neighborhood design plays in
determining travel behavior. In particular, advocates of "neo-traditional neighborhoods" and town
planning have argued that such neighborhoods should encourage more walk and transit trios. shorter
trips. etc. thereby contributing to reductions in auto VMT' emissions etc.
- -r - ~
Me literature dealing with this issue which has been reviewed in this report is' again quite
mixed in its findings, again probably for many of the reasons previously cited (lack of
comprehensive analysis' methodological weaknesses data limitations' etc.). The key issue here
however. is almost certainly that our "activity- time-suace prisms" extend well beyond the local
nei abhorEc)od given both the levels of accessibility available to many people and their
~,
expectations/needs concerning their part1clpatlon in activities ot various types. Jobs-houslng
balances, for example, are virtually impossible to achieve in practice, given the nature and dynamics
of our complex labor markets. Similarly, our desire for the widest possible range of consumer
goods, experiences, etc. Bleary that our "action spaces" will inevitably extend well beyond the
neighborhood boundary.
This is not to say that the details of neighborhood design are not important. It is such design
details, after all, which determine the residential and employment densities within the neighborhood,
the ease and attractiveness of walking' and the ease and efficiency of providing transit services
~ O
- 30 -
OCR for page 63
TCRP H-17 Final Report
within the neighborhood. to name just a few- important issues. From a modeling point of view,
however, the key point is. again, the need for a comprehensive, integrated view of the problem --
neighborhood design within the overall transportation - land-use interaction -- which is cntical to
achieve, bow In order to few ur~derstand the nature of the interactions involved, arid then to generate
useful analyses and forecasts based upon this understar~ding.
Figure 3.2 represents one rather messy but still overly simplified summary of this discussion.
In this figure, act~vity/trave] behavior is shown to be We "outcome" of a complex set of interactions
among the various factors discussed above. This figure is intended to illustrate two key points:
1.
''Urbar1 form" or "larld-use" or "physical design" (as represented by residential
density, employment density and neighborhood designs provides a context for human
behavior, which, in this case, includes location decisions (residence, job locations),
auto ownership decisions, and. ultimately, activi07/trave! decisions. That is,
increased residential density does not directly "cause" reductions in auto VMT.
Rather, under the right circumstances, it may attract a resident population with
particular socio-economic charactenstics and desired activity patterns who Bill make
auto ownership and travel decisions that will result in increased transit/walL usage,
reduced VMT. etc. relative to what they might do in other urban form contexts.
Numerous "supply-demand" or "feedback" Interactions exist within the system (travel
decisions affect road congestion levels, which, in turn, affect travel decisions,
residential densities combined With attributes of the resident population affect the
level of transit service provided, which, in turn, affects the attractiveness of the
residential area for people of different types; etc.~. Ignoring these complex
interactions and ar~alyzing the system In a partial. overly simplified was- almost
inevitably leads to misleading or even erroneous results.
3.2.2 Impacts of Transit on Urban Form
To this point the focus has been on discussing urban form impacts on travel behavior.
Reversing the direction of causation, Table 3.2 provides a brief summary of the empirical evidence
conceding transit system impacts on urban foe which was discussed in greater detail in Appendix
A. In this case, findings are summanzed within the table by urban area given the "case studs-" nature
of this literature. As in Table 3. i. the citation for each paper reviewed is provided, as is the page
number in Appendix A where a more detailed discussion of the paper can be found.
Again, a diversity of findings is reported concerning the impacts of light rail, subway and
commuter rail lines and stations on residential density, employment density, property values, etc.
Without attempting to discuss or reconcile these results in any detail, there are at least four main
points to be drawn from Table 3.2 which have implications for integrated urban modeling.
~9
OCR for page 64
TCRP H-12 Final Report
Figure 3.2
Urban Form Impacts on Activity and Travel
~ \~
(Accessibility)
(Demographic)
-
-
- 40
,~
\
I ACTIVITY /
TRAVEL
OCR for page 65
TCRP H-17 Final Report
Table 3.2 Summary of Empirical Evidence Concerning
Transit Impacts on Urban Form
Atlanta
Population adjacent to raid stations
decreased; employment increased but
employment share decreased relative
to the rest of the region
San Francisco/Bay Area
After ~ year of operation, only 35% of B ART users
were former car users; 50% previously used bus
Transportation access to jobs seldom a factor in job
location choice; access to BART a minor consideration
in household location choice; BART unpact on
land-use after 5 years of operation was insignificant
After 20 year of operation, land-use unpacts
of BART focussed in downtown San Francisco and
Oakland' plus a few suburban stations; neighborhood
opposition and/or poor local real estate markets
prevented significant impacts elsewhere; general
freeway-oriented growth occurred despite BART's
presence; stronger public policy initiatives
would be required to generate a compact, multi-
centered urban fonn around BART
Property values increased as one moves closer to
a commuter rail station and decreased with
proximity to a freeway, for a selected B ART station
Boston
Nielson & Sanchez, 1997, pg. 198]
[Webber, 1976, pg. 196]
[Giuliano' 199S, pg. 196]
[Cervero & Landis, 1997, pg. 197]
[Workman & Brod. 1997, pg. 197]
Rail extensions/improvements had a strong [Knight & Trygg, 1977, pg. 199]
influence on development within North Quincy along with
other complementary factors; elsewhere, little impacts observed
- 41
OCR for page 66
TCRP H-12 Final Report
Table 3e2, contld
C~g~'
Residential location preferences strongly
affected by transportation system; significant
premium exists for housing within walking
distance of a light rail station (Calgary)
Chicago
Rail system improvements generated no
discernable land-use impacts, although one
must note He high level of CBD development
high cost of land, and the "mature" nature of
the existing system
Cleveland
Subway construction had minimal land-use
impact; note lack of complementary policies/factors
relative to the Toronto case
Glasgow.
One year after major rail upgrades, planning
applications in adjacent areas increased; some
reversal of population decline observed; fairly
small number of new induced trips and diversions from auto
Montreal
Nature and intensity of CBD retail activities
influenced by the Metro, but, as in Toronto,
other development factors were also of unpor~nce;
outside the CBD, little development impact
generally occurred.
tHunt, en al., 1994, pg. 28]
[Knight & Trygg, 1977. pg. 31]
[Knight & Trygg, 1977. pg. 31]
[Gentlemen, et al., 1983, pg. 29]
[Knight & Trygg, 1977, pg. 31]
- 42
OCR for page 67
TCRP H-12 Final Report
-
Table 3e2, cont~d
Philadelphia
Substantial increase in property values along
the Lindewold tine; influenced zoning and
actual investments ~ suburban office and
apa~-ln~ent buildings although growth along the line
was not higher than in other portions of the urban region
Portland
Property value Trip acts of Me Portland
light-rai! line difficult to identify
No statistically significant impact of
proximity to rail stations on property
values for Portland stations
Significant new development observed
adjacent to Portland light rail stations
Toronto
Significant development occured around many
stations; some evidence of increased proper
values and densities adjacent to stations;
however, much of this results from aggressive zoning,
joint public-private development, and, in general
carefully coordinated transit and land-use development
Washington, D.C.
1979-1982, 54% of nonresidential construction
in the metropolitan areas occured within 0.7 mi.
Of a Metro station.
Rail corridors have developed more than other
places, with majority of this growth occuring
Knight & Trygg, 1977, pg. 199]
"Workman & Brod, 1997, pg. 197]
[Al-Mosaind, et al., 1993, pg. 197]
Barrington, 1989, pg. 197]
Knight & Trygg, 1977, pg. 199]
[Baker, 1983, pg.198]
[CIreen and James, 1993, pg. 198]
- 43
OCR for page 68
TCRP H-12 Final Report
First, notable by their absence is any mention of land-use impacts of shared-nght-of-way bus
systems. HOV systems, etc. This reflects a widely held belief that it is only major fixed-guideway
infrastructure which wall have a discernable impact on urban form development. Bus routes can
charge on a mon~-to-month basis, HOV Cartes can generally be converted to general traffic usage
with minimal effort. Further, such systems (especially bus systems) are often fairly ubiquitous,
providing a "background, base" service over an extended area. Such flexible relativeiv ubiquitous
systems Carl provide reasonably high levels of accessibility, which, in nine influence location and
Ravel choices, but which, In and of themseIves. are assumed to be unlikely to stimulate major land
development decisions.
Second, many of the studies cited suffer Tom methodological and/or data limitations. Two
of these are particularly notable:
I. Many studies involve very short analysis time penods (the extreme being Webber's
1976 study of BERT impacts one year after beginning operations. Urban form
evolution, however, operates on a time scale of decades; short-run impacts are
inevitably negligible and short-run responses need not be indicative of Ton~-run
impacts. Practical difficulties exist, however, in long-run empirical studies.
. .
. . . . .
is,
First,
simply being able to observe a system over a very long penoa or time IS a major
undertaking. Second, over the Tong run many factors change. making unambiguous
determination of the transit system impact difficult, if not impossible.
All empirical studies are either "before-and-after" studies of the impact of a given
facility in a given location or (less commonly) a comparison of two "comparable"
locations, one in which the facility was constructed, and one in which it was not.
Both are intended to approximate a "w~th-and-without" analysis, which is what is
really required in order to sort out the actual impacts of the transit facility. Neither
the before-and-after or the comparable location approaches, however. provide an
adequately controlled "experiment" to oronerIv isolate the transit facility impact:
. ~. ~· .
],_ _ ~ ,, ~
~e~ore-ano-atter comparisons are Inevitably confounded by all the other evolutionary
factors at work within the given urban area and no two combers cities. etc. are ever
so "identical" that one can conclude that the only explanation for differences between
the two locations' development is the presence of the transit facility. It can be argued
that the only way to identify the "true" impacts of raid transit systems on a long-run,
with-and-without basis is through the use of an integrated transportation - land-use
model. That is, it is argued that it is only through the exercising of such a model that
one can trace the long-r~ evolution of an urban system within a controlled setting
(i.e., one Carl control other exogenous inputs into the urban system being modeled).
Further, with such a mode! one can test the sensitivity of the transit impacts to
changes in these "exogenous inputs" so as to explore the extent to which robust,
generalizable results Carl be obtained.
- 44
OCR for page 69
TCRP H-12 Final Report
. . .
Third, it is noteworthy that We focus of many of these studies is on land value. particularly
given the total absence of land or housing prices from the "urban form impact" literature summarized
in Table 3.1. Land development, building stock supply, and residential and commercial location
decisions all occur within economic markets, within which supply-demand processes (for land.
buildings, etc.) exist and are reconciled through the determination of market-clearing prices. The
development of land and the supply of building stock is primarily the task of private sector agents
seeking to maximize profits (within whatever "rules of the game" exist with respect to zoning, etc.),
subject to anticipated demand levels. The decision concerning where to live or where to locate one's
Finn depends in no small way on the prices of different building types at different locations. Given
the absolutely fundamental importance of market processes in both land development and location
choice, it is essential that these processes be explicitly included in any integrated model of
transportation and land-use, if a proper understanding of urban system dynamics and evolution is to
be achieved.
Finally, Me one absolutely clear result which emerges Mom Table 3.2 is that transportation
in general =d rail transit In particular is a facilitator of development, not a cause: a "necessary but
not sufficient condition" for development to occur. Although note- over twenty years old, the work
of Knight and Try go remains the most comprehensive and definitive study of this issue to this day.
In their 1977 study, they build a compelling case that transit ins estment is but "one piece of the
pu=]ett~ and Mat loch land-use policies, over government policies. the local and regional economic
climate, etc. all must interact in a mutually reinforcing way in order for positive land development
impacts to occur. Much earlier, Spen;,ler L1Y301 came to very star conclusions in his study ofthe
impact of the New York subway system on land values dunng the period 1909-1930. As
summarized by Libicki t1 975], these conclusions w-ere:
"a. transit served to shift as much as augment land values;
b. neighborhoods which were 'marked out' for certain Acres of development and already
so developed would show little increase in land value;
most of the large increase in land value arose not from the rail transit system, but
from subdividing arid building ore land;
d.
e.
declining neighborhoods did not stop declining, stag t neighborhoods did not stop
stagnating and the character of art area around a station strongly resembled the
character of the surrounding neighborhood in an`; case, so that consequently;
rather than to be considered a cause of land value changes, a transit facility should
more properly be regarded as a construction which permits or facilitates under certain
circumstance. an emergence of tared values, the values being determined largely by
other factors." ~LJibicki, 1975, pp. 6-7]
- 4:
OCR for page 70
TCRP H-12 Final Report
Figure 3.3 is taken Tom Knight and Trygg [1977] and summarizes their findings concerning
the land development process. This figure illustrates both the complexity of this process and the role
which transit ~nvestrnent can play within it. Some of the actors and their interactions shown in this
figure Carl (at least In principle) be modeled within an Integrated urban model, some almost certainly
can not. The lesson, however is Me same as has been stated at venous points throughout this section:
the need for integrated urban models as the only to feasible method for properly analyzing and
forecasting the land development process.
- 46
OCR for page 71
| Ease of Private
L Assembly! I
I Infrastructure |
| Capacity
. Public
Facilities
| Urban l
| Renewal ~OTHER NEW
NEARBY LAND
Prig ate ~ INVESTMENTS
Dcvclopment
l
cighborhood ~LOCAL LAND
I Aninades I USE POLICIES
_ 1 /
Zoning & / I
Development
Incentives
TCRP H-12 Final Report
Figure 3.3
Factors Influencing Land-Use Impact
(Source: Knight aIld TIygg t19774)
Cost of Land
&` Site
Preparation
Public Assembly
Activities
· Urban Renewal
Transit Excess
(Young
Amend,T~cut)
i
AVAILABILITY OF
DEVELOPABLE
LAND Phvsical
. Blimp
\ ATTRACTIVENESS
OF SITE FOR
\ DEVELOPMENT
COMM}'TMENT ~ ~ ~ ~_
ITO SPECIFIC J ~ ~ | DECISION IS TO |~ IMPACT |
OF TRANSIT ~_
IMPROVEMENT ~ IMPROVEMENT IN
ACCESSIBILITY'
~ /,
l OTHER
/ . GOVERNMENT
~, / POLICIES ~ ~ Taxation and 1
Goals of Larger | / ~ other l
Community, ~ | \ | Assessment
Growth I ~
Social Character | I Compliance | ~nfras~uculrc |
Plan & Dollar I | withother | | Provisions |
Pnont~es ~ I Public I
I Programs
I Environmental
| Impact Review
~ · Equal
| Opponunit~
- 47
|REGION S DEMAND|
I FOR NEW L
| DEVELOPMENT
L ~ | Economy |
Regional |
I Economy I
OCR for page 72
TCRP H-12 Final Report
- 48
Representative terms from entire chapter:
auto ownership
