Peng 2017

JID: TRB
ARTICLE IN PRESS [m3Gsc;March 31, 2017;0:19]
Transportation Research Part B 0 0 0 (2017) 1–22
Contents lists available at ScienceDirect
Transportation Research Part B

journal homepage: www.elsevier.com/locate/trb
Transit-oriented development in an urban rail transportation

corridor
Ya-Ting Peng a, Zhi-Chun Li a,∗, Keechoo Choi b
a
School of Management, Huazhong University of Science and Technology, Wuhan 430074, China
b
Department of Transportation Engineering, Ajou University, San 5 Woncheon-Dong, Yeongtong-Ku, Suwon, 442-749 South Korea
a r t i c l e i n f o a b s t r a c t
Article history: Transit-oriented development (TOD) has been recognized as an important avenue for cre-
Received 30 June 2016 ating a green transportation system. This paper addresses TOD investment issue in terms
Revised 21 March 2017
of the location, number and size of the TOD zones along a rail line. An urban system equi-
Accepted 21 March 2017
librium problem with TOD investment is first formulated. Two social welfare maximization
Available online xxx
models, which take into account different investment regimes for TOD projects (i.e., public
Keywords: and private), are then proposed for optimizing TOD investment schemes along a rail line
Transit-oriented development and train service frequency on that line. In the public regime model, the government is re-
Rail transportation corridor sponsible for the investment cost of TOD projects, which is borne by the private property
Households’ residential location choices developers in the private regime model. The proposed models explicitly consider the inter-
Housing market actions among the government, property developers and households in the urban system,
Public and private regimes together with the effects of the TOD investment on households’ residential location choices
Social welfare and housing market. The population thresholds for investing in a TOD project under the
public and private regimes are also identified. The findings show that the TOD investment
can cause population agglomeration at the TOD zones and a compact city; households and
the society can benefit from the TOD investment; and the private TOD investment regime
outperforms the public regime in terms of total social welfare of urban system.
© 2017 Elsevier Ltd. All rights reserved.
1. Introduction
In response to rapid growth in the number of motorized vehicles and thus in traffic congestion, transit-oriented devel-
opment (TOD) has been suggested as an effective tool for mitigating auto-related problems (e.g., traffic congestion and en-
vironmental issues) and controlling excessive urban sprawl (Cervero et al., 2004; Cervero and Day, 2008; Papa and Bertolini,
2015). TOD refers to medium- and high-density housing along with complementary public uses, jobs, retail, and services in
mixed-use development around transit stations (Calthorpe, 1993). The TOD investment projects, as an important avenue for
creating a green transportation system, have received considerable attention from the relevant authorities of many coun-
tries or regions in the world, such as United States, South Korea, and Hong Kong (Lund, 2006; Cervero and Day, 2008; Loo
et al., 2010; Sung and Oh, 2011). Recently, the Chinese government has been developing the TOD projects in some densely
populated cities, such as Guangzhou, Shenzhen, and Wuhan.
The TOD schemes can attract households to reside nearby transit stations by providing them with amenities for conve-
nience, enjoyment, or comfort, and thus improve the land-use efficiency and the value of properties nearby transit stations
∗
Corresponding author.
E-mail address: smzcli@gmail.com (Z.-C. Li).
http://dx.doi.org/10.1016/j.trb.2017.03.011
0191-2615/© 2017 Elsevier Ltd. All rights reserved.
Please cite this article as: Y.-T. Peng et al., Transit-oriented development in an urban rail transportation corridor, Trans-
portation Research Part B (2017), http://dx.doi.org/10.1016/j.trb.2017.03.011
JID: TRB
2 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22
Table 1
Contributions to the land use and transportation models.
Household’s residential Housing Transit service TOD design in terms of TOD investment regime
Citation location choice market optimization location, number and size (public or private)
Eliasson and Mattsson (20 0 0) × × × ×

Chang and Mackett (2006) × × × ×
Martínez and Henríquez (2007) × × ×
Li et al. (2012b) × ×
Li et al. (2012c) × × × ×
Li et al. (2013) × × ×
Ma and Lo (2013) × ×
Li et al. (2015) × × ×
Ng and Lo (2015) × × ×
Wang and Lo (2016) × × × ×
This paper
Note: “” means that the associated item is considered, whereas “×” means that the associated item is not considered.
(Kay et al., 2014). The pedestrian-friendly environment inside TOD zones encourages residents to commute by public transit
modes, which can effectively alleviate the auto-related problems (Cervero, 2007; Loo et al., 2010). However, the development
of TOD neighborhoods around transit stations may be costly due to investment in the public amenities inside the TOD zones
(e.g., pedestrian lane design, school, hospital, shopping centers and sports facilities) (for more details, see Nelson and Niles,
1999; Lee et al., 2016). This leads to a tradeoff between cost and benefit generated by the TOD investment projects.
In the TOD projects, the location, number and size of the TOD zones along a rail transit line play an important role
in the cost and benefit analyses of the TOD investment. As a matter of fact, different locations, numbers or sizes of TOD
zones on a rail transit line imply different residential distributions along that rail line and thus different passenger demands.
They also imply different housing supplies and land development costs for the TOD facilities due to differential land values
along the rail line and thus different costs and benefits of the TOD investment projects. Therefore, there is indeed a need
to determine the optimal location, number and size of the TOD zones on a rail transit line such that the TOD investment
projects are economic viable and cost-effective from an input-output perspective. The present study addresses the TOD
investment issues for strategic planning purposes.
In the literature, there are some studies involving TOD design or investment issues (see e.g., Cervero, 1994; Bernick and
Cervero, 1997; Cervero and Kockelman, 1997). However, these previous related studies mainly focused on the strategies and
principles/rules for TOD design. For example, Cervero and Kockelman (1997) summarized the TOD strategies and design prin-
ciples as high-density, mixed land use and pedestrian-friendly development around transit stations. Lund (2006) conducted
a survey to find out the motivations that people move into the TOD zones and their travel mode choices. Cervero (2007) ad-
dressed the effects of TOD investment on the transit ridership of residents inside the TOD zones by using a statistical anal-
ysis method. Loo et al. (2010) revealed the important factors of affecting the transit ridership inside a TOD zone through
a comparison of the case studies of New York and Hong Kong. Kay et al. (2014) carried out a hedonic regression analysis
to examine the effects of the TOD investment and the amenities provided by the mixed land development on residential
properties nearby the transit stations. However, these existing relevant studies usually adopted empirical and/or statistical
approaches to address the TOD investment related issues, and few studies adopted an analytical modeling method. An ex-
ception is Lin and Gau (2006), which developed a multi-objective model to determine the optimal investment intensity (or
the ratio of floor space to site space) inside a given TOD zone. However, their study only focused on a specific TOD station
on a rail line, and did not concern the optimization problems of the location, number and size of the TOD zones along a rail
line. The present study aims to fill the gap by developing an analytical model for investigating the TOD investment issues
along a rail corridor.
In the literature, the interactions between urban land use and transportation system have been widely studied. For the
convenience of readers, we have provided in Table 1 some principal contributions to the land use and transportation re-
search, in terms of household’s residential location choice, housing market, transit service frequency/fare optimization, and
the TOD investment. Table 1 shows that the previous related studies have mainly focused on urban land use (in light of res-
idential location choice and housing market) and/or transit service optimization issues. In particular, the studies of Li et al.
(2012b), Ma and Lo (2013) and Li et al. (2015) incorporated the effects of rail line investment on the value of properties,
households’ residential location choices and housing market along the rail line. However, these relevant studies were not
concerned with the TOD design problems in terms of the location, number, and size of TOD zones, together with the in-
teractions among the TOD investment, households’ residential location choices, and housing market. Some empirical studies
(see e.g., Lund, 2006; Hess and Almeida, 2007; Duncan, 2011; Dröes and Rietveld, 2015) have shown that the TOD schemes
can affect households’ residential location choices and passenger demand distribution along the rail line, which in turn af-
fect the decisions on the TOD investment and the train service frequency along the rail line. It is, therefore, important to
consider such interactions among the TOD investment, households’ residential location choices and train service frequency
in the TOD design problems.
JID: TRB
Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 3
Fig. 1. The rail line configuration and TOD zones along the rail corridor.
The government
Objective: maximizing social
welfare of urban system
(Decision variables: location,
number and size of TOD zones, and
train headway)
Land value, passenger

TOD design scheme
demand, utility level,
and train headway
and city boundary
Housing market equilibrium
Households Property developers

Objective: maximizing utility Housing demand Objective: maximizing net profit
(Decision variables: residential (Decision variables: capital
location and housing/non-housing Housing supply investment density)
consumption)
Fig. 2. Interactions among three types of agents in the urban economy.
In addition, these existing related studies never concern the source of the capitals for TOD projects. The TOD facilities
usually consist of housing and amenities (e.g., shopping centers or sports facilities that are provided for people’s conve-
nience, enjoyment, or comfort). In the cities of mainland China, the housing inside the TOD zones is usually invested and
built by the private property developers. The amenities (e.g., shopping centers or sports facilities) inside the TOD zones, as a
public infrastructure, are usually invested by the government. However, there are also some cases that the amenities inside
the TOD zones are invested by the private property developers. In this paper, the investment regimes that the amenities
are invested by the government and by the private property developers are referred to as the public and private invest-
ment regimes, respectively. It is, thus, meaningful to examine the effects of the amenities’ investment regimes on the TOD
investment decisions. For ease of presentation, “TOD’s amenities” and “TOD”, and thus “amenities’ investment” and “TOD
investment” are used without difference in this paper.
In view of the above discussions, this paper proposes analytical models for investigating the TOD investment issues in a
rail transportation corridor. The problems to be addressed in this paper are stated as follows: given a rail line configuration
(as shown in Fig. 1), how does the government determine such decision variables as the optimal location, number and size
of the TOD zones along a rail line and the train service headway on that rail line so as to maximize the social welfare of
the urban system? What are the effects of the TOD investment regimes on the TOD design?
The main contributions of this paper are twofold. First, we present an urban system equilibrium problem with the TOD
investment consideration, in which the residents inside and outside of the TOD zones are differentiated by their utility
functions. Compared to the residents living outside the TOD zones, the residents living inside the TOD zones enjoy an extra
amenity utility due to a better accessibility to the train stations and the constructed amenities inside the TOD zones. The
households’ residential location choices, land value across the city, and the housing market in terms of housing rental price
and space can be endogenously determined by the urban system equilibrium problem. Second, TOD design models under
the public and private investment regimes are proposed, respectively. In the public and private regimes, the TOD investment
cost is, respectively, borne by the government and the private property developers. The interactions among the government,
property developers, and households are explicitly considered, as indicated in Fig. 2. Specifically, the government seeks to
maximize the total social welfare of the urban system by determining the optimal location, number and size of the TOD
JID: TRB
Fig. 3. The shape of amenity function (x).
zones along a rail line corridor and the optimal train headway on that rail line. The private property developers aim to
maximize their own net profit by providing the residents (or households) with housing. Households choose their residential
locations to maximize their own utility within their budget constraints. The effects of the TOD investment on the urban
system are examined, together with a comparison of public and private TOD investment regimes and no TOD investment.
Sensitivity analyses of some factors that affect the TOD investment decisions are also conducted, including the TOD invest-
ment cost and the total number of households.
The remainder of this paper is organized as follows. The next section describes some basic assumptions. Section 3 for-
mulates basic components of the models, including passenger travel cost, passenger demand for each railway station, house-
holds’ residential location choices, property developers’ housing production behavior and housing demand-supply equilib-
rium. Section 4 presents TOD design models under different investment regimes for determining the optimal location, num-
ber, and size of the TOD zones and the train headway on the rail line. A heuristic solution algorithm for solving the proposed
models is also proposed. In Section 5, a numerical example is provided to illustrate the applications of the proposed models
and the model properties. Finally, Section 6 provides conclusions and recommendations for further studies.
2. Assumptions
To facilitate the presentation of essential ideas without loss of generality, this paper makes the following basic assump-
tions.
A1. The city is assumed to be linear, closed, and monocentric, which implies that the population of the city is exoge-
nously given and fixed and all job opportunities are located in a highly compact central business district (i.e., CBD). It is
also assumed that the land is owned by absentee landlords. The value of land at/beyond the city boundary equals the agri-
cultural rent or opportunity cost of the land. These assumptions have been widely adopted in the previous studies (see e.g.,
Alonso, 1964; Muth, 1969; Mills, 1972; Fujita, 1989; O’ Sullivan, 20 0 0; McDonald, 20 09; Li et al., 2013, 2015; Li and Peng,
2016; Li and Guo, 2017).
A2. There are three types of agents in the urban economy: the government, households and private property developers.
The government aims to determine the location, number and size of TOD zones along a rail line corridor and the train
headway to maximize the total social welfare of the urban system. The private property developers seek to maximize their
own net profit generated by housing supply. Each property developer follows a Cobb-Douglas housing production function
(see e.g., Beckmann, 1974; Quigley, 1984; Li et al., 2013, 2015; Li and Guo, 2017).
A3. All households are assumed to be homogeneous, implying that the income level is identical for all households. Each
household has a quasi-linear form of utility function (see e.g., Song and Zenou, 2006; Kono et al., 2012). A household’s
income is spent on transportation, housing and non-housing goods. The objective of each household is to maximize its
own utility by choosing a residential location, size of housing space, and amount of non-housing goods within its budget
constraint (see e.g., Solow, 1972, 1973; Beckmann, 1969, 1974; Anas, 1982; Fujita, 1989).
A4. The TOD facilities provide the residents living in the TOD zones with amenities for convenience, enjoyment, or com-
fort. The amenity level decreases with the distance from the train station (i.e., the center of the TOD zone). In this paper,
the following form of amenity function is adopted: (x ) = 1 + a1 e−a2 |D j −x| , where Dj is the distance from TOD station j
to the CBD and a1 and a2 are positive parameters, which can be calibrated by survey data (see e.g., Wu and Plantinga,
20 03; Wu, 20 06; Kovacs and Larson, 2007). The amenity function is discontinuous with the TOD’s boundary as discontinu-
ous/interruption point, as shown in Fig. 3.
JID: TRB
A5. This study mainly focuses on the commuters’ home-based work trips, which is a compulsory (or obligatory) activity,
and thus the household’s number of trips is not affected by various factors, such as the household’s income level and/or
housing rents. The average number of commuters (or workers) per household is assumed to be exogenously given and fixed,
and is represented as η. Every day, each worker makes a round commuting journey between his/her place of residence and
workplace located in the CBD. Thus, the average daily number of trips to the CBD per household is η. For example, η = 1
means that each household makes an average of one trip to the CBD per day. This assumption of one worker per household
has also been adopted by the previous related studies (see e.g., Anas and Xu, 1999; Song and Zenou, 2006; Li et al., 2013,
2015; Li and Guo, 2017).
A6. An exponential elastic demand function is used to capture passengers’ responses to the quality of the rail transit
service (Ortuzar and Willumsen, 2001), which is measured by a generalized travel cost that is a weighted combination of
the access time to station, waiting time at station, in-vehicle time, and fare. The responses include the decision to switch
to an alternative transportation mode (e.g., auto or bus) and the decision not to make the journey at all (Lam and Zhou,
20 0 0; Li et al., 2012c). In order to capture the effects of the TOD’s amenities, without loss of generality it is assumed that
the passengers residing inside the TOD zones are less sensitive to the rail travel cost than those who reside outside the TOD
zones (Lund et al., 2004; Lund, 2006; Cervero, 2007).
3. Components of the models
Consider a linear rail line corridor in which TOD projects may be introduced. Referring to Fig. 1, the rail line is described
by an ordered sequence of stations, represented as {1, 2, …, M + 1}. Ds represents the distance between station s and the
CBD, and D1 represents the length of the rail line. The TOD investment divides the rail line corridor into two kinds of
regions: internal regions inside the TOD zones and external regions outside the TOD zones. We denote J as the catchment
area of all the internal regions and J¯ as the catchment area of all the external regions. Let B be the distance of the city
boundary from the CBD (i.e., the corridor’s length). Thus, J ∪ J¯ = B holds. Let Xs be a 0–1 indicator variable which represents
whether station s is selected as a TOD station, s = 1, 2,…, M. Xs = 1 if station s is selected as a TOD station, and Xs = 0
otherwise. Let s be the size (or radius) of TOD station s. If station s is a TOD station, then the catchment area of TOD
station s can be represented by [Ds − s ,Ds + s ]. Consequently, J can further be expressed as
s=1 [Xs (Ds − s ), Xs (Ds + s )].

J = ∪M (1)
In the following, we in turn formulate the passenger travel cost, passenger demand for each station, households’ residen-
tial location choices, property developers’ housing production behavior and housing market equilibrium.
3.1. Passenger travel cost
Let x denote the distance of a location from the CBD and c(x) denote the travel cost of a passenger from location x to
the CBD. It comprises the walking/access time to the railway station, the wait time at the station, the in-vehicle travel time
and the fare, expressed as
c (x ) = τa As (x ) + τwWs + τt Ts + Fs , ∀x ∈ [0, B], s = 1, 2, . . . , M, (2)

where As (x) is the average passenger access time to station s from location x, which is dependent on the distance of location
x from station s. Ws is the average passenger wait time at station s. Ts is the average passenger in-vehicle time from station
s to the CBD. Fs is the fare for traveling from station s to the CBD. τ a , τ w , and τ t are the values of the access time, wait
time, and in-vehicle time, respectively.
The average passenger access time As (x) depends on the walking distance between location x and station s and the
average passenger walking speed Va . It is expressed as
|Ds − x|
As ( x ) = , ∀x ∈ [0, B], s = 1, 2, . . . , M. (3)
Va
With an assumption of constant headway between trains and a uniform random passenger arrival distribution (Lam and
Morrall, 1982), the average passenger wait time at station s, Ws , can be estimated by
Ws = 0.5H, ∀s = 1, 2, . . . , M, (4)
where H is the train headway.
The passenger in-vehicle time, Ts , from station s to the CBD can be calculated by
Ds
Ts = + κ0 (M + 1 − s ), ∀s = 1, 2, . . . , M, (5)
Vt
where Vt is the average train travel speed, and κ 0 is the average train dwell time at a station, which can be calibrated by
survey data. The first term on the right-hand side of Eq. (5) is the non-stop line-haul travel time along the corridor and the
second term is total train dwell time at all stations on the corridor.
JID: TRB
The train ticket fare Fs from station s to the CBD is defined as a linear function of the distance of station s from the CBD,
similar to Li et al. (2012b, c). It is represented as
Fs = f0 + f Ds , ∀s = 1, 2, . . . , M, (6)
where f0 and f are the fixed and variable components of the train ticket fare, respectively.
In light of the above, the annual passenger cost, ϕ (x), at location x can be expressed as
ϕ (x ) = 2ρ c(x ) = 2ρ (τa As (x ) + τwWs + τt Ts + Fs ), ∀x ∈ [0, B], s = 1, 2, . . . , M, (7)
where “2” denotes a round trip between the CBD and location x, and ρ is the average number of trips to the CBD per
household per year, which can be calibrated by survey data.
3.2. Passenger demand for each station
Note that any two consecutive stations on any segment of the rail line are competing for passengers between those two
stations. Consequently, there is a passenger watershed line that divides the line segment between two adjacent stations into
two sub-segments (see Fig. 1). The passengers in the two sub-segments use the upstream and downstream stations of the
line segment, respectively. Let ls be the passenger watershed line between stations s and s + 1, and Ls be the distance of the
passenger watershed line ls from the CBD. According to Vuchic and Newell (1968), Vuchic (1969) and Li et al. (2012b), the
passenger shed line ls is located so that the walking time from the shed line to the downstream station s + 1 equals the
sum of the walking time from the shed line to the upstream station s and the riding time from station s to s + 1, that is,
Ls − Ds+1 Ds − Ls Ds − Ds+1
= + , ∀s = 1, 2, . . . , M. (8)
Va Va Vt
From Eq. (8), we obtain
Vt + Va Vt − Va
Ls = Ds + Ds+1 , ∀s = 1, 2, . . . , M, (9)
2Vt 2Vt
where DM + 1 = 0. The catchment area of station s is thus [Ls ,Ls − 1 ] for any s = 1, 2,…, M.
Let q0 (x) be the daily density of potential (latent) passenger demand (i.e., the potential number of passengers per unit of
distance) at location x. According to A5, the average number of daily trips per household is η. Hence, we have q0 (x) = ηn(x),
where n(x) is the household residential density at location x defined later. The passenger demand for the rail line service
is elastic because it is sensitive to the travel cost by rail. Following A6, an exponential elastic demand density function is
adopted to model the effects of the passenger demand elasticity. However, the passengers residing outside the TOD zones
are more sensitive to the rail travel cost than those residing inside the TOD zones. The elastic demand density functions
inside and outside the TOD zones are specified as

q0 (x ) exp (−ω1 c (x ) ), ∀x ∈ J,
q (x ) = (10)
q0 (x ) exp (−ω2 c (x ) ), ∀x ∈ J¯,
where q(x) is the daily density of (actual) passenger demand for rail transit travel at location x. ω1 and ω2 are parameters
that reflect the demand sensitivity to the rail travel cost inside and outside the TOD zones, respectively. Herein, ω 1 < ω2
holds, implying that the demand elasticity of the passengers for rail service inside the TOD zones is lower than that outside
the TOD zones, as assumed in A6.
Hence, the daily number of passengers boarding trains at station s, denoted as Qs , can be calculated by
Ls−1
Qs = q(x )dx, ∀s = 1, 2, . . . , M, (11)
Ls
where Ls can be given by Eq. (9) and L0 = B.
3.3. Households’ residential location choices
We first define the households’ utility function. As stated before, the TOD facilities provide residents living inside the TOD
zones with public amenities (e.g., shopping centers or sports facilities), which incur an extra utility (called amenity utility
in this paper) to the residents inside the TOD zones compared to those living outside the TOD zones. Following Song and
Zenou (2006) and Kono et al. (2012), we assume in this paper that the household utility function takes a quasi-linear form,
but with an additional consideration of the amenity utility for the households inside the TOD zones. The households’ utilities
are specified as
U (x ) = z(x ) + α log g(x ) + β log (x ), α , β > 0, (12)
where U(x) is the utility of the households at location x. z(x) is the consumption of non-housing goods at location x, of
which the price is normalized to 1. g(x) is the consumption of housing goods at location x, measured in square meters
of floor space. α and β are positive constants. As specified in A4, the amenity level, (x), at location x is determined
by
JID: TRB

1 + a1 e−a2 |D j −x| , ∀x ∈ J,
(x ) = (13)
1, ∀x ∈ J¯,
where a1 and a2 are positive parameters, which can be calibrated by survey data. Note that when the amenity level, (x),
at location x is 1 (i.e., the locations outside the TOD), the amenity utility, β log (x), at location x equals zero.
According to A3, each household within the city chooses a residential location and housing / non-housing consump-
tions that maximize its utility subject to the income budget constraint. The household utility maximization problem can be
expressed as
max U (z, g) = z(x ) + α log g(x ) + β log (x ), (14)
subject to
z ( x ) + p( x ) g ( x ) = Y − ϕ ( x ) , ∀x ∈ [0, B], (15)
where p(x) is the average annual housing rental price per unit of housing area at location x, Y is average annual income of
households in the city, and ϕ (x) is the average annual travel cost of a household living at location x.
Substituting z(x) in Eq. (15) into Eq. (14) and setting the derivative of U(•) with regard to g equal to zero (i.e., dU /dg = 0),
one obtains
p( x ) g ( x ) = α , (16)
and
z (x ) = Y − ϕ (x ) − α . (17)
When the households’ residential location choice equilibrium state is reached, all households in the city have the same
utility level regardless of their residential locations. Let u be the common utility level of all households at the equilibrium
state. Substituting Eq. (17) into z(x) + α log g(x) + β log (x) = u, we have
1
g(x ) = exp (u − Y + ϕ (x ) + α − β log (x )) . (18)
α
Substituting Eq. (18) into Eq. (16), we obtain
1
p(x ) = α exp − (u − Y + ϕ (x ) + α − β log (x )) . (19)
α
Eqs. (17) and (18) describe the equilibrium consumption of non-housing goods and the equilibrium amount of housing
floor space per household at location x, respectively. Eq. (19) defines the equilibrium housing rental price per unit of housing
floor space at location x.
Given the common utility level u, TOD investment scheme and the train headway, one can then obtain the rail travel cost
ϕ (x) by Eq. (7) and the amenity level (x) by Eq. (13). Subsequently, the equilibrium consumption of non-housing goods
z(x), the equilibrium amount of housing floor space per household g(x), and the equilibrium housing rental price p(x) per
unit of housing floor space at location x can be determined by Eqs. (17)-(19), respectively.
3.4. Property developers’ housing production behavior
According to A2, the following Cobb-Douglas form of the housing production function is adopted by the property devel-
opers.
θ
h ( S ( x ) ) = θ1 ( S ( x ) ) 2 , θ1 , θ2 ∈ (0, 1 ), x ∈ [0, B], (20)
where h(S(x)) represents the housing supply per unit of land area at location x, S(x) denotes the capital investment per unit
of land area at location x, and θ 1 and θ 2 are positive parameters.
3.4.1. Property developers’ profit under public TOD investment regime
As previously pointed out, the public TOD investment regime in this paper implies that the TOD project is invested by the
government. That is, the government is responsible for the TOD investment cost (i.e., the construction cost of amenities in-
side the TOD zones). Therefore, under the public regime, the TOD investment cost is included in the government’s objective
of social welfare maximization, which is defined later. The property developers under the public investment regime receive
the revenue from housing supply along the rail line and pay the land rent and the opportunity cost of the capital used for
housing construction. The net profit of the property developers under the public TOD investment regime, represented by
ψ PUB (x), is thus expressed as
ψPUB (x ) = p(x )h(S(x ) ) − (r (x ) + kS(x )), ∀x ∈ [0, B], (21)
where the rental price per unit of housing space, p(x), is given by Eq. (19). k is the price of capital (i.e., the interest rate).
The term p(x)h(S(x)) on the right-hand side of Eq. (21) denotes the total revenue received by housing supply. These two
terms, r(x) and kS(x), in the bracket are the land rent cost and the capital cost, respectively.
JID: TRB
3.4.2. Property developers’ profit under private TOD investment regime
For the private investment regime, the TOD investment cost is undertaken by the private property developers. The TOD
investment cost is closely related to the value of the land used for constructing the amenities and the amenity level. In
general, the higher the value of the land or the amenity level is, the higher the TOD investment cost is, and vice versa.
Without loss of generality, we assume that the TOD investment cost at location x inside a TOD zone is a linear function of
the land value and the amenity level at that location. Specifically, the TOD construction cost π (x) at location x is given by
π (x ) = c0 + c1 r (x ) + c2 (x ), ∀x ∈ J, (22)
where r(x) is the land value at location x, which will be defined later. c0 is the fixed component of the TOD investment cost.
c1 and c2 are the variable components of the TOD investment cost, which are relevant to the land value and the amenity
level, respectively.
Let ψ PRI (x) represent the net profit of the property developers due to the provisions of the housing and/or TOD amenities
at location x under the private investment regime. For a location x within a TOD zone, the private property developers have
to pay the TOD investment cost. ψ PRI (x) can be defined as

p(x )h(S(x ) ) − (r (x ) + kS(x )) − π (x ), ∀x ∈ J,
ψPRI (x ) = (23)
p(x )h(S(x ) ) − (r (x ) + kS(x )), ∀x ∈ J¯,
where the TOD investment cost, π (x), at location x is defined by Eq. (22).
3.4.3. Profit maximization problem for property developers
From A2, the private property developers in the housing market aim to determine the capital investment intensity to
maximize their own net profit. The profit maximization problem under the public or private regime can be expressed as
max ψPUB (S ) or ψPRI (S ). (24)

S
The first-order optimality condition of maximization problem (24) yields
p(x )θ1 θ2 Sθ2 −1 − k = 0. (25)
From Eq. (25), we obtain the equilibrium capital investment intensity as follows.
1 / ( 1 − θ 2 )
S(x ) = p(x )θ1 θ2 k−1 . (26)
It should be pointed out that the expression for the equilibrium capital investment intensity S(x) is identical for the
public and private TOD investment regimes. However, their values for a given location x are generally different because the
value of the housing rental price p(x) at location x is generally different for these two regimes. This will be illustrated in the
numerical study section.
Note that under the perfect competition, the property developers earn zero profit (i.e., ψ PUB = ψ PRI = 0). The equilibrium
land values under the public and private investment regimes can thus be, respectively, expressed as
1 1/(1−θ2 )
rPUB (x ) = k −1 p(x )θ1 θ2 k−1 , ∀x ∈ [0, B], (27)
θ2
and
⎧ 1/(1−θ2 )
⎨ k 1 − 1 p(x )θ1 θ2 k−1 − c0 − c2 (x ) /(1 + c1 ), ∀x ∈ J,
θ2
rPRI (x ) = (28)
⎩k 1 − 1 p(x )θ θ k−1 1/(1−θ2 ) , ∀x ∈ J¯.
θ2 1 2
From Eqs. (27) and (28), one can immediately obtain the following property, which reveals the relationships among the
land value, the housing rental price, and the TOD investment cost.
Proposition 1. (i) For the public and private TOD investment regimes, as the housing rental price p(x) increases, the land value
r(x) increases, and vice versa. (ii) For the private TOD investment regime, given the housing rental price p(x), as c0 or c2 in the
TOD investment cost function (22) increases, the land value r(x) inside the TOD zones decreases, and vice versa.
The proof of Proposition 1 is provided in Appendix A. It should be pointed out that the effects of c1 in the TOD invest-
1/ (1−θ2 )
ment cost function (22) on the land value r(x) is ambiguous because the relationship between k( θ1 − 1 )( p(x )θ1 θ2 k−1 )
2
and c0 +c2 (x) is indeterminate.
JID: TRB
3.5. Housing demand-supply equilibrium
The equilibrium household residential density, n(x), at location x according to Eqs. (18), (19), (20) and (26) can be given
by

h (S (x ) ) u − Y + α + ϕ (x ) − β log (x )
n (x ) = = σ exp − , (29)
g( x ) α ( 1 − θ2 )
θ
where σ = (θ1 (αθ2 k−1 ) 2 )1/(1−θ2 ) is a constant.
At the equilibrium, all households in the city are inside the urban area, which can be given by

n(x )dx + n(x )dx = N. (30)
J J¯
The first and second terms on the left-hand side of Eq. (30) represent the numbers of households inside and outside all
the TOD zones, respectively.
From J ∪ J¯ = B and substituting Eq. (29) into Eq. (30), we have

B
u − Y + α + ϕ (x ) − β log (x )
σ exp − dx = N. (31)
0 α ( 1 − θ2 )
According to A1, the equilibrium rent per unit of land area devoted to housing at the fringe of the city equals the
exogenous agriculture rent or the opportunity cost of the land, i.e., r(B) = rA . Combining Eqs. (19) and (27) and r(B) = rA ,
one obtains

u − Y + α + ϕ (B )
α (1 − θ2 )σ exp − = rA . (32)
α ( 1 − θ2 )
Given the households’ annual income Y, the annual travel cost ϕ (x) and the amenity level (x), one can determine the
equilibrium household utility level u and the city boundary B by solving the system of Eqs. (31) and (32) and thus the
functions z(x), g(x) and p(x) in terms of Eqs. (17)–(19), and S(x), r(x) and n(x) in terms of Eqs. (26)–(29), respectively. The
closed-form solutions of the equilibrium household utility level u and the city boundary B are given as follows (its proof is
provided in Appendix B).
Proposition 2. Given the annual travel cost ϕ (x), household income Y and the opportunity cost of land rA , the common utility u
and city boundary B can be, respectively, given by

σ (α (1 − θ2 )Va 1 + 2ρτa 2 )
u = Y − α + α (1 − θ2 ) log (33)
2ρτa N + Va rA
and

Va α ( 1 − θ2 ) α (1 − θ2 )(2ρτa N + Va rA )
B = D1 + log − c ( D1 ) (34)
τa 2ρ rA (α (1 − θ2 )Va 1 + 2ρτa 2 )
where
M M
ϕ ( LM ) ϕ ( Ds ) ϕ ( Ls ) ϕ (Ls−1 )
1 = 1 − exp − + 2 exp − − exp − − exp − (35)
α ( 1 − θ2 ) s=1
α ( 1 − θ2 ) α ( 1 − θ2 ) s=2
α ( 1 − θ2 )
and

M D s + s
ϕ (x ) β log (x )
2 = Xs exp − exp − 1 dx. (36)
s=1 D s −s α ( 1 − θ2 ) α ( 1 − θ2 )
The following proposition further reveals the effects of total number of households, and location and size of TOD station
on the household utility level and city boundary. Its proof is given in Appendix C.
Proposition 3. The household utility level u increases with the size (or radius) of this TOD station j , but decreases with the
total number of households in the city N and the distance of the TOD station from the CBD Dj . The city boundary B increases with
N and Dj , but decreases with j . These effects are summarized in Table 2.
4. TOD design models under public and private investment regimes
In this section, two models, a public and a private regime, are presented for the TOD design along a rail transit line.
As previously stated, the layout and design of the TOD system are made by the government. However, the TOD investment
cost may be undertaken by the government or the private property developers in practice. From the perspective of the
JID: TRB
Table 2
Effects of total number of households, location and size of TOD station
on the household utility level u and city boundary B.
N Dj j
u − − +
B + + −
Note: “+” means a positive correlation and “−” means a negative corre-
lation.
planner (i.e., the government), the objective of the TOD design is to maximize the social welfare of the urban system by
determining the optimal location, number and size of the TOD zones along a rail line corridor and the train headway on
that rail line. The social welfare is the total benefits of all parties in the urban system, and is defined as the sum of the total
utility of all households in the urban system, the aggregate land rent received by the absentee landlords, and the total fare
revenue, minus the train operating cost and the TOD investment cost (only for the public investment regime in which the
government pays the TOD investment cost).
Let U represent the total utility of all households in the city. It can be given by
U = uN, (37)
where the equilibrium household utility u can be calculated by Eq. (33).
Let RENT represent the total land rents for constructing the rail line or housing and FARE represent the total fare
revenue generated by the rail operations, respectively. They can be, respectively, expressed as
B
RENT = (r (x ) − rA )dx, (38)
0
and

M
F ARE = 2ρ Fs Qs , (39)
s=1
where the rent or value r(x) per unit of land area at location x is determined by Eqs. (27) and (28) for the public and private
regimes, respectively. The rail fare, Fs , from station s to the CBD is given by Eq. (6). The passenger demand, Qs , for station s
is given by Eq. (11).
Let COPE represent the annual train operating cost, which comprises the fixed operating cost and variable operating cost
(Li et al., 2012b, c), expressed as

COPE = λ0 + λ1 , (40)
H
where λ0 is the annual fixed operating cost and λ1 is the annual operating cost per train. represents the vehicle round
journey time, and thus H represents the fleet size or number of vehicles on the rail line. The round journey time consists
of the terminal time, the line-haul travel time and the train dwell time at stations. It is given as
= ζ t0 + 2(t1 + t2 ), (41)
where t0 is the constant terminal time on the circular line and ζ is the number of terminal times on the line. t1 and t2 are,
respectively, the total line-haul travel time and total dwell time at stations (i.e., t1 = D1 /Vt and t2 = κ 0 M).
Let CTOD represent the total construction cost of all the TOD zones along the rail line. According to Eqs. (1) and (22), CTOD
can be expressed as

M D s + s
CT OD = π (x )dx = Xs (c0 + c1 r (x )+c2 (x ))dx, (42)
J s=1 D s −s
where the TOD construction cost, π (x), at location x is determined by Eq. (22). The total catchment area J of all the TOD
zones along the rail line is given by Eq. (1).
In light of the above, the social welfare maximization problem under the public and private TOD investment regimes can
be formulated as
U
+ RENT + F ARE − COPE − CTOD , for public regime,
max SW (X, , H ) =
U + RENT + F ARE − COPE , for private regime,
⎧ B M
⎪
⎪uN + 0 (r (x ) − rA )dx + 2ρ s=1 Fs Qs − λ0 + λ1
⎨ D s + s
H

= − M s=1 Xs Ds −s (c0 + c1 r (x )+c2 (x ) )dx, for public regime, (43)
⎪
⎪
⎩ B M
uN + 0 (r (x ) − rA )dx + 2ρ s=1 Fs Qs − λ0 + λ1 H , for private regime,
JID: TRB
s.t. Constraints (30)–(32),

M
KV EH
μQ s ≤ , (44)
H
s=1
Xs ≤ s ≤ Xs
¯, ∀s = 1, 2, . . . , M, (45)

1, if s is selected as a TOD station, s = 1, 2, . . . , M,
Xs = (46)
0, Otherwise,
where SW(·) denotes the total social welfare of the urban system. X = (Xs ,s = 1, 2,…, M) is the vector of the 0–1 indicator

variables. M s=1 Xs is thus the total number of the TOD stations along the rail line. = (s ,s = 1, 2,…, M) is the vector of the
sizes (or radii) of the TOD stations along the rail line. H is the train headway on the rail line. Constraints (30)–(32) represent
the housing market equilibrium. μ is the peak-hour factor, i.e., the ratio of peak-hour demand to the daily passenger de-
mand, which is used to convert the passenger volume from a daily basis to an hourly basis. μQs is the peak-hour passenger
demand at station s. KVEH is the capacity of vehicles (i.e., the maximum number of passengers allowed in a vehicle, both
seated and standing). Eq. (44) is the line capacity constraint, which guarantees that the rail service supply satisfies the asso-
ciated (peak-hour) passenger demand. and ¯ in Eq. (45) are the lower and upper bounds for the TOD size, respectively.
Eq. (45) reveals that s > 0 if and only if Xs = 1 (i.e., station s is a TOD station), and s = 0 otherwise.
The social welfare maximization model (43)–(46) is a mixed 0–1 integer programming problem, which is usually non-
linear and non-convex, making it difficult to find its global optimal solution. In the following, a heuristic greedy approach
M
s=1 Xs , optimal location X and optimal size of the TOD stations along the rail
is used to find the optimal number ∗ ∗ ∗
line, where the asterisk represents the optimal solution of associated variable. The basic idea behind the greedy approach is
to decompose the optimization problem with M candidate TOD stations into M sub-problems, and then sequentially solve
the M sub-problems with one sub-problem at one time. Note that the number of the TOD stations is an integer variable,
which makes it difficult to solve. Fortunately, the number of the TOD stations on a rail line is a finite number. Therefore,
a simple approach to find the optimal number of the TOD stations is to compare the resultant objective function values
with different numbers of the TOD stations. The radius of the TOD station is a continuous variable usually within a range
of 100 m to 10 0 0 m (see the definition of TOD in Calthorpe (1993)). For simplification, in this paper we treat it as a discrete
variable with a range from 100 m to 10 0 0 m, and the step size for searching its solution is set as 10 m. The step-by-step
procedure of the heuristic greedy approach is described as follows.
Step 1. First loop operation. Let j be the counter of the TOD stations, and j begins with 1, i.e., j = 1.
Step 2. Second loop operation. Determine the optimal location and size of the jth TOD station.
Step 2.0. Choose an initial location Xj(1) and an initial size j(1) for the jth TOD station. Calculate the amenity level j(1)
and set the outer loop iteration counter ξ = 1.
Step 2.1. Third loop operation. Choose an initial value for the train headway H(1) . Set the inner loop iteration counter to i = 1.
Step 2.2. Calculate the annual travel cost ϕ (i) , and then determine the common utility u(i) and the city boundary B(i) by
Eqs. (33)–(36). Calculate the values of the vectors p(i) , g(i) , S(i) , r(i) and n(i) by Eqs. (18)–(20) and (26)–(29), and
then the aggregate land value RENT(i) , the total fare revenue FARE(i) , the annual train operating cost COPE (i) and
the total TOD investment cost CTOD (i) by Eqs. (38)–(42), respectively. Calculate the resultant passenger demand,
Qs (i) , for railway station s by Eq. (11).
Step 2.3. Check whether the resultant passenger demand exceeds the rail service supply, i.e., whether the capacity con-
straint Eq. (44) is satisfied. When Eq. (44) is active (or binding), then solve the auxiliary train headway Hˆ (i ) =
(i )
KV EH / M s=1 μQs .
Step 2.4. Update the train headway according to H (i+1 ) = H (i ) + (Hˆ (i ) − H (i ) )/i.
Step 2.5. Convergence check for the third loop operation. If the relative gap H (i+1 ) − H (i ) /H (i ) is smaller than a pre-
specified tolerance, then stop. Otherwise, set i = i + 1 and go to Step 2.2.
Step 2.6. Convergence check for the second loop operation. Change the values of the location Xj(ξ ) and size j(ξ ) for the jth
TOD station, and repeat Steps 2.1–2.5. Check the resultant social welfare: if no better solution can be found, then
stop and output SW(j) = max (SW(ξ ) ). Otherwise, set ξ = ξ + 1, and go to Step 2.1.
Step 3. Convergence check for the first loop operation. Repeat Steps 2.0–2.6, and check the social welfare. If SW(j + 1) < SW(j) ,
then stop and output the optimal social welfare SW∗ = SW(j) . Otherwise, set j = j + 1, and go to Step 2.
It should be pointed out that during the iterations, once constraints (44) and/or (45) are violated, the headway H and/or
the TOD’s radius s are then set at the corresponding bounds. The proposed heuristic solution algorithm can arrive at a
solution by making a sequence of decisions, each of which looks the best at that moment (that is, each decision is locally
optimal). Thus, a global optimum solution cannot be guaranteed.
5. Numerical study
In this section, a numerical example is used to illustrate the applications of the proposed models and the contributions
of this paper. The example is intended to compare the optimal solutions under different TOD investment regimes and to
JID: TRB
Table 3
The input parameters for the numerical example.
Symbol Definition Baseline value
M Total number of stations on the rail line 10

D1 Length of the rail line (km) 20
τa Value of access time (RMB/h) 30
τw Value of waiting time (RMB/h) 40
τt Value of in-vehicle time (RMB/h) 20
Va Average walking speed of passengers (km/h) 5.0
Vt Average train cruise speed (km/h) 50
κ0 Average train dwell time at a rail station (h) 0.01
f0 Fixed component of railway fare (RMB) 3.0
f Variable component of railway fare (RMB/km) 0.3
ρ Average annual number of trips to the CBD per household 365
η Average number of daily trips per household 1.0
ω1, ω2 Sensitivity parameter in elastic demand function inside and outside TOD zones 0.001, 0.08
α, β Parameters in household utility function 15,0 0 0, 80,0 0 0
a1, a2 Parameters in amenity function 0.1, 0.001
Y Average annual income of households (RMB/year) 60,0 0 0
θ 1, θ 2 Parameters in housing production function 0.04, 0.7
k Interest rate 0.05
c0, c1, c2 Fixed and variable components of TOD investment cost (RMB/year, RMB/km/year) 160,0 0 0, 0.9, 10 0,0 0 0
N Total number of households in the city 60 0,0 0 0
rA Opportunity cost of land at the city boundary (RMB/km/year) 150,0 0 0
λ0, λ1 Fixed and variable components of annual train operating cost (million RMB/year, million RMB/vehicle/year) 50, 20
ζ Number of terminal times on the line 2.0
t0 Constant terminal time on the circular line (h) 0.1
μ Peak-hour factor, i.e., the ratio of peak-hour flow to daily average flow 0.1
KVEH Capacity of vehicles (passengers/vehicle) 1500
,
¯ Lower and upper bounds for the radius of TOD zones (km) 0.1, 1.0
Note: “RMB” stands for Chinese currency and US$1.0 approximates RMB6.88 on 1 January 2017.
reveal the effects of introducing the TOD projects on urban spatial structure in terms of household’s residential distribution,
housing space, housing rental price, capital investment density and passenger demand for rail service. The effects of the
TOD investment cost and the total number of households in the city on the optimal TOD design schemes are also explored,
together with the urban population thresholds for investing in a TOD project under the public and private regimes. The
proposed solution algorithm is coded in MATLAB and run on a ThinkPad T410 computer with an Intel(R) Core(TM) i5 CPU
(2.4 GHz) and 4GB of RAM. The pre-specified tolerance is set as 0.0 0 0 01. The numerical experiment takes about 40 min of
CPU time. In the following analysis, unless specifically stated otherwise, the values of the input parameters are the same as
those in Table 3.
5.1. Comparison of solutions under public and private TOD investment regimes
Fig. 4 shows the changes of the annual social welfare with the TOD investment under the public and private TOD in-
vestment regimes and the resultant optimal TOD design schemes along the rail line corridor. It can be seen in Fig. 4a that
as the number of the invested TOD stations in the corridor increases, the annual social welfares of the urban system under
these two TOD investment regimes first increase and then decrease. The optimal social optimum solutions, respectively, oc-
cur at 5 and 8 TOD stations for the public and private investment regimes, with the maximum social welfares of RMB63.68
billion and RMB64.73 billion per year. Herein, “RMB” stands for Chinese currency and US$1.0 approximates RMB6.88 on 1
January 2017. The no TOD case leads to the lowest social welfare of RMB63.50 billion per year. This implies that the public
and private TOD investment regimes can bring an increase in the social welfare by RMB0.18 billion and RMB1.23 billion per
year, respectively. It can also be seen in Fig. 4a that the social welfare curve with the private TOD investment regime is
always above that with the public TOD investment regime. This means that the private TOD investment regime is superior
to the public TOD investment regime in terms of the social welfare. It is, therefore, more appropriate for the authority to
confer the development rights of the TOD projects to the private property developers from the society’s perspective of social
welfare maximization.
Fig. 4b and c depict the optimal TOD design schemes along the rail line corridor under the public and private TOD in-
vestment regimes, respectively. It can be noted that the number and size of the TOD zones and their investment importance
rankings are different for the two regimes. Specifically, under the public regime (see Fig. 4b), 5 TOD zones are required to
construct along the rail line, and the first TOD zone to be constructed is at the second nearest railway station from the CBD
with a radius of 230 m, and the size of the TOD zones along the rail line first increases and then decreases. However, under
the private regime (see Fig. 4c), 8 TOD stations to be invested are in turn constructed from the urban central area to the
suburban area, and the size of the TOD zones always decreases outwards along the rail line. Such changes are a result of
JID: TRB
Fig. 4. (a) Changes of annual social welfare with number of TOD stations under public and private TOD investment regimes; (b) optimal location and size
(or radius) of TOD stations under public regime; and (c) optimal location and size (or radius) of TOD stations under private regime.
the trade-off between the accessibility to the CBD and the TOD investment cost which is related to the land value and the
amenity level.
5.2. Effects of TOD investment on urban system
Fig. 5 displays the equilibrium household residential density, housing space, housing rental price, capital investment den-
sity, land value and daily passenger demand at each railway station under the public and private TOD investment regimes
and the no TOD case. It can be seen in Fig. 5a that in contrast to the no TOD case, the TOD project investments under the
public and private regimes lead to a higher residential density inside the TOD zones but a lower residential density outside
the TOD zones. This is because the TOD zones can provide residents with a better accessibility to the railway station and
the TOD amenities, such as the opportunities of entertainment and shopping. As a result, some residents would like to move
into the inside of the TOD zones from the outside of the TOD zones. Consequently, the housing rental price and the capital
investment density inside the TOD zones increase, while those outside the TOD zones decrease, as shown in Fig. 5c and d,
respectively. Thus, the housing space inside the TOD zones decreases and that outside the TOD zones increases, as shown
in Fig. 5b. It can also be seen in Fig. 5e that compared to the no TOD case, the TOD investment under the public regime
causes an increase in the land value inside the TOD zones, but a decrease in the land value outside the TOD zones. However,
under the private TOD investment the land value across the city always decreases. This is because in the public regime, the
government invests in the amenities inside the TOD zones and the private property developers invest in the housing. This
leads to a competition for the land between them. As a result, the land value inside the TOD zones increases. However, in
the private regime, the private property developers invest in both the housing and the amenities, thus causing a lower land
value. These observations are consistent with the results in Proposition 1.
Fig. 5f depicts the changes of the passenger demand at each railway station under the scenarios with and without TOD
project investment. It can be noted that as the distance from the CBD increases, the rail passenger demand decreases. In
contrast with the no TOD case, the TOD investment can cause an increase in the passenger demand close to the CBD area
JID: TRB
Fig. 5. (a)–(f) represent household residential density, housing space per household, housing rental price, capital investment density, land value, and daily
passenger demand at railway station with and without TOD projects, respectively.
JID: TRB
Table 4
Equilibrium solutions of urban system with and without TOD investment.
No TOD Public investment Private investment

Solution project regime regime
Number of TOD stations – 5 8

Total catchment area of all the TOD zones (km) – 1.66 6.18
Total number of households inside the TOD zones – 277,634 450,282
Total number of households outside the TOD zones 60 0,0 0 0 322,366 149,718
Train headway (min) 3.27 2.50 1.90
Length of corridor (km) 25.68 25.23 24.86
Household utility level (RMB/year) 101,140 103,328 105,099
Average household residential density (households/km) 23,358 23,783 24,139
Inside TOD zones – 167,250 72,861
Outside TOD zones – 13,677 8015
Average housing space (m2 /household) 71.89 66.31 67.06
Inside TOD zones – 45.28 57.30
Outside TOD zones – 84.43 96.42
Average housing rental price (RMB/m2 /year) 208.66 226.21 223.68
Average capital investment intensity (million RMB/km) 4905.26 4994.40 5069.10
Inside TOD zones – 35,122.41 15,300.86
Average land value (million RMB/km) 105.11 107.02 34.94
Total daily passenger demand (million passengers/day) 0.18 0.36 0.47
Annual social welfare (billion RMB/year) 63.50 63.68 64.73
B
Note: Average household residential density = N/citysize, Average housing space = 0 h (S (x ))dx/N, Average housing rental
B B B B
price = 0 p(x )h (S (x ) )dx/ 0 h (S (x ) )dx, Average capital investment intensity = 0 S (x )dx/city size, Average land value = 0 r (x )dx/city size.
but a decrease in the passenger demand in the suburb. That is, there is a critical location such that the passenger demand
increases on its left-hand side but decreases on its right-hand side due to the TOD investment. Specifically, the critical
locations for the public and private investment regimes are 12 km and 18 km, respectively.
Table 4 further summaries the effects of the TOD investment on the urban system under different investment regimes.
Table 4 shows that compared to the no TOD case, the (public or private) TOD investment can cause an increase in the house-
hold utility level, average household residential density, average housing rental price, average capital investment intensity,
total passenger demand and the annual total social welfare of the urban system, but a decrease in the city size and average
housing space. Specifically, there are 277,634 and 450,282 households to move into the TOD zones under the public and pri-
vate TOD investment regimes, respectively. Consequently, the city size decreases by 0.45 km (from 25.68 km to 25.23 km) for
the public regime and by 0.82 km (from 25.68 km to 24.86 km) for the private regime, implying that the TOD investments
lead to a more compact city and can thus restrain the urban sprawl. The public and private TOD investments, respectively,
cause an increase in the rail passenger demand by 0.18 million (from 0.18 million to 0.36 million) and 0.29 million (from
0.18 million to 0.47 million). As a result, the train headway decreases (or equivalently, the train frequency increases) by 46 s
and 82 s under the public and private regimes, respectively. In addition, in contrast with the no TOD case, the household
utility level and the social welfare, respectively, increase by RMB2188 per year per household and RMB0.18 billion per year
for the public investment regime, and by RMB3959 per year per household and RMB1.23 billion per year for the private
investment regime. This implies that the TOD investment along the rail line benefits both households and the society.
Table 4 also shows that in comparison with the public regime, the total catchment area of the TOD zones under the
private regime increases by 4.52 km from 1.66 km to 6.18 km. This is because the private investment regime decreases the
land value inside the TOD zones (see e.g., Fig. 5e) and thus the TOD investment cost, which leads to a larger total catchment
area of the TOD zones. It can also be seen in Table 4 that in contrast to the public investment regime, the number of
households inside the TOD zones, average household utility level, total daily passenger demand and social welfare under
the private regime increase by 172,648 households, RMB1771 per year, 0.11 million passengers per day, and RMB1.05 billion
per year, respectively. However, the average land value decreases by RMB72.08 million per kilometer.
5.3. Effects of TOD investment cost on TOD design schemes under public and private regimes
To look at the effects of the TOD investment cost on the optimal TOD design scheme, we conduct numerical experiments
by scaling the basic value of c0 in the TOD investment cost function (22) 0.8 time down and 1.2 times up for the public and
private TOD investment regimes. Fig. 6 shows that for the public investment regime, the TOD investment begins with the
second nearest station from the CBD, and the size of the TOD zones along the rail line first increases and then decreases.
However, the TOD investment importance rankings under the private regime are in turn from the urban central area to the
JID: TRB
CBD 2nd 1st 3rd 4th 5th 6th
230m 260m 200m 160m 120m 100m
(a) Fixed cost of TOD investment = 0.8c0
CBD 2nd 1st 3rd 4th 5th
200m 230m 170m 130m 100m
(b) Fixed cost of TOD investment = c0
CBD 2nd 1st 3rd 4th 5th
170m 190m 140m 110m 100m

(c) Fixed cost of TOD investment = 1.2c0
Fig. 6. Optimal TOD design scheme versus TOD investment cost under public regime (c0 = 160 0 0 0).
Fig. 7. Optimal TOD design scheme versus TOD investment cost under private regime (c0 = 160 0 0 0).
suburban area, and the size of the TOD zones along the rail line always decreases, as indicated in Fig. 7. Figs. 6 and 7 also
show that as the TOD investment cost increases from 0.8c0 to 1.2c0 , the optimal number of the TOD zones, respectively,
decreases from 6 to 5 for the public regime, and from 9 to 8 for the private regime. These results mean that the TOD
investment cost significantly affects the TOD investment decisions under different regimes.
Table 5 further summarizes the effects of the TOD investment cost c0 on the urban system performance under the public
and private investment regimes. It can be seen in Table 5 that for any of the both regimes, the increase of the TOD invest-
ment cost from 0.8c0 to 1.2c0 leads to a decrease in the total catchment area of the TOD zones, and thus in the total number
of households inside the TOD zones, household utility level, total passenger demand and annual social welfare. As a result
of decreased number of households inside the TOD zones, the total number of households outside the TOD zones and thus
the train headway increase, leading to a growth in the city’s size.
JID: TRB
Table 5
Effects of TOD investment cost on urban system performance (c0 = 160 0 0 0).
Urban system performance Fixed cost of TOD investment (RMB/year)
0.8c0 c0 1.2c0
Public regime Number of TOD stations 6 5 5

Total catchment area of all the TOD zones (km) 2.14 1.66 1.42
Total number of households inside the TOD zones 307,760 277,634 254,139
Total number of households outside the TOD zones 292,240 322,366 345,861
Private regime Number of TOD stations 9 8 8
Fig. 8. Optimal TOD design scheme versus total number of households under public regime (N = 60 0 0 0 0).
5.4. Effects of number of households on TOD design schemes under public and private regimes
Figs. 8 and 9 show the optimal TOD design schemes with different numbers of households under the public and private
TOD investment regimes, respectively. It can be noted that the total number of households in the city significantly affects
the number and size of the TOD zones. As the total number of households increases, the total number of the TOD zones and
their associated sizes generally increase for each regime. The investment order of the TOD neighborhoods under the private
regime is successively from the urban central area to the suburban area. However, it is not so for the public regime. Again,
this is a result of the trade-off between the accessibility to the CBD and the TOD investment cost.
Table 6 further indicates the effects of the total number of households on the urban system under the public and private
investment regimes. It can be seen that for each investment regime, the increase in the total number of households leads to
an increase in the total catchment area of all the TODs, the numbers of households inside and outside the TOD zones, city
size, total daily passenger demand, and social welfare, but a decrease in the household utility level and the train headway.
Fig. 10 depicts the changes of the net profit of the TOD investment projects with the total number of households in
the city under the public and private TOD investment regimes. The net profit of the TOD investment projects is defined as
the difference of the social welfare with optimal TOD investment scheme minus the social welfare without TOD investment
projects. Fig. 10 shows that as the total number of households increases from 50,0 0 0 to 450,0 0 0, the net profit for each
of the two regimes increases from a negative value to a positive value. The break-even point (i.e., the point with zero net
∗
profit) is 330,0 0 0 households for the public regime (i.e., NPUB = 330 0 0 0) and 70,0 0 0 households for the private regime (i.e.,
JID: TRB
Fig. 9. Optimal TOD design scheme versus total number of households under private regime (N = 60 0 0 0 0).
Table 6
Effects of total number of households on urban system performance (N = 60 0 0 0 0).
Urban system performance Total number of households
0.8N N 1.2N
Public regime Number of TOD stations 4 5 6

Private regime Number of TOD stations 8 8 9
∗ = 70 0 0 0). This means that the population threshold for investing in a TOD project under the public regime is larger
NPRI
than that under the private regime. Thus, the public investment regime can cause a late investment in the TOD projects.
6. Conclusion and further studies
This paper presented an analytical modeling approach to address the TOD investment issues in a rail transportation
corridor. The TOD system, as a public facility, was assumed to be designed by the government, aiming to maximize the
total social welfare of the urban system. The amenities inside the TOD zones can be invested by the government or the
private property developers, leading to a public and a private investment regime, respectively. On the basis of an analysis of
the urban system equilibrium with the TOD investment, social welfare maximization models under the public and private
investment regimes were proposed to simultaneously determine the optimal location, number and size of the TOD zones
and the optimal train headway on the rail line. In the proposed models, the effects of the TOD investments on the urban
spatial structure in terms of households’ residential location choices and housing market were explicitly taken into account.
JID: TRB
Fig. 10. Changes of net profit of TOD investment projects with total number of households under public and private regimes.
Some important findings and new insights are obtained. First, the TOD investment can lead to a more compact city and
thus effectively control the urban sprawl through changing households’ residential location choices and housing market. It
can also lead to an increase in the total rail passenger demand (and thus an increase in the train frequency), household
utility level, and social welfare. The TOD investment can thus benefit both households and the society. Second, the private
investment regime outperforms the public regime in terms of total social welfare of the urban system. Thereby, in practice
it is more appropriate for the government to confer the development rights of the TOD projects to the private property
developers from the society’s perspective of social welfare maximization. In contrast to the private investment regime, the
public investment regime can cause a late investment in the TOD projects. Third, the TOD investment cost and the total
number of households in the city can significantly affect the TOD investment decisions in terms of the number and size of
TOD neighborhoods. The proposed models can serve as a useful tool for addressing TOD investment issues from an economic
viable and cost-effective perspective.
Although the proposed models in this paper provide new revenues for examining the effects of the TOD investment on
the urban system, some further extensions should be made.
1. The city concerned in this paper was assumed to be monocentric. However, modern cities generally have multiple
business and commercial centers. Therefore, there is a need to extend the proposed models to consider a polycentric
urban structure.
2. It is assumed in this paper that all households were homogeneous in the value of time and income level. However,
in reality, people with different income levels have different attitudes/preferences towards residential location choices
(Li and Peng, 2016). Therefore, it is worthwhile to investigate the effects of heterogeneous commuters with different
income levels on the TOD design scheme.
3. The proposed models in this paper assumed that the rail line was operated by the government, with the objective of
maximizing the total social welfare of urban system. Recently, the Chinese government has tried to encourage private
sectors to take part in the public infrastructure investment projects through offering various confessional terms, such
as Build-Operate-Transfer and Public-Private-Partnership. The objective of the private sectors is however to maximize
their own profit. It is, thus, worthwhile to extend the proposed models to consider the interests of the private and
public investors with different objectives.
4. The governmental regulations, such as regulations on urban population density, building height, city size, and floor
area ratio, have a direct effect on the housing supply and households’ housing floor space (Brueckner, 2007; Kono
et al., 2012) and thus on the TOD investment schemes. Therefore, it is meaningful to extend the proposed models to
examine the effects of the governmental regulations on the urban spatial structure and the TOD investment decisions.
5. This paper focused mainly on rail mode, and thus the competition and substitution effects between private transporta-
tion and transit modes was ignored. It is thus meaningful to extend the proposed models to consider different modes
of transportation and to address the effects of TOD investments on a multi-modal transportation system (Li et al.,
2012a; Ma and Lo, 2013).
6. In the proposed models, rail line configuration (e.g., rail line length, station location and spacing) was assumed to be
pre-given and fixed. However, in reality the TOD investment scheme affects the residential distribution along the rail
line and thus the passenger demand for the rail transit service, which in turn affects the layout of the rail transit
line configuration. Therefore, it is meaningful to investigate the integrated optimization issue of the rail transit line
configuration and the TOD development scheme.
JID: TRB
7. The proposed heuristic solution algorithm in this paper cannot guarantee to find a global optimum solution of the
models proposed in this paper. It is, thus, meaningful to develop an efficient solution algorithm for solving the models
in a further study.
Acknowledgments
We are grateful to Professor Hai Yang, Professor William H.K. Lam, four anonymous referees, and participants at the
8th International Workshop on Computational Transportation Science for their helpful comments and suggestions on earlier
versions of the paper. The work described in this paper was jointly supported by grants from the National Natural Science
Foundation of China (71525003, 71222107), the Major Program of National Social Science Foundation of China (13&ZD175),
Huazhong University of Science and Technology (50 0130 0 0 01), and the National Research Foundation of Korea (NRF) Grant
funded by the Korea government ( MSIP ) (NRF -2010-0029446).
Appendix A. Proof of Proposition 1
According to Eqs. (27) and (28), one can obtain

∂ rPUB θ2 / ( 1 − θ 2 )
= θ1 pθ1 θ2 k−1 , (A.1)
∂p
θ / ( 1 − θ2 )
∂ rPRI θ1 pθ1 θ2 k−1 2 /(1 + c1 ), ∀x ∈ J,
= (A.2)
∂p θ pθ θ k−1 2
θ / ( 1 − θ 2 )
, ∀x ∈ J¯,
1 1 2
∂ rPRI 1
=− , ∀x ∈ J, (A.3)
∂ c0 1 + c1
∂ rPRI (x )
=− , ∀x ∈ J. (A.4)
∂ c2 ( 1 + c1 )
∂ rPUB ∂ rPRI ∂ rPRI ∂ rPRI
Since θ 1 , θ 2 , k and c1 are positive parameters and (x) > 0 holds, we have ∂ p > 0, ∂ p > 0, ∂ c0 < 0, x ∈ J and ∂ c2 <
0, x ∈ J.
Appendix B. Proof of Proposition 2
According to Eqs. (29) and (30) and J ∪ J¯ = B, we obtain

u−Y +α
ϕ (x ) ϕ (x )−β log (x )

− − −
N= σe α ( 1 − θ2 ) e α (1−θ2 ) dx + e α ( 1 − θ2 ) dx
J¯ J
u−Y +α
B ϕ (x ) ϕ (x )−β log (x ) ϕ (x )

− − − −
= σ e α ( 1 − θ2 ) e α (1−θ2 ) dx + e α ( 1 − θ2 ) − e α ( 1 − θ2 ) dx . (B.1)
0 J
From Fig. 1, we have

ϕ (x ) ϕ (x )
ϕ (x ) ϕ (x )

B
−
LM
−
M Ds
−
Ls−1
−
e α (1−θ2 ) dx = e α (1−θ2 ) dx + e α (1−θ2 ) dx + e α (1−θ2 ) dx , (B.2)
0 0 s=1 Ls Ds
where Ls is given by Eq. (9), and L0 = B.

According to Eq. (7), one obtains
ϕ (x )
ϕ ( LM )

LM
− α (1 − θ2 )Va −
e α (1−θ2 ) dx = 1−e α ( 1 − θ 2 ) . (B.3)
0 2ρτa
For any s = 1, 2, …, M, we have

Ds
−
ϕ (x )
α (1 − θ2 )Va − αϕ(1(−Dsθ)2 ) −
ϕ ( Ls )
e α (1−θ2 ) dx = e − e α ( 1 − θ2 ) , (B.4)
Ls 2ρτa
and

Ls−1
−
ϕ (x )
α (1 − θ2 )Va − αϕ(1(−Dsθ)2 ) −
ϕ (Ls−1 )
e α (1−θ2 ) dx = e − e α ( 1 − θ2 ) . (B.5)
Ds 2ρτa
JID: TRB
Substituting Eqs. (B.3)–(B.5) into Eq. (B.2) yields

M
ϕ (x ) ϕ ( LM ) ϕ ( Ds ) ϕ ( Ls ) − ϕ (Ls−1 ) ϕ (B )
B
− α (1 − θ2 )Va −
M
− − −
e α (1−θ2 ) dx = 1 − e α ( 1 − θ2 ) + 2 e α ( 1 − θ2 ) − e α ( 1 − θ2 ) − e α ( 1 − θ2 ) − e α ( 1 − θ2 ) . (B.6)
0 2ρτa
s=1 s=2
Substituting Eqs. (32), (35), (36) and (B.6) into Eq. (B.1) yields Eq. (33).
From Eqs. (32) and (33), we obtain Eq. (34). This completes the proof of Proposition 2.
Appendix C. Proof of Proposition 3
The partial derivatives of u and B with regard to N, Dj and j can be obtained as follows:
∂u 2ρτa α (1 − θ2 )
=− , (C.1)
∂N 2ρτa N + Va rA

∂u 4 ρ 2 τa D j + j
ϕ (x ) β log (x ) ∂ c j (x )
=− exp −
exp −1 dx, (C.2)
∂Dj α (1 − θ2 )Va 1 + 2ρτa 2 D j − j α ( 1 − θ2 ) α ( 1 − θ2 ) ∂Dj

∂u 4ρτa α (1 − θ2 ) ϕ (D j + j ) β log (D j + j )
= exp − exp −1 , (C.3)
∂ j α (1 − θ2 )Va 1 + 2ρτa 2 α ( 1 − θ2 ) α ( 1 − θ2 )
∂B Va α (1 − θ2 )
= , (C.4)
∂ N 2ρτa N + Va rA
D j + j
∂B 2ρVa ϕ (x ) β log (x ) ∂ c j (x )
= exp − exp −1 dx, (C.5)
∂ D j α (1 − θ2 )Va 1 + 2ρτa 2 D j − j α ( 1 − θ2 ) α ( 1 − θ2 ) ∂Dj

∂B 2α (1 − θ2 )Va ϕ (D j + j ) β log (D j + j )
=− exp − exp −1 . (C.6)
∂j α (1 − θ2 )Va 1 + 2ρτa 2 α ( 1 − θ2 ) α ( 1 − θ2 )
∂ c (x )
Noted that 0 < θ 2 < 1 and ∂ jD > 0 hold. Therefore, we can easily obtain
j
∂u ∂u ∂u ∂B ∂B ∂B
< 0, < 0, > 0, > 0, > 0, and < 0. (C.7)
∂N ∂Dj ∂j ∂N ∂Dj ∂j
References
Alonso, W., 1964. Location and Land Use: Toward a General Theory of Land Rent. Harvard University Press, Cambridge, MA.
Anas, A., 1982. Residential Location Markets and Urban Transportation. Academic Press, New York.
Anas, A., Xu, R., 1999. Congestion, land use, and job dispersion: a general equilibrium model. J. Urb. Econ. 45 (3), 451–473.
Beckmann, M.J., 1969. On the distribution of urban rent and residential density. J. Econom. Theory 1 (1), 60–67.
Beckmann, M.J., 1974. Spatial equilibrium in the housing market. J. Urb. Econ. 1 (1), 99–107.
Bernick, M., Cervero, R., 1997. The Transit Village in the 21st Century. McGraw-Hill, New York.
Brueckner, J.K., 2007. Urban growth boundaries: an effective second-best remedy for unpriced traffic congestion? J. Hous. Econ. 16 (3), 263–273.
Calthorpe, P., 1993. The Next American Metropolis: Ecology, Community, and the American Dream. Princeton Architectural Press, New York.
Cervero, R., 1994. Transit-based housing in California: evidence on ridership impacts. Transp. Pol. 1 (3), 174–183.
Cervero, R., 2007. Transit-oriented development’s ridership bonus: a product of self-selection and public policies. Environ. Plan. A 39 (9), 2068–2085.
Cervero, R., Day, J., 2008. Suburbanization and transit-oriented development in China. Transp. Pol. 15 (5), 315–323.
Cervero, R., Kockelman, K., 1997. Travel demand and the 3Ds: density, diversity, and design. Transport. Res. Part D 2 (3), 199–219.
Cervero, R., Murphy, S., Ferrel, C., et al., 2004. Transit-Oriented Development in the United States: Experiences, Challenges, and Prospects, Transit Cooperative
Research Program (TCRP) Report 102. Transportation Research Board, National Academy Press, Washington.
Chang, J.S., Mackett, R.L., 2006. A bi-level model of the relationship between transport and residential location. Transport. Res. Part B 40 (2), 123–146.
Dröes, M.I., Rietveld, P., 2015. Rail-based public transport and urban spatial structure: The interplay between network design, congestion and urban form.
Transport. Res. Part B 81, 421–439.
Duncan, M., 2011. The impact of transit-oriented development on housing prices in San Diego, CA. Urb. Stud. 48 (1), 101–127.
Eliasson, J., Mattsson, L.G., 20 0 0. A model for integrated analysis of household location and travel choices. Transport. Res. Part A 34 (5), 375–394.
Fujita, M., 1989. Urban Economic Theory: Land Use and City Size. Cambridge university press.
Hess, D.B., Almeida, T.M., 2007. Impact of proximity to light rail rapid transit on station-area property values in Buffalo, New York. Urb. Stud. 44 (5-6),
1041–1068.
Kay, A.I., Noland, R.B., DiPetrillo, S., 2014. Residential property valuations near transit stations with transit-oriented development. J. Transp. Geogr. 39,
131–140.
Kono, T., Joshi, K.K., Kato, T., Yokoi, T., 2012. Optimal regulation on building size and city boundary: an effective second-best remedy for traffic congestion
externality. Region. Sci. Urb. Econ. 42 (4), 619–630.
Kovacs, K.F., Larson, D.M., 2007. The influence of recreation and amenity benefits of open space on residential development patterns. Land Econ. 83 (4),
475–496.
Lam, W.H.K., Morrall, J., 1982. Bus passenger walking distances and waiting times: a summer-winter comparison. Transport. Q. 36, 407–421.
Lam, W.H.K., Zhou, J., 20 0 0. Optimal fare structure for transit networks with elastic demand. Transp. Res. Rec. 1733, 8–14.
Lee, J., Choi, K., Leem, Y., 2016. Bicycle-based TOD as an alternative to overcome the criticisms of the conventional TOD. Int. J. Sustain. Transport. 10,
975–984.
Li, Z.C., Chen, Y.J., Wang, Y.D., Lam, W.H.K., Wong, S.C., 2013. Optimal density of radial major roads in a two-dimensional monocentric city with endogenous
residential distribution and housing prices. Region. Sci. Urb. Econ. 43 (6), 927–937.
JID: TRB
Li, Z.C., Guo, Q.W., 2017. Optimal time for implementing cordon toll pricing scheme in a monocentric city. Papers Region. Sci. 96, 163–190.
Li, Z.C., Guo, Q.W., Lam, W.H., Wong, S.C., 2015. Transit technology investment and selection under urban population volatility: a real option perspective.
Transport. Res. Part B 78, 318–340.
Li, Z.C., Lam, W.H.K., Wong, S.C., 2012a. Modeling intermodal equilibrium for bimodal transportation system design problems in a linear monocentric city.
Transport. Res. Part B 46 (1), 30–49.
Li, Z.C., Lam, W.H.K., Wong, S.C., Choi, K., 2012b. Modeling the effects of integrated rail and property development on the design of rail line services in a
linear monocentric city. Transport. Res. Part B 46 (6), 710–728.
Li, Z.C., Lam, W.H.K., Wong, S.C., Sumalee, A., 2012c. Design of a rail transit line for profit maximization in a linear transportation corridor. Transport. Res.
Part E 48 (1), 50–70.
Li, Z.C., Peng, Y.T., 2016. Modeling the effects of vehicle emission taxes on residential location choices of different-income households. Transport. Res. Part
D 48, 248–266.
Lin, J.J., Gau, C.C., 2006. A TOD planning model to review the regulation of allowable development densities around subway stations. Land Use Pol. 23 (3),
353–360.
Loo, B.P., Chen, C., Chan, E.T., 2010. Rail-based transit-oriented development: lessons from New York City and Hong Kong. Landsc. Urb. Plan. 97 (3), 202–212.
Lund, H., 2006. Reasons for living in a transit-oriented development, and associated transit use. J. Am. Plan. Assoc. 72 (3), 357–366.
Lund, H., Cervero, R., Willson, R.W., 2004. Travel Characteristics of Transit-Oriented Development in California. California Department of Transportation,
Sacramento.
Ma, X., Lo, H.K., 2013. On joint railway and housing development strategy. Transport. Res. Part B 57, 451–467.
Martínez, F.J., Henríquez, R., 2007. A random bidding and supply land use equilibrium model. Transport. Res. Part B 41 (6), 632–651.
McDonald, J.F., 2009. Calibration of a monocentric city model with mixed land use and congestion. Region. Sci. Urb. Econ. 39 (1), 90–96.
Mills, E.S., 1972. Urban Economics. Scott Foresman, Glenview, Illinois.
Muth, R.F., 1969. Cities and Housing. University of Chicago Press, Chicago.
Nelson, D., Niles, J., 1999. Market dynamics and nonwork travel patterns: obstacles to transit-oriented development? Transp. Res. Rec. 1669, 13–21.
Ng, K.F., Lo, H.K., 2015. Optimal housing supply in a bid-rent equilibrium framework. Transport. Res. Part B 74, 62–78.
Ortuzar, J.D., Willumsen, L.G., 2001. Modelling Transport, Third edition John Wiley, Sons, Chichester.
O’Sullivan, A., 20 0 0. Urban Economics. Irwin/McGraw-Hill, Boston.
Papa, E., Bertolini, L., 2015. Accessibility and transit-oriented development in European metropolitan areas. J. Transp. Geogr. 47, 70–83.
Quigley, J.M., 1984. The production of housing services and the derived demand for residential energy. Rand J. Econ. 15 (4), 555–567.
Solow, R.M., 1972. Congestion, density and the use of land in transportation. Swed. J. Econ. 74 (1), 161–173.
Solow, R.M., 1973. Congestion cost and the use of land for streets. Bell J. Econ. Manage. Sci. 4 (2), 602–618.
Song, Y., Zenou, Y., 2006. Property tax and urban sprawl: theory and implications for US cities. J. Urb. Econ. 60 (3), 519–534.
Sung, H., Oh, J.T., 2011. Transit-oriented development in a high-density city: identifying its association with transit ridership in Seoul, Korea. Cities 28 (1),
70–82.
Vuchic, V.R., 1969. Rapid transit interstation spacing for maximum number of passengers. Transport. Sci. 3 (3), 214–232.
Vuchic, V.R., Newell, G.F., 1968. Rapid transit interstation spacing for minimum travel time. Transport. Sci. 2 (4), 303–339.
Wang, D.Z., Lo, H.K., 2016. Financial sustainability of rail transit service: the effect of urban development pattern. Transp. Pol. 48, 23–33.
Wu, J., 2006. Environmental amenities, urban sprawl, and community characteristics. J. Environ. Econ. Manage. 52 (2), 527–547.
Wu, J., Plantinga, A.J., 2003. The influence of public open space on urban spatial structure. J. Environ. Econ. Manage. 46 (2), 288–309.

Peng 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Peng 2017

Uploaded by

Copyright:

Available Formats

JID: TRB

ARTICLE IN PRESS [m3Gsc;March 31, 2017;0:19]

Transportation Research Part B 0 0 0 (2017) 1–22

Contents lists available at ScienceDirect

Transportation Research Part B

Transit-oriented development in an urban rail transportation

2 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

Eliasson and Mattsson (20 0 0) × × × ×

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 3

Land value, passenger

Households Property developers

Fig. 2. Interactions among three types of agents in the urban economy.

4 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

Fig. 3. The shape of amenity function (x).

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 5

3. Components of the models

s=1 [Xs (Ds − s ), Xs (Ds + s )].

3.1. Passenger travel cost

c (x ) = τa As (x ) + τwWs + τt Ts + Fs , ∀x ∈ [0, B], s = 1, 2, . . . , M, (2)

6 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

3.2. Passenger demand for each station

where Ls can be given by Eq. (9) and L0 = B.

3.3. Households’ residential location choices

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 7

3.4. Property developers’ housing production behavior

3.4.1. Property developers’ proﬁt under public TOD investment regime

8 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

3.4.2. Property developers’ proﬁt under private TOD investment regime

3.4.3. Proﬁt maximization problem for property developers

max ψPUB (S ) or ψPRI (S ). (24)

The ﬁrst-order optimality condition of maximization problem (24) yields

p(x )θ1 θ2 Sθ2 −1 − k = 0. (25)

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 9

3.5. Housing demand-supply equilibrium

4. TOD design models under public and private investment regimes

10 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 11

s.t. Constraints (30)–(32),

12 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

Symbol Deﬁnition Baseline value

M Total number of stations on the rail line 10

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 13

5.2. Effects of TOD investment on urban system

14 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 15

No TOD Public investment Private investment

Number of TOD stations – 5 8

16 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

CBD 2nd 1st 3rd 4th 5th 6th

230m 260m 200m 160m 120m 100m

(a) Fixed cost of TOD investment = 0.8c0

CBD 2nd 1st 3rd 4th 5th

200m 230m 170m 130m 100m

(b) Fixed cost of TOD investment = c0

CBD 2nd 1st 3rd 4th 5th

170m 190m 140m 110m 100m

Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22 17

Urban system performance Fixed cost of TOD investment (RMB/year)

Public regime Number of TOD stations 6 5 5

18 Y.-T. Peng et al. / Transportation Research Part B 000 (2017) 1–22

Urban system performance Total number of households

Public regime Number of TOD stations 4 5 6

6. Conclusion and further studies