bài báo nghiên cứu trung quốc PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

International Journal of Sustainable Transportation

ISSN: 1556-8318 (Print) 1556-8334 (Online) Journal homepage: https://www.tandfonline.com/loi/ujst20

How app-based ride-hailing services influence


travel behavior: An empirical study from China

Bao-Jun Tang, Xiao-Yi Li, Biying Yu & Yi-Ming Wei

To cite this article: Bao-Jun Tang, Xiao-Yi Li, Biying Yu & Yi-Ming Wei (2019): How app-based
ride-hailing services influence travel behavior: An empirical study from China, International Journal
of Sustainable Transportation, DOI: 10.1080/15568318.2019.1584932

To link to this article: https://doi.org/10.1080/15568318.2019.1584932

Published online: 14 Mar 2019.

Submit your article to this journal

Article views: 37

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ujst20
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION
https://doi.org/10.1080/15568318.2019.1584932

How app-based ride-hailing services influence travel behavior: An empirical


study from China
Bao-Jun Tanga,b,c,d,e, Xiao-Yi Lia,b,c,d, Biying Yua,b,c,d,e, and Yi-Ming Weia,b,c,d,e
a
Center for Energy and Environmental Policy Research, Beijing Institute of Technology, Beijing, China; bSchool of Management and
Economics, Beijing Institute of Technology, Beijing, China; cBeijing Key Lab of Energy Economics and Environmental Management, Beijing,
China; dCollaborative Innovation Center of Electric Vehicles in Beijing, Beijing, China; eSustainable Development Research Institute for
Economy and Society of Beijing, Beijing, China

ABSTRACT ARTICLE HISTORY


App-based ride-hailing is a newly emergent mode of travel that is expected to influence passen- Received 13 September 2017
gers’ travel behavior significantly. In this study, we define app-based ride-hailing services to Revised 15 February 2019
include hailing of taxis through smartphone apps and sharing of private vehicles (car sharing is Accepted 17 February 2019
not included). This paper is one of the first quantitative studies to examine how app-based ride
KEYWORDS
hailing would impact passengers’ choice of travel mode and change car-purchasing behaviors. App-based ride-hailing
China is taken as the empirical context, and the data are from a large-scale app-based survey con- service; car-purchasing
ducted via the largest app-based ride-hailing platform in the world. Drawing on these rich survey behavior; China; discrete
data, the investigation covers both the short-term travel mode choices and long-term car-purchas- choice model; travel
ing behaviors of app-based ride-hailing users. The research findings are helpful in understanding mode choice
changes in urban residents’ travel behaviors under in context of app-based ride hailing, and pro-
vide valuable in-depth insights for both governments and app-based ride-hailing service providers
(i.e., enterprises) for the management and regulation of ride hailing.

1. Introduction private vehicles through app-based ride-hailing services.


App-based ride hailing provides on-demand ride services for
The emergence of ride-hailing services has been regarded as
travelers, and allows private car owners to share their cars;
the product of growing travel demand and technological
this has changed the way people hail and share car rides.
innovation. The development of technology has allowed
However, the reasons that ride-hailing app users choose
smartphone users to access ride-hailing apps easily, and at
app-based ride hailing instead of their original mode of
the same time, has created a market for transport services
travel, and the factors that cause passengers to embrace this
based on sharing private cars (Anderson, 2014). On-demand
ride-hailing service platforms such as Uber, Lyft and DIDI new travel mode, remain unclear. It is necessary to discover
take advantage of smartphones, mobile payments and other the factors that influence users to get improve our under-
technologies to provide more diverse and differentiated serv- standing of potential strategies to develop this novel travel
ices, and have become rapidly integrated into people’s lives. mode. Furthermore, from a long-term perspective, the emer-
According to Morgan Stanley (2016), taxis and cars operated gence of app-based ride hailing may change demand for pri-
by ride-hailing companies already accounted for 4% of glo- vate cars, which may influence the willingness to buy new
bal miles traveled by 2015, and this proportion is estimated cars or replace old ones. To fully investigate the reasons and
to reach 26% by 2030. Although app-based ride hailing has changes in residents’ travel behavior induced by app-based
only risen in popularity in the past few years, it has had a ride-hailing services, this study takes China as its empirical
significant impact on existing transport systems, overall car context because it is the largest app-based ride-hailing mar-
usage and urban residents. App-based ride-hailing services ket in the world. In December 2016, the number of online
satisfy the travel demand that existing public transport taxi-hailing services users reached 226 million, and those
modes cannot. Especially in megacities like New York, users of other online ride-hailing services numbered 168
Beijing and Shanghai, public transport during rush hour, is million (CNNIC, 2017). We conducted a joint research
extremely crowded. App-based ride-hailing services offer study with DIDI Chuxing, the largest on-demand ride-hail-
more choices for people. To some extent, they compensate ing services platform in China and one of the largest in the
for insufficient public transport capacity. world. Unlike popular US ride-hailing apps (e.g., Uber and
In this study, we define app-based ride hailing to include Lyft), Chinese ride-hailing apps provide not only private car
hailing of taxis through smartphone apps and sharing of ride-hailing services, but also online taxi-hailing services.

CONTACT Biying Yu yubiying_bj@bit.edu.cn Center for Energy and Environmental Policy Research, Beijing Institute of Technology, Beijing 100081, China
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/ujst.
ß 2019 Taylor & Francis Group, LLC
2 B.-J. TANG ET AL.

Table 1. Definition of app-based ride-hailing services.


App-based taxi
Service type hailing service Hitching service Express service ExpressPool service Premium service
Car ownership Taxi company Private Private Private Private or Car company
Price Same as traditional taxi Set by the driver; Similar to taxi Total price is the same as Higher than all
Significantly Cheaper express service but other services
than taxi shared by a group
of passengers
Destination Single destination Similar destinations Single destination Similar destinations Single destination
with driver’s among passengers
Time consumption Less than traditional taxi Take more time for driver Same as app-based Spend more time on sys- Same as app-based
to confirm taxi hailing tem matching taxi hailing
Form of employment Full-time Part-time Part-time/Full-time Part-time/Full-time Part-time/Full-time
Characteristic Allows taxi driver and pas- Picks up passengers while Private car owners drive Private car owners drive Offers standardized, high
senger to find each on their own trips with around to make money around to make quality and profes-
other on apps passengers who have money; Allows private sional services
similar destinations car driver to pick up a
or routes group of passengers
who have the similar
destinations or part
of route
Chinese 出租车 (Chuzuche) 顺风车(Shunfengche) 快车(Kuaiche) 快车拼车(Kuaichepinche) 专车(Zhuanche)

Currently, app-based ride-hailing services including app- users (SUMC, 2017). In addition to traditional modes of
based Taxi hailing, Hitching service, Express service, public transport – taxis and limousines – a variety of
ExpressPool service and Premium service, have been innovative travel modes have emerged in recent decades,
launched on the Chinese market. Although in general, trad- including car sharing (Millard-Ball, 2005; Steininger, Vogl,
itional taxis and app-based taxis are the same travel mode, & Zettl, 1996; Teal, 1987), bicycle sharing (DeMaio, 2009;
they lead to different travel patterns (e.g., passengers choose Shaheen, 2010), ride-sourcing (Rayle, Dai, Chan, Cervero, &
taxis depending on their information about their availability) Shaheen, 2016; Rayle, Shaheen, Chan, Dai, & Cervero, 2014;
and show distinct passenger characteristics. Therefore, to Yu et al, 2017a, 2017b), ride-sharing (Chan & Shaheen,
highlight the differences of app-based ride-hailing services 2011). App-based on-demand ride services are now the most
from traditional travel modes, we define app-based ride-hail- rapidly evolving area of shared mobility. A large body of
ing services as new travel modes. Table 1 lists the main research indicates that different kinds of shared mobility
characteristics of different app-based ride-hailing services in reduce travel costs and fuel consumption, mitigate pollution
China provided by DIDI Chuxing. A questionnaire was spe- emissions, improve efficiency and reduce pressure on park-
cifically designed and sent out to frequent ride-hailing users ing lots among other benefits (Fagnant & Kockelman, 2014).
via the DIDI app-based platform. From the survey data, On the other hand, some studies find that shared mobility
there were two questions to be answered: (1) What factors stimulates demand for travel, which cause worsens traffic
make people turn to app-base ride-hailing services instead congestion and causes other issues (Cervero & Yu-Hsin,
of their original mode of transportation (i.e., travel mode 2003; Gao, An, & Quan, 2016; Schaller, 2016).
choice), and (2) How will app-base ride-hailing services Car sharing is a mode of car rental that provides people
change car-purchasing behavior (i.e., vehicle ownership)? with access to cars for short periods of time. As one of the
This paper is organized as follows. The second section earliest examples of shared mobility, car-sharing has gained
reviews the related literature. The third section describes the much academic attention. The impact of car-sharing on
research methods, including the models and interpretation travel behavior and car ownership are two main issues
of selected variables. The fourth section introduces the sur- (Martin & Shaheen, 2011; Martin, Shaheen, & Lidicker,
vey design and data collection. Section five discusses the 2010). According to research by Cervero, Golub, and Nee
model results. The final section concludes with the main (2007), more than 60% of car-sharing users would change
findings related to travel behavior and the influence of app- their car-purchasing plans, if their mobility and travel range
based ride-hailing services. Suggestions for ride-hailing plat- were enhanced. Martin et al. (2010) and Klincevicius et al.
forms and authorities are also discussed. (2014) found that households without private cars are active
users and that the vehicle ownership rate of ride-sharing
users is lower than that of nonusers. Utilizing a San
2. Literature review
Francisco area subsample and considering transit frequency,
Ride-hailing apps in China usually provide two kinds of walk, bike trips and daily driving distance, Mishra, Clewlow,
services: online taxi hailing and private car hailing. App- Mokhtarian, and Widaman (2015) built on previous studies
based taxi hailing allows taxi drivers and passengers to find to evaluate further the short-term impact of car sharing on
each other on mobile apps. Private car hailing services can travel behavior. Moreover, they assessed the impact of car or
be regarded as a typical form of shared mobility, because ride sharing on vehicle ownership. The results show that
people either share rides or their private car with others car-sharing users tend to engage in more sustainable travel
(Clewlow & Mishra, 2017). Shared mobility is a term used patterns, have lower rates of car ownership and take more
to describe transportation services that are shared among foot and bicycle trips.
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 3

Ride sharing (or ridesplitting) means that private car choice and car purchases. Another limitation of current
drivers share their cars on their own trips with passengers studies is data availability; most are based on small-scale
who have similar destinations or routes. As an innovative questionnaire survey data or trip survey information instead
and popular means of transportation, ride sharing, in com- of large-scale data collected from a ride-hailing platform.
mon with car sharing, widens mobility choices, offering Moreover, they do not directly target frequent users of app-
diverse forms of travel mode. Participants can benefit from based ride hailing which may overestimate its impact. Given
shared travel costs, saving waiting and travel time by using these concerns, we draw on the extensive app-based data
high-occupancy vehicle lanes and reducing commuting stress collected from the DIDI platform, which include 9762 valid
(Chan & Shaheen, 2011). With the tremendous development questionnaire responses from ride-hailing app users in
of ride sharing, scholars have shed light on this area; how- China (more details regarding the data can be found in
ever, research remains limited. Most studies have focused on Section 4). This paper aims to reveal the characteristics of
legal and compliance issues. Some scholars believe that ride app-based ride-hailing service users to determine which
sharing represents an innovative approach to reducing ser- users are likely to use app-base ride-hailing services instead
ious urban transportation problems and environmental of using the original travel mode, and the factors influencing
issues. A limited number of studied concern ride sharing their future car purchase decisions. The ultimate goal is to
from the perspective of travel behavior (Chen, Zahiri, & improve understanding of app-based ride-hailing behavior
Zhang, 2017). Yu et al. (2017a) evaluate the direct and indir- and services.
ect environmental benefits of ride-sharing and indicate that
from a long-term perspective, shifts in travel mode triggered
3. Travel behavior models for app-based
by ride-sharing will contribute to substantial emission
ride-hailing service users
reductions. In Beijing, for example, research findings show
that ride-sharing services could encourage people to reduce The emergence of app-based ride-hailing services may have
their private car usage and could change their decisions on a great impact on peoples’ travel behavior. In the short-
new car purchases. In the present study, app-based Hitching term, passengers tend to choose app-based ride hailing
service is considering a ride-sharing service. instead of buses, taxis, private cars, the metro or others serv-
“Ride sourcing” refers to on-demand ride-matching serv- ices; that is to say, people’s travel mode choice behavior has
ices in which drivers do not share a destination with their been changed by the availability of app-based ride hailing.
passengers (Shaheen, Cohen, & Zohdy, 2016). It has roots in In the long-term viewpoint, app-based ride hailing is likely
ride sharing, but shares the characteristics of traditional taxi to influence people’s demand for private vehicles owing to
services (Rayle et al., 2016). To some extent, it has replaced the convenience of app-based service, which in turn will
demand for private cars and public transportation, especially change their intention to purchase cars (whether new or
taxis. Rayle et al. (2016) collected 380 surveys in San replacement vehicles). Although the impact of app-based
Francisco and focused on the competition between app- ride-hailing services on travel behavior has been realized,
based on-demand ride sourcing and traditional taxi services. the form and triggers for changes in travel mode and car
Their study found that these two services have overlapping purchasing may differ between individuals according to their
but different markets. Liu and Wang (2016) attempt to sociodemographic attributes, travel behavior and motiva-
show how travelers’ mode choices have changed in a short tions, residential environmental characteristics, and sur-
period, following the emergence of a shared mobility plat- rounding transportation infrastructure. Consequently, two
form, focusing on the influence on urban traffic congestion. discrete choice models—a multinomial logit (MNL) model
However, owing to a lack of real-world data, their mathem- and a binary logit model (Koppelman & Bhat, 2006)—are
atical example failed to reflect the current situation accur- applied here to determine and quantify the effects of these
ately. Few studies of shared mobility have used real world factors on individuals’ travel mode choices behavior and
data. Based on two questionnaire surveys distributed in car-purchasing behaviors, respectively. The goal of discrete
Beijing, Gao et al. (2016) estimate the changes in Beijing choice analysis is to estimate the parameters of the
residents’ travel mode choices. They find that ride-sourcing value function.
in Beijing was enormously popular, reaching 3.526 million
rides per day. In Beijing, 58% of the total number of ride-
3.1. Choice model for app-based ride-hailing users’
sourcing trips are by people who would otherwise use public
alternative travel mode
transportation or cyclists. Ridesourcing is one type of app-
based ride-hailing service. In the present study, app-based A MNL model is used to investigate what alternative travel
Express service, ExpressPool service and Premium service mode the user would have taken if there are no app-based
are more similar to the abovementioned ride-sourcing ser- ride-hailing services. Based on this, we can further analyze
vice, whereby private cars are used to provide a taxi service. what induces travelers to give up their original modes of
As can be seen, some existing studies investigate the transportation and switch to app-based ride-hailing service.
influence of app-based shared mobility on people’s travel The theoretical framework of the (MNL) model has each
behavior, but mainly target a specific type of shared mobility ride-hailing app user i faced with j alternative travel modes.
service. However, these may be less informative in regard to Each resident chooses to a use ride-hailing service instead of
how app-based ride hailing typically affects travel mode six other travel modes. Here we set the alternative travel
4 B.-J. TANG ET AL.

mode: private car as j ¼ 1; traditional taxi 1as j ¼ 2; bus Table 2. Distribution of selected samples by city.
as j ¼ 3; metro as j ¼ 4; walking or cycling as j ¼ 5 and City Valid surveys City scalea Geographical distribution
other modes as j ¼ 6: Then, the users’ overall utility of choos- Shenyang 561 Supercity Northeast
Beijing 2743 Megacity Northern
ing particular alternative travel mode can be specified as: Tianjin 1895 Megacity
Uij ¼ bj Xij þ eij (1) Shijiazhuang 1741 Big city
Langfang 518 Medium-sized city
Shanghai 411 Megacity Eastern
Xij is a vector of explanatory variables expected to influence Jinan 583 Big city
users to change their travel mode to app-based ride-hailing Wuhan 541 Supercity Central
services. The explanatory variables are classified into four Shenzhen 435 Megacity Southern
Chengdu 334 Supercity Southwest
categories: sociodemographic attributes, travel behavior and a
The standard of division for city scale is used in accordance with ‘Notice on
motivations, residential environmental characteristics varia- regulating urban scale classification’ issued by state council in 2014.
bles and city-specific variables which are related to urban
transportation infrastructure. First, travel mode choice may users’ overall utility of allowing the app-based ride-hailing
vary according to annual household income, educational services to influence their car willingness to purchase a car
background and household type. In addition, we assume can be specified as:
that choosing app-based ride hailing is closely related to res-
idents’ travel behaviors and motivations, such as travel fre- Vik ¼ kk Yik þ eik (3)
quency, time and travel purpose, and their reasons for using where Yik is a vector of explanatory variables such as
app-based ride hailing. In addition, owning to differences in income, travel time, private car ownership, travel frequency
the time required to walk to the nearest public transport sta- and other variables that are expected to affect users’ car
tion and the characteristics of their residential areas, people purchase choices. The vector kk represents the coefficients
may have different choices of travel mode. Different spatial for the vector of explanatory variables Yik. The probability
contexts (e.g., cities) have different travel conditions, so of a user changing car-purchasing behavior based on the
transportation infrastructure, length of bus and metro oper- availability of app-based ride-hailing service can be calcu-
ating routes, number of operation taxis and urban conges- lated as:
tion indexes are also taken into consideration. The vector bj
expðkk Yik Þ
represents the coefficients for the vector of explanatory vari- P¼ (4)
ables Xij for alternative j. In an MNL model, it is assumed 1 þ expðkk Yik Þ
that the error terms eij, are independent and follow Gumbel The parameters (kk) are estimated by maximizing the log
distribution. Under this assumption, the probability that a of the likelihood function. The explanatory variables are pre-
user would choose alternative travel mode j if there were no sented in Table 3.
app-based ride-hailing service, which can also be explained
as the probability that a user will switch to app-based ride
hailing from alternative travel mode j, is calculated by equa- 4. App-based survey and data
tion (2): In China, there are five main app-based ride-hailing services:
  App-based Taxi hailing service, Hitching service, Express
exp bj Xij
Pij ¼ P   (2) service, ExpressPool service and Premium service. The def-
j exp bj Xij inition of each service is given in Table 1.
The parameters (bj) are estimated by maximizing the log
likelihood function. The goal of model estimation is to find
4.1. App-based survey
a set of parameters b such that the probability of the
observed choices being realized is highest. To understand better the phenomenon of app-based ride
hailing and the potential changes of in travel behaviors and
motivations caused by such services, we designed a ques-
3.2. Choice model of app-based ride-hailing users’ tionnaire for frequent users (who use app-based ride-hailing
car-purchasing behavior services at least twice per week) and conducted a large-scale
The binary logit model is utilized in this study to represent app-based survey through the DIDI Chuxing platform from
the car-purchasing choices of ride-hailing app users. May to November 2016. Users received and responded to
Similarly, the theoretical framework of the binary logit survey from their DIDI Chuxing app on their mobile
model has each ride-hailing user i faced with two choices: phones. DIDI Chuxing completed a total of 1.43 billion
(1) is likely to or will change their willingness to buy a new rides through its platform in 2015 (Tam, 2016), making it
car/replace the old car (k ¼ 1) or (2) otherwise (k ¼ 0). the most popular app-based ride-hailing company in the
People receive a certain level of utility from a car purchase world. Therefore, the data collected from the DIDI Chuxing
and choose the alternative that maximizes that utility. The platform may to some extent represent the overall situation
of China’s app-based ride-hailing market. According to
1 CBNDATA (2016) statistics, online app-based ride-hailing
Traditional taxi: trip generated from cab stands, by street-hailing or by
telephone instead of through ride-hailing apps. Traditional taxi and app-based services cover more than 400 cities in China. The penetra-
taxi hailing are the same travel mode, but have different dispatch modes. tion rate of this service in first- and second-tier cities is
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 5

Table 3. Descriptive statistics of variables.


Variables Proportion/Mean (S.D.c)
Sociodemographic attributes
Annual household income (RMB) Less than 50 thousand 15.3
50–100 thousand 33.9
100–150 thousand 22.0
150–250 thousand 17.2
More than 250 thousand 11.6
Education background Received higher education 47.4
Have not received higher education 52.6
Household type Live alone 19.4
Live with spouse 28.8
Two-generation family 31.5
Three-or more-generation family 12.3
Other 8.0
Travel behavior and motivations
Selected app-based ride-hailing service for travel Premium service 3.2
Hitching service 5.8
Taxi hailing service 2.9
Express service 69.9
ExpressPool service 18.1
Preferred app-based ride-hailing service Premium service 3.4
Hitching service 5.2
  Taxi hailing service 1.1
  Express service 70.2
  ExpressPool service 20.0
Reason for using app-based ride-hailing Lack of public transportation 19.1
Time saving 47.3
Economical 21.8
Limitations on private car 4.9
Parking issue 3.7
Other 3.1
Travel frequency of app-based ride-hailing  More than once a day 25.6
3–4 times a week 37.4
  1–2 times a week 27.5
  1–2 times a month or lesser 9.5
Travel purpose Commute to work/school 27.3
Commute to home 13.3
Entertainment 33.0
Business 9.6
Other 16.8
Alternative travel mode Private car 17.3
Traditional taxi 35.1
  Bus 22.2
  Metro 15.2
  Walking/Cycling 6.4
  Other 3.8
Travel time of alternative travel mode Less than 10 min 15.3
11–20 min 29.2
21–30 min 23.9
31–40 min 14.5
41–50 min 6.6
51–60 min 5.4
More than 60 min 5.1
Private car ownership – 1.86 (0.75)
Car purchasing behavior Would not change 16.8
  May not change 31.6
  May change 45.0
  Would change 6.6
Residential environment characteristics
Walking time to nearest public transport station Less than 5 min 35.8
6–10 min 37.8
11–20 min 16.7
21–30 min 5.4
31–60 min 2.4
More than 60 min 1.8
Living area Central area/CBD 48.8
  Other 51.2
City-specific factors
Length of operating routes of buses/ per capita(1000km)a 12.79 (7.37)
Length of operating routes of metro/ per capita(100km)a 2.18 (2.23)
Number of operating taxi/per capitaa 3.31 (2.43)
Urban congestion indexb 1.55 (0.1)
Cities’ internet penetration rate in transportationd 54.25 (27.35)
a
Provided by National Report on Urban Passenger Transport Development, 2016.
b
Provided by the Transportation Report of Main Cities in China 2015, 2016.
c
S.D.: standard deviation.
d
Provided by Amap. Detailed index algorithm can be found at http://www.useit.com.cn/thread-13477-1-1.html.
6 B.-J. TANG ET AL.

4.2. Descriptive statistics


In this section, key findings from the survey, including user
demographics, travel behavior and residential environmental
characteristics, are discussed. Using descriptive statistical
analysis, the preferences, motives and attitudes toward app-
based ride-hailing services can be explored and studied.
Table 3 shows the descriptive statistics of the variables used
in this study.

4.2.1. Respondents’ sociodemographic attributes


As can be seen in Table 3, the educational backgrounds of
app-based ride-hailing services users did not show signifi-
cant differences. Unlike Rayle et al.’s (2016) finding that
Figure 1. Selected app-based ride-hailing service and service preferences of
respondents. 84% of ride-sourcing services users in San Francisco were
well educated, most Chinese users had no higher education,
significantly higher than that in small cities, and the app- implying that app-based ride-hailing services are widely used
based ride-hailing trips in first- and second-tier cities in China. The annual household income of more than 55%
account for almost 97% of the total. Therefore, based on the of the survey respondents was between 50 and 150 thousand
geographical distribution and scale of each city, 10 cities RMB (approximately equivalent to 7–21 thousand USD).
were selected to represent the major urban areas of China. People who live in two-generation households or with a
This survey included a total sample of 150,000 medium- and spouse are the main users, accounting for 31.5% and 28.8%,
high-frequency users. The sample size in each city was respectively. People who live alone (19.4%) are also a target
decided according to the popularity of app-based ride-hail- group for app-based ride-hailing services.
ing services. After removing responses with missing and
invalid answers, a total of 9762 valid survey responses from
4.2.2. Respondents’ travel behavior and motivations
passenger in the 10 main cities of China were used for this
analysis. Table 2 shows the distribution of valid survey The travel behavior and motivations for using app-based
responses disaggregated by city. City-specific information ride-hailing services are observed in the sample. Figure 1
contains the following four aspects, therein, the city specific shows the demand for selected existing app-based ride-hail-
information related to urban transportation infrastructure ing services and preferred services. It seems that respondents
was collected outside the questionnaire survey. The survey choose their preferred app-based ride-hailing services now,
information contains the following four aspects: given similar distributions of the existing choices and pre-
ferred services. More than two-thirds of the respondents
(1) Users’ sociodemographic attributes: annual household chose Express service, which is the mainstay of the app-
income, education background and household type; based ride-hailing services. Express service, ExpressPool and
(2) Travel behavior and motivations: app-based ride-hail- Premium service are more likely to be chosen by app-based
ing service selected for the latest trip, preferred app- ride-hailing users, while the hitching service and online taxi-
based ride-hailing services, reason for using app-based hailing services are less competitive. Long waiting time for
ride hailing, frequency of travel via app-based ride order matching may explain why the Hitching service is
hailing, travel purpose, alternative travel mode, the less favored.
travel time of alternative travel mode, private car own- When into respondents’ demographics are examined in
ership and car-purchasing behavior; relation to their preferred services, users with higher house-
(3) Residential environmental characteristics: living area hold income prefer using the Premium and App-based Taxi
and walking distance to nearest public trans- hailing services. The well-educated make up a higher pro-
port station; portion of Expresspool service and Hitching service users;
(4) City-specific information related to urban transporta- these are regarded as environmentally friendly because they
tion infrastructure: bus system, metro routes, number promote sharing of the same car for the same trip (Yu et al.,
of taxis, congestion condition and Internet penetration 2017a). This implies that people with higher education are
rate in transportation2 (CBNDATA, 2016). more likely to make conscious choices of more environmen-
tally friendly app-based ride-hailing services.
The questionnaire asks directly about the reasons for
2
using app-based ride-hailing services, the purposes of trips
The Internet penetration rate in transportation measures and describes the
development level of the urban Internet in the transportation industry, based and the alternative travel modes that are replaced by app-
on the penetration of GPS navigation , the number of roads with real-time based ride-hailing services. As expected, user place greatest
road condition systems, the penetration of app-based ride-hailing services, the value on time saving, similar to Rayle et al’s (2016) finding.
number of bus stations with online bus timetables and so on. A detailed
index algorithm can be found at http://www.useit.com.cn/thread-13477–1- Economics and lack of public transportation are also the
1.html. main reasons for using app-based ride-hailing services. In
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 7

Figure 2. Selected app-based ride-hailing services and travel purposes.

Figure 3. Selected app-based ride-hailing services and travel time.

terms of trip purpose, 40% were for commuting between unavailable, 35.1% indicated taxi, implying fierce competi-
work and home. A large percentage (33%) of trips were for tion between traditional taxis and app-based ride-hailing
entertainment, and only 9.6% of the trips were for business. services, while 22.2% said bus and 17.3% said private car. It
Varied by purpose of using app-based ride-hailing, the is worth mentioning that among the respondents who chose
selected services of respondents are different (Figure 2). Of other modes, 0.4% answered that if the app-based ride-hail-
those who used the typical travel-sharing mode (Hitching ing services had not been available, they would not have
service and ExpressPool service), more than 50% were com- made the trip. Although only a few people would have
muters. Premium service is more attractive to people travel- abondoned the trip, this still suggests that app-based ride
ing for business purposes. hailing increases travel demand.
Because travel time differ, the app-based ride-hailing Comparing the original travel mode and selected app-
services selected by passengers vary slightly. More than half based ride hailing, Premium service users are mainly pri-
of the respondents use the services for 11 to 30-minute trips. vate car users and taxi passengers and account for the
With different travel time, the preferred services of users are highest percentage of those who would use other app-based
somewhat different. As can be seen in Figure 3, for longer ride-hailing services (Figure 4). ExpressPool service attracts
trips with increased travel time, Hitching service is used, more bus passengers than other app-based ride-hail-
while fewer people choose Express service. This reflects that ing services.
the more affordable and comfortable travel that the It is surprising that more than two-thirds of the ride-hail-
Hitching service provides for long-distance travelers. At the ing app users owned private cars; 16.9% of respondents
same time, it also offers private car users on long drives owned more than one car. Reasons for using app-based
with a way to reduce their travel costs without changing ride-hailing services varied according to private car owner-
their travel mode. ship (Figure 5): 7.1% and 5.4% of the private car owners did
When respondents were asked which travel mode you not drive their cars because of legal limits on private cars
would have taken if app-based ride-hailing service were and difficulties in finding parking. Respondents were asked
8 B.-J. TANG ET AL.

if the traffic, car purchase limitation and parking issues and the time required to walk to nearest public transport
become increasingly more severe but the app-based ride- station. A large number of respondents (73.2%) lived no
hailing services would be permanently provided, would they more than 10 minutes’ walk from their nearest public trans-
change their minds on car purchasing plan including buying port stations; however, they still chose to take a car.
new cars or replacing the old cars. App-based ride hailing
appears to have made 51.6% of respondents reconsider
car purchases. 5. Model results and discussion
5.1. How app-based ride-hailing services impact
passengers to abandon their original travel mode
4.2.3. Respondents’ residential environmental
characteristics The MNL estimates of alternative travel mode choices are
The residential environmental characteristics of respondents presented in Table 4, and the marginal effects on probability
were also investigated in the survey, including living area changes as a result of a one-unit change in an independent

Figure 4. Alternative travel modes of app-based ride-hailing users by service type.

Figure 5. Reasons for using app-based ride-hailing services differed by car ownership.
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 9

Table 4. MNL estimates of the changes of passenger’s travel mode choices.


Variables Private car Taxi Bus Metro Walking/Cycling
  Coef. (Z-stat)
Sociodemographic attributes
Annual household income
Less than 50 thousand Fixed Fixed Fixed Fixed Fixed
100–150 thousand 0.101 0.0575 0.662 0.0307 0.395
(0.554) (0.214) (3.454) (0.278) (1.889)
150–250 thousand 0.283 0.132 0.895 0.0259 0.268
(1.332) (0.589) (4.225) (0.346) (1.253)
More than 250 thousand 0.444 0.176 1.062 0.204 0.617
(1.870) (0.647) (4.433) (0.969) (2.335)
Household type
Live alone 0.604 0.746 0.392 0.641 0.622
(2.680) (3.721) (1.891) (2.935) (2.531)
Live with spouse 1.212 0.697 0.497 0.631 0.645
(5.769) (3.673) (2.533) (3.032) (2.744)
Two generations family 1.189 0.843 0.649 0.567 0.625
(5.650) (4.458) (3.326) (2.723) (2.650)
Three or more generations family 1.088 0.608 0.427 0.231 0.519
(4.572) (2.785) (1.890) (0.951) (1.918)
Other Fixed Fixed Fixed Fixed Fixed
Travel behaviors and motivations
Reason of using
Lack of public transportation 0.522 0.834 1.157 0.682 0.691
(1.833) (3.172) (4.024) (2.294) (2.059)
Time saving 0.855 0.946 1.292 0.932 0.841
(3.242) (3.885) (4.800) (3.354) (2.660)
Economical 0.564 0.428 1.071 0.619 0.483
(2.062) (1.680) (3.841) (2.147) (1.476)
Limitation of private car 0.588 0.305 0.004 0.163 0.079
(1.809) (0.991) (0.01) (0473) (0.190)
Parking issue 1.342 0.391 0.238 0.875 0.798
(3.417) (1.012) (0.557) (2.083) (1.701)
Other Fixed Fixed Fixed Fixed Fixed
Travel time
Less than 10 min Fixed Fixed Fixed Fixed Fixed
11–20 min 0.328 0.362 1.414 1.178 1.145
(1.992) (2.321) (7.548) (5.626) (5.580)
21–30 min 0.197 0.029 1.995 1.876 1.290
(1.111) (0.181) (10.204) (8.770) (5.990)
31–40 min 0.168 0.403 2.222 2.126 0.745
(0.834) (2.088) (10.369) (9.212) (2.991)
41–50 min 0.075 0.359 2.441 2.705 0.994
(0.261) (1.28) (8.338) (8.926) (2.934)
51–60 min 0.892 0.769 2.909 3.163 0.043
(0.040) (2.328) (8.587) (9.085) (0.094)
More than 60 min 0.892 1.750 1.801 2.096 1.209
(3.403) (6.745) (7.106) (7.861) (2.663)
Travel purpose
Commute to work/school 0.188 0.091 0.618 0.303 0.831
(1.097) (0.561) (3.637) (1.714) (4.169)
Commute to home 0.084 0.144 0.873 0.562 0.548
(0.390) (0.715) (4.160) (2.580) (2.188)
Entertainment 0.351 0.211 0.447 0.403 0.313
(2.115) (1.333) (2.712) (2.349) (1.566))
Business 0.357 0.253 0.305 0.258 0.119
(1.582) (1.175) (1.301) (1.118) (0.410)
Other Fixed Fixed Fixed Fixed Fixed
City specific factors
Length of operating routes of buses/per capita (1000 km) 0.010 0.053 0.057 0.076 0.018
(0.743) (2.811) (1.977) 3.741 (1.307)
Length of operating routes of metro/per capita (100 km) 0.050 0.046 0.057 0.164 0.066
(0.939) (1.516) (3.85) (1.912) (0.105)
Urban congestion index 3.489 4.161 4.664 2.888 3.980
(2.232) (2.754) (3.021) (1.738) (2.339)
CONS 5.406 4.952 7.947 7.466 6.943
(2.230) (2.123) (3.322) (2.901) (2.61)

variable are presented in Table 5. The statistical test favors 5.1.1. Respondents’ sociodemographic attributes
the random-effects MNL model, which relaxes the As can be seen from the estimation results in Tables 4 and
Independence of Irrelevant Alternatives assumption (Galizzi, 5, the wealthier bus passengers show a lower probability of
2004; Hausman & McFadden, 1984). This model was used using app-based ride hailing. Regarding household types,
to analyze the determinants of changes in travel mode in people who live with spouses and those with more than
the presence of app-based ride-hailing services. two-generation families are more likely to choose app-based
10 B.-J. TANG ET AL.

Table 5. Marginal effect (computed from Table 4).


Variables Private car Taxi Bus Metro Walking/Cycling
Sociodemographic attributes
Annual household income
Less than 50 thousand 0.016 0.037 0.062 0.013 0.004
50–100 thousand 0.036 0.059 0.108 0.024 0.016
150–250 thousand 0.063 0.074 0.149 0.019 0.011
More than 250 thousand 0.098 0.093 0.167 0.000 0.030
Household type
Live alone 0.001 0.051 0.378 0.013 0.003
Live with spouse 0.078 0.003 0.035 0.004 0.003
Two generations family 0.061 0.021 0.017 0.021 0.008
Three generations family 0.080 0.008 0.020 0.037 0.002
Travel behavior and motivations
Reason of using
Lack of public Transportation 0.036 0.028 0.069 0.014 0.006
Time saving 0.012 0.008 0.061 0.004 0.007
Economical 0.001 0.042 0.086 0.003 0.007
Limitations on private car 0.065 0.011 0.031 0.008 0.019
Parking issue 0.153 0.093 0.057 0.028 0.008
Travel time
11–20 min 0.044 0.086 0.085 0.036 0.034
21–30 min 0.074 0.223 0.181 0.103 0.041
31–40 min 0.109 0.308 0.270 0.160 0.009
41–50 min 0.127 0.347 0.262 0.232 0.010
51–60 min 0.142 0.436 0.347 0.300 0.029
More than 60 min 0.129 0.433 0.313 0.278 0.031
Travel purpose
Commute to work/school 0.004 0.099 0.070 0.001 0.046
Commute to home 0.039 0.067 0.096 0.015 0.009
Entertainment 0.009 0.030 0.024 0.009 0.001
Business 0.034 0.032 0.646 0.019 0.012
City-specific factors
Length of operating routes of buses/per capita (1000 km) 0.001 0.014 0.005 0.009 0.000
Length of operating routes of metro/per capita (100 km) 0.010 0.009 0.008 0.005 0.001
Urban congestion index 0.096 0.249 0.181 0.121 0.010

ride hailing instead of private cars, while people who live as the short supply of public transport and disorganized bus
alone are more likely to have taken taxis in the past. This stations, make the current transport systems of most cities
may be because when there are more family members, pri- unable to guarantee the supply of public transport (MTO,
vate cars may not satisfy everyone’s travel needs. Users who 2016), which eventually forces bus passengers to choose
live with larger families are the target groups for raising app-based ride hailing. The results also show that users who
awareness of ride-hailing services. walk or cycle may tend to use app-based ride-hailing serv-
ices because of lack of public transportation or parking
issue. The reasonable interpretation is that more than 40%
5.1.2. Travel behavior and motivations of people who chose walking or cycling for their last alterna-
Of the reasons cite for using app-based ride hailing, parking tive transportation own private cars. Therefore, when they
issues are the major factor that leads private car drivers choose the mode they travel, they may shift to use ride-hail-
abandon their original travel mode. This reflects the com- ing due to the lack of public transportation or parking
mon shortage of parking lots that has been identified in difficulties.
most of Chinese cites (NDRC, 2015). Limitations on private Travel time has significant effects on all travelers using
cars are also a main reason for drivers to stop driving their various travel modes. App-based ride hailing has to some
own cars and opt to app-based ride hailing. Among the cit- extent replaced private cars and taxis for short trips, suggest-
ies where the survey was conducted, five had already ing that private car users with short travel distances and taxi
imposed limits on private vehicle use. Thus, app-based ride passengers are a potential market for the service. One plaus-
hailing provides an alternative means of transportation for ible explanation for taxi passengers is that with the increased
people who own cars but are not allowed to drive. The low travel time, the price advantage of app-based ride hailing is
cost of some ride-hailing services has become the most no longer significant. Taxis are more dependable in terms of
attractive factor for former bus passengers, followed by lack service quality, sanitation and safety than app-based ride
of public transportation and time-saving. It is worth men- hailing, so users prefer to use taxis on long-distance trips.
tioning that during the survey period, the ride-hailing plat- By contrast, original bus and metro passengers show a
form implemented different levels of subsidies for potential higher probability of changing to app-based ride hailing
users, which allowed bus passengers to spend just a little when their travel time increases. The implication here is
more money to take a faster and more convenient means of that the public transport serves a limited area and usually
transportation. The problems of public transportation, such requires transfers, which increases discomfort. For travel
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 11

Table 6. Estimates and marginal effect of app-based ride-hailing users’ car purchasing behavior.
Coef. coefficients
Variables (z-stat) Marginal effect
Sociodemographic attributes
Household type
Live alone Fixed
Live with spouse 0.109 (1.651) 0.027
Two generations family 0.067 (1.125) 0.019
Three or more generations family 0.148 (1.875) 0.037
Other 0.143 (1.629) 0.036
Travel behavior and motivations
Alternative travel mode
Private car Fixed
Taxi 0.136 (2.337) 0.034
Bus 0.182 (2.562) 0.045
Metro 0.195 (2.398) 0.049
Walking/Cycling 0.338 (3.574) 0.084
Other 0.235 (1.938) 0.059
Travel time
Less than 10min Fixed
11–20 min 0.115 (1.780) 0.029
21–30 min 0.104 (1.555) 0.026
31–40 min 0.212 (2.793) 0.053
41–50 min 0.092 (1.026) 0.023
51–60 min 0.221 (2.213) 0.055
More than 60 min 0.237 (2.386) 0.059
Residential environmental characteristics
Walking time to nearest public transport station
Less than 5 min Fixed
6–10 min 0.098 (2.041) 0.025
11–20 min 0.063 (1.017) 0.016
21–30 min 0.190 (2.056) 0.047
31–60 min 0.293 (2.181) 0.073
More than 60 min 0.178 (1.178) 0.044
City-specific factors
Length of operating routes of metro (100 km) 0.022 (4.022) 0.027
Number of operating taxi (thousand) 0.011 (3.245) 0.027
Internet penetration rate in transportation 0.005 (2.863) 0.012

purposes, bus passengers have a high likelihood of changing original taxi passengers turning to app-based ride hailing.
to app-based ride hailing for urban commutes. This may be By using a ride-hailing platform, passengers can hail a cab
the result of shortages, overcapacity and long waiting time more easily and make full use of their waiting time.
for public transport during rush hour. However, Rayle et al. Therefore, original metro passengers are more likely to stick
(2016) observed that because some individuals may have a to taking underground transportation to avoid the incon-
drink after work, they must forgo driving and choose alter- venience of road congestion.
native travel modes.

5.1.3. City-specific factors 5.2. How app-based ride-hailing services influence users’
Surprisingly, an increase in the length of operating routes of car purchasing behavior
buses has a positive effect by encouraging original bus pas- The survey data were used to build a binary logit model.
sengers to adopt app-based ride hailing. A reasonable The results were used to analyze determinants of willingness
explanation is that the length of bus operating routes is to purchase cars given the availability of app-based ride-hail-
commonly determined by demand and city scale. Even ing services. Coefficients and marginal effects are shown in
though cities have longer bus operating routes, they still
Table 6. From the coefficient estimates, we gain an under-
cannot meet the demand by residents in cities such as in
standing of the factors that influence car purchase willing-
Beijing. Notably, residents who live in cities with longer
ness positively or negatively. Moreover, the marginal effect
metro operating routes are less likely to use app-based ride
suggests how independent variables influence car pur-
hailing instead than the metro. To be more precise, a 1%
increase in the length of metro operating routes decreases chase decisions.
the probability of choosing app-based ride hailing instead of
the metro by 0.5%. Metro has the advantage of a time guar-
antee and lower cost, especially compared with long-distance 5.2.1. Respondents’ sociodemographic attributes
car travel. If metro coverage were higher, it would reduce The results shown in Tables 6 suggest that if the app-based
the need for app-based ride hailing. The urban congestion ride hailing services were permanently available, car pur-
index represents the traffic congestion in a city. The higher chases would not be influenced by users’ sociodemographic
the congestion index, the worse the urban traffic. The attributes, but would be significantly influenced by their
increase in the urban congestion increases the probability of travel behavior.
12 B.-J. TANG ET AL.

5.2.2. Travel behavior and motivations hailing users, whose travel behavior has been explored in this
As the app-based ride-hailing travel time increases, the like- paper based on large-scale app-based survey data. Research has
lihood of users changing their minds about buying a new brought clarity to the current landscape of app-based ride hail-
car also increases. In other words, if longer journeys must ing. Several conclusions can be drawn as follows.
be taken using app-based ride hailing, the demand for pri- App-based ride-hailing services attract people who pursue
vate cars decreases. As expected, users’ original travel modes fast, affordable and around 10 to 30-minute point-to-point trans-
that would be replaced by app-based ride hailing are an port. Express service and ExpressPool service are the most popu-
important determinant of car purchases. In comparison with lar app-based ride-hailing services in China, and demand for
those who were originally private car users, app-based ride- these services is expected to increase. Hitching service is most
hailing users who would have used other modes are more favored by long-distance travelers. App-based ride hailing also
likely to change their minds over buying a new car. helps mitigate the inconvenience of limits on private car usage.
Residents who originally traveled mode is on foot or by In the short-term, the emergence of app-based ride hail-
bicycle have the highest probability (8%) of changing their ing has changed travel behaviors. More than 35% of app-
minds. Original bus and metro passengers also tend to based ride-hailing users would have taken traditional taxi
change their minds. That indicates that if the travel needs of services; this reflects the competitive relationship between
people who take public transport and cycle or walk can be app-based ride hailing and traditional taxi services. In add-
met by app-based ride hailing, then their demand for private ition, 37% of app-based ride hailing users would have taken
cars may weaken. Promoting app-based ride hailing to these public transportation (bus and metro); this shows that it is
residents will effectively reduce their willingness to buy cars. necessary to improve current public transport systems and
reach a balance between app-based ride hailing and public
transportation. In this study, various factors that influence
5.2.3. Residential environmental characteristics passengers’ travel mode choice in relation to app-based ride
Walking time to nearest public transport station has notice- hailing are explicitly stated. Factors that influence passengers
able impact on willingness to purchase cars. If residents to choose app-based ride hailing instead of their original
need to take a 21 to 60-minute walk to reach the nearest travel modes are related to their household income and
public transport and app-based ride hailing is available, they household type. Analyzing the ‘reason for using app-based
are more likely to reconsider their car-purchasing decisions. ride-hailing’, we find that private car users are more likely
to cite issues of parking and car usage limitations, whereas
5.2.4. City-specific factors bus passengers use it mainly because it is economical and
App-based ride-hailing users’ car-purchasing willingness also time saving. In terms of travel time, original private car
differs according to quality of urban transportation infrastruc- drivers and traditional taxi passengers are more likely to
ture. Length of operating routes of metro per capita has a choose app-based ride-hailing services for short trips,
negative and significant effect; a 1% increase in metro operat- whereas users who used to take buses and metro transport
ing routes per capita decreases the probability of changing are more likely to choose app-based ride hailing for long
willingness to purchase a car by 2.7%. This suggests that if trips. The findings in this study show that people who were
app-based ride-hailing services were popularized in cities with originally private car drivers with higher incomes, who live
shorter metro lines, people in these cities may reconsider car with more family members, or who travel for short trips,
purchases. As for taxis, residents who live in cities with more travel less frequently or live in cities with longer bus and
taxis are more likely to forgo future car purchases App-based metro operating routes are target groups more likely to give
ride hailing allows conventional taxi drivers to use an online up driving and turn to use app-based ride hailing.
hailing service to reach customers instead of aimlessly cruis- Taking a long-term perspective, app-based ride hailing is
ing, which improves operational efficiency would be likely to change people’s future willingness to purchases
improved. The Internet penetration rate in cities’ transporta- cars. More than half of app-based ride-hailing users showed
tion sectors is positively correlated with changes in car-pur- a change in attitude toward car purchases, and 6.6% of the
chasing decisions. Enhancing traditional ride hailing through respondents affirmed that they would not purchase new pri-
mobile apps could make urban mobility systems more con- vate cars if app-based ride-hailing services were permanently
venient and efficient. They emphasize car availability instead available. This shift in willingness is mainly correlated with
of car possession, reducing people’s willingness to buy cars. the time required to reach the nearest public transport,
travel time and their original travel modes. Compared with
those who were originally private car travelers, former users
6. Conclusions and policy implications of other travel modes are more likely to change their will-
ingness to purchase cars. Therefore, our study finds that
6.1. Conclusions
respondents who used to take buses, the metro, taxis,
App-based ride hailing as a newly emergent travel mode is bicycles or foot as their travel modes, or who live in the cit-
expected to continue its rapid evolution. However, the impact of ies with shorter metro routes, more operating taxis and
this new travel mode on passengers’ short- and long-term travel greater Internet penetration rates in transportation, are
behavior remains unknown. This study seeks to extend existing more likely to forgo purchasing cars. The development of
studies by understanding the travel behavior of app-based ride- app-based ride hailing makes it easier for people to travel
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 13

and reduces their need for private car ownership. With the urban public transport system is still an essential issue that
growing availability of app-based ride hailing, the use of government authorities should resolve.
existing transport resources can be maximized and demand
for private cars may be further reduced.
Acknowledgements
The authors acknowledge financial support received through National
6.2. Policy implications Key R&D Program of China (2016YFA0602603), the National Natural
Science Foundation of China (Grant Nos. 71822401, 71603020,
This study takes China as its empirical context; however, the 71573013, 71521002, and 71642004), the Beijing Natural Science
development of app-based ride hailing as a future mode of Foundation of China (Grant No. 9152014), Special Items Fund for
travel in China still requires strong and coordinated efforts Cultivation and Development of Beijing Creative Base (Grant No.
between app-based ride-hailing enterprises, government Z171100002217023), Project of Beijing Social Science Foundation
Research Base (Grant Nos.15DJA084 and 18JDGLB039), and the Joint
authorities and passengers. The conclusions and implications
Development Program of Beijing Municipal Commission of Education.
presented here may be applied in other developing countries
as well.
Implications for app-based ride-hailing service providers
(enterprises): the findings of travel behavior of app-based Disclosure statement
ride-hailing users provide these enterprises with a deeper No potential conflict of interest was reported by the authors.
understanding of user demand. Express services are the most
popular among users. Service providers should maintain its
advantages of time saving and economy. Hitching services as
a shared travel mode need to be further promoted to passen- References
gers. If the matching process and waiting times can be short- Anderson, D. N. (2014). “Not just a taxi”? For-profit ridesharing,
ened, more users will try this mode of travel. In addition, driver strategies, and VMT. Transportation, 41(5), 1099–1117. doi:
short-distance private car and taxi travelers, as well as those 10.1007/s11116-014-9531-8
Chan, N., & Shaheen, S. (2012). Ridesharing in North America: Past,
who live in areas without sophisticated metro systems are
present, and future. Transport Reviews, 32(1), 93–112.
more likely to choose app-based ride-hailing services. Chen, X. M., Zahiri, M., & Zhang, S. (2017). Understanding ridesplit-
Therefore, app-based ride-hailing platforms could incentivize ting behavior of on-demand ride services: An ensemble learning
drivers to operate in these areas, and marketing campaigns approach. Transportation Research Part C: Emerging Technologies,
could be designed for short-distance travelers. Future 76, 51–70. doi:10.1016/j.trc.2016.12.018
Cervero, R., & Yu-Hsin, T. (2003). San Francisco City CarShare:
research on app-based ride-hailing service demand, users’
Travel-demand trends and second-year impacts (No. 2003, 05).
travel behavior, impact analysis and relevant policy-making Working paper. University of California, Institute of Urban and
should fully consider price and incentive factors. App-based Regional Development.
ride-hailing service providers are encouraged to share more Cervero, R., Golub, A., & Nee, B. (2007). City CarShare: Longer-term
of their data with regulators. This in turn would help author- travel demand and car ownership impacts. Transportation Research
Record: Journal of the Transportation Research Board, 1992(1),
ities understand the problems of urban transport systems by
70–80. doi:10.3141/1992-09
analyzing app-based ride-hailing user behavior. CBNDATA. (2016). 2016 intelligent travel data report. Retrieved from
Implications for government authorities: authorities can http://www.199it.com/archives/556606.html
perceive the problems of urban transport systems exposed Clewlow, R. R., & Mishra, G. S. (2017). Disruptive transportation: The
by analyzing app-based ride-hailing user behavior. The find- adoption, utilization, and impacts of ride-hailing in the United
States (Vol. 7). Research Report–UCD-ITS-RR-17.
ings indicate that people who live in cities with longer metro China Internet Network Information Center (CNNIC). (2017). The
operating routes are more likely to reconsider purchasing 39th China Internet network development statistic report (in
new cars. This suggests that cities with relatively advanced Chinese). Retrieved from https://cnnic.com.cn/IDR/
public transport systems and app-based ride-hailing services ReportDownloads/201706/P020170608523740585924.pdf
may reduce their need for private cars. Therefore, authorities DeMaio, P. (2009). Bike-sharing: History, impacts, models of provision,
and future. Journal of Public Transportation, 12(4), 3.
should take steps to improve public transport systems, espe- Fagnant, D. J., & Kockelman, K. M. (2014). The travel and environ-
cially metro construction, to encourage residents to give up mental implications of shared autonomous vehicles, using agent-
their cars. In those cities where the transportation infra- based model scenarios. Transportation Research Part C Emerging
structure cannot be upgraded in the short-term, local Technologies, 40(1), 1–13. doi:10.1016/j.trc.2013.12.001
authorities could seek temporary partnerships with app- Gao, Y., An, J., & Quan, Y. (2016). The impact of online ride sourcing
on the choice of transportation and transport operation. Urban
based ride-hailing enterprises to promote this new travel Transportation, 5, 1–8 (in Chinese)
mode. App-based ride-hailing services should be required to Hausman, J. A., & McFadden, D. (1984). A specification test for the
share more of their data with regulators. This in turn allows multinomial logit model. Econometrica, 52(5), 1219–1240. doi:
authorities to have a better understanding of the problems 10.2307/1910997
Klincevicius, M., Morency, C., & Trepanier, M. (2014). Assessing the
of urban transport system through analyzing app-based ride-
impact of carsharing on household car ownership in Montreal.
hailing users’ behavior. However, in the long-term, govern- Emerging and Innovative Public Transport and Technologies.
ment authorities should not rely on app-based ride-hailing CD-ROM. Transportation Research Board of the National
services to ease urban traffic problems. Optimizing the Academies, Washington, DC, USA.
14 B.-J. TANG ET AL.

Koppelman, F. S., & Bhat, C. (2006). A self-instructing course in mode Appendix


choice modeling: Multinomial and nested logit models. Retrieved
from http://www.caee.utexas.edu/prof/bhat/courses/lm_draft_ A1. Questionnaire on the behavior of app-based
060131final-060630.pdf ride-hailing users
Liu, M., & Wang, X. (2016). Modeling the impacts of ride-sourcing
platforms on urban transportation market. Logistics Sci-Tech, 10, 1. Select your city
79–82. (in Chinese). 2. Select the app-based ride-hailing service you took.
Galizzi, M. M. (2004). The economics of car-pooling: A survey for
Europe. Workshop paper: Highways—Cost and Regulation in Europe. (1) Premium service
Bergamo, Italy: Universita degli Studi di Bergamo. (2) Hitching service
Martin, E., Shaheen, S., & Lidicker, J. (2010). Impact of carsharing on (3) App-based taxi hailing service
household vehicle holdings: Results of a North American shared-use (4) Express service
vehicle survey. Transportation Research Record: Journal of the (5) ExpressPool service
Transportation Research Board, 2143(1), 150–158. doi:10.3141/2143-
19 3. What service would you prefer to use in the future?
Martin, E. W., & Shaheen, S. A. (2011). Greenhouse gas emission
impacts of carsharing in North America. IEEE Transactions on
(1) Premium service
Intelligent Transportation Systems, 12(4), 1074–1086. doi:10.1109/
(2) Hitching service
TITS.2011.2158539
(3) App-based taxi hailing service
Ministry of Transport the People’s Republic of China (MTO). (2016).
(4) Express service
National report on urban passenger transport development.
(5) ExpressPool service
Morgan Stanley. (2016). Shared mobility on the road of the future.
Retrieved from https://www.forbes.com/sites/morganstanley/2016/07/
20/shared-mobility-on-the-road-of-the-future/#284ac64f1cae 4. Why do you use app-based ride-hailing services?
Millard-Ball, A. (2005). Car-sharing: Where and how it succeeds.
Transportation Research Board, 108. (1) Lack of public transportation
Mishra, G. S., Clewlow, R. R., Mokhtarian, P. L., & Widaman, K. F. (2) To save time
(2015). The effect of carsharing on vehicle holdings and travel (3) Economical
behavior: A propensity score and causal mediation analysis of the (4) Limitations on private car use
San Francisco Bay Area. Research in Transportation Economics, 52, (5) Parking difficulties
46–55. doi:10.1016/j.retrec.2015.10.010 (6) Other
National Development and Reform Commission (NDRC). (2015).
China still needs more than ten million parking lots. Retrieved from
http://www.sohu.com/a/82082757_364316 5. How frequently do you use app-base ride-hailing services?
Rayle, L., Shaheen, S., Chan, N., Dai, D., & Cervero, R. (2014). App-
based, on-demand ride services: Comparing taxi and ridesourcing (1) More than once a day
trips and user characteristics in San Francisco. University of (2) (2) 3–4 times a week
California Transportation Center (UCTC). UCTC-FR-2014-08. (3) (3) 1–2 times a week
Rayle, L., Dai, D., Chan, N., Cervero, R., & Shaheen, S. (2016). Just a (4) (4) 1–2 times a month or less
better taxi? A survey-based comparison of taxis, transit, and ride-
sourcing services in San Francisco. Transport Policy, 45, 168–178.
doi:10.1016/j.tranpol.2015.10.004 6. What is the purpose of your trip?
Schaller, B. (2016). Unfinished business: A blueprint for Uber, Lyft and
Taxi regulation. New York: Schaller Consulting. (1) Commute to work/school
Shaheen, S. A., Guzman, S., & Zhang, H. (2010). Bikesharing in (2) Commute to home
Europe, the Americas, and Asia: past, present, and future. (3) Entertainment
Transportation Research Record, 2143, 159–167. (4) Business
Shaheen, S., Cohen, A., & Zohdy, I. (2016). Shared mobility: Current (5) Other
practices and guiding principles (No. FHWA-HOP-16-022).
Shared-Use Mobility Center (SUMC). (2017). What is shared-use
mobility? Retrieved from http://sharedusemobilitycenter.org/what-is- 7. If you were not using an app-based ride-hailing service, which travel
shared-mobility/ mode would you have taken?
Steininger, K., Vogl, C., & Zettl, R. (1996). Car-sharing organizations:
The size of the market segment and revealed change in mobility (1) Private car
behavior. Transport Policy, 3(4), 177–185. doi:10.1016/S0967- (2) Traditional taxi
070X(96)00024-8 (3) Bus
Tam, I. (2016, January 12). Didi Kuaidi completed 1.4bn rides in 2015, (4) Metro
surpassing Uber. Marketing Interactive. Retrieved from http://www. (5) Walking/Cycling (6) Other
marketing-interactive.com/didi-kuaidi-completed-1-4bn-rides-in-
2015-surpassing-uber/.
Teal, R. F. (1987). Carpooling: Who, how and why. Transportation 8. If you used the form of transport mentioned in last question, how
Research Part A: General, 21(3), 203–214. doi:10.1016/0191- long would it take you to reach your destination?
2607(87)90014-8
Yu, B., Ma, Y., Xue, M., Tang, B., Wang, B., Yan, J., & Wei, Y. M. (1) Less than 10 minutes
(2017a). Environmental benefits from ridesharing: A case of Beijing. (2) 11–20 minutes
Applied Energy, 191, 141–152. (3) 21–30 minutes
Yu, B., Xue, M., Du, Y., Wang, B., Tang, B., Guan, D., & Wei, Y. M. (4) 31–40 minutes
(2017b). Roadmap of low-emission ride-sourcing travel in a global (5) 41–50 minutes
megacity: The case of Beijing. Working paper. (6) 51–60 minutes or more
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 15

9. What is your walking time to the nearest public transport station? 13. What is your annual household income range (RMB, after taxes)?

(1) Less than 10 minutes (1) Less than 50,000


(2) 11–20 minutes (2) (2) 50–100,000
(3) 21–30 minutes (3) (3) 100–150,000
(4) 31–40 minutes (4) (4) 150–250,000
(5) 41–50 minutes (5) (5) More than 250,000
(6) 51–60 minutes
(7) More than 60 minutes
14. What is your highest education level?
10. How many private cars do you own? ( )
11. If traffic, limits on car purchases and parking issues became
increasingly more severe, but app-based ride-hailing services were (1) Below high school
permanently available, would you change your mind regarding (2) High school or equivalent
plans to buy a car (new car or replacement for an old car)? (3) University or equivalent
(4) Graduate school
(1) Would not change
(2) May not change
15. What is your household type?
(3) May change
(4) Would change
(1) Live alone
12. Do you live in CBD/central area? (2) Live with spouse
(3) Two-generation family
(1) Yes (4) Three- or more generation family
(2) No (5) Other

You might also like