Content uploaded by Sylvia He
Author content
All content in this area was uploaded by Sylvia He on Apr 09, 2020
Content may be subject to copyright.
1
Evaluating Hong Kong’s spatial planning in new towns from the perspectives of
job accessibility, travel mobility, and work–life balance
Sylvia Y. He a, *, Sui Tao b, Mee Kam Ng a,b, Hendrik Tieben c
*
Corresponding author
a
Department of Geography and Resource Management
The Chinese University of Hong Kong
Shatin, NT, Hong Kong
Tel.: +852 3943 6646
Fax: +852 2603 5006
sylviahe@cuhk.edu.hk
b
Institute of Future Cities
The Chinese University of Hong Kong
Shatin, NT, Hong Kong
c
School of Architecture
The Chinese University of Hong Kong
Shatin, NT, Hong Kong
2
Abstract
Problem, Research Strategy, and Findings: During the 1970s–1990s, the Hong
Kong government developed several new towns to alleviate pressure on contemporary
urban areas and provide housing for a rapidly growing population. Although self-
containment is a key objective in building new towns, no research has assessed
residents’ behavioral outcomes, limiting the objective assessment of the spatial
planning of new towns. In this study, we hypothesize that adequate job accessibility
would offer residents shorter commutes and hence more time for non-commute travel
and activities and better work–life balance. Drawing on census and household travel
survey data, we assess the spatial planning of new towns by investigating the effects
of job accessibility on commute and non-commute travel durations. We find that (1)
there is a disparity in self-containment and access to job opportunities between these
towns and urban areas, (2) job accessibility strongly impacts commute duration, and
(3) prolonged commutes can reduce non-commute travel duration, particularly for
public-transportation users, suggesting that this may harm work–life balance for
workers with long commutes.
Takeaway for Practice: Building a self-contained new town requires strategic spatial
planning and a concrete plan to develop the local economy. To cultivate local
employment, planners should develop a regional plan that differentiates main
industries in different new towns. A good starting point would be a thorough
understanding of the history, spatial distribution of existing industries and firms, and
skills of local workers in these towns. We recommend more proactive efforts to 1)
establish more self-contained new town communities, 2) relocate jobs in certain
sectors (e.g., government offices) to new towns; 3) strengthen an integrated transit
system, and 4) adopt alternative work schedules (e.g., telecommuting, flexible work
hours) in certain industries to relieve commute burdens, improving both commute
experience and work–life balance.
Keywords: commute; job accessibility; new town; travel mobility; work–life balance
3
Introduction
The official creation of and legislation for new towns in the UK dates to the
enactment of the New Town Acts in 1946 (Osborn & Whittick, 1977; Wannop, 1999),
when the population of London increased dramatically and severe urban problems,
such as overcrowding, pollution, and congestion surfaced. The original principal
purpose of new towns was to accommodate the “overspill” of populations from
metropolitan areas (Wannop, 1999). Early generations of new towns were created
largely following Ebenezer Howard’s (1898) envisioned garden cities, where
urbanites could “escape from the biting reality of our complex urban problems”
(Alonso, 1970a, p. 3).
Achieving the status of “self-containment” is often a key criterion for new
towns to be considered a success (Alonso, 1970b; Wannop, 1999). For example,
Letchworth, UK, was regarded as a “conspicuous success” partially because it
balanced employment and quality of life the workers who lived nearby (Osborn &
Whittick, 1977, p. 24). A similar claim was made in a report published by the New
Towns Committee (cited in Wannop, 1999). If new towns become self-contained, a
direct behavioral outcome is short commutes for workers, along with relatively ample
time for non-commute travel and activities such as leisure and entertainment, which
arguably can lead to a better work–life balance (Kalliath & Brough, 2008; Abendroth
& Dulk, 2011). However, despite over 70 years of practice in new town development,
relatively limited behavior-oriented appraisal research has focused on both commute
experience (e.g., Lau & Chiu, 2013; Yang et al., 2017) and work-–life balance.
To fill this research gap, we assess spatial planning from the perspectives of
job accessibility and commuters’ travel patterns, which serve as dimensional
measures of their work–life balance. Job accessibility and its impact on the daily
commute have been examined extensively in the Euro-American context (Levinson,
1998; Shen 2000; Wang, 2000; Kawabata & Shen, 2007; Hu, 2015); however, more
evidence is needed in the Asian context (e.g., Fan, Allen, & Sun, 2014) to better
understand how job accessibility affects commute experience in high density, transit-
oriented cities. Moreover, job accessibility, which reflects the spatial ease of reaching
and distribution of job opportunities, arguably plays a key role in how people arrange
their non-work travel and activities. Hence, the influences of job accessibility and
4
urban development on commute and non-commute travel patterns deserve further
investigation with a particular focus on new towns.
We chose Hong Kong, a former British colony in which spatial planning still
reflects the influence of its past colonizer (Osborn & Whittick, 1977; Forsyth &
Peiser, 2019). As Osborn & Whittick (1977) observed, the concept of new towns has
seen both increasing adoption in planning practice and widespread scholarly interest.
Since the 1970s, soon after the peak of interest in new towns in the UK (Wannop,
1999), the Hong Kong government initiated a series of new town development
projects to accommodate the rapidly growing population. Yet, due to economic
restructuring, new towns have seen a substantial reduction of manufacturing jobs,
which has resulted, in turn, in longer-distance job searches and commutes for new-
town residents (Lau, 2010; Tao, He, Kwan, & Luo, 2020). We need to learn how to
build new towns that are flexible and able to cope with economic restructuring.
Achieving such flexibility has implications for new town developments in other
regions.
Despite the importance of coordinating a job and housing balance, it is
unknown whether or not spatial planning in Hong Kong has contributed to the spatial
barriers to job opportunities and influenced work–life balance, particularly for new-
town commuters (Forrest, La Grange, & Yip, 2004; Kwok & Yeh, 2004; Hui & Lam,
2005). Investigating these issues in Hong Kong adds complementary insights to new-
town literature in Western contexts (e.g., Cervero, 1995a, 1995b). While Hong
Kong’s new-town planning was heavily influenced by the British approach (Forsyth
& Peiser, 2019), its highly transit-oriented urban context warrants separate
investigation.
In filling these research gaps, this study addresses the following questions: a)
To what extent has job accessibility changed in new towns between 2001 and 2011?
b) How did job accessibility influence commute duration in 2002 and 2011? c) How
might job accessibility have influenced non-commute travel for new-town and urban-
area commuters in 2002 and 2011? d) How do the implications for work–life balance
differ for these two groups, particularly for the former?
5
Drawing on the 2002 and 2011 Hong Kong Travel Characteristics Surveys
(TCS), this study first examined self-containment (via the independence index) and
job accessibility across new towns in comparison with urban areas. Next, linear
regression models were estimated to investigate the influence of residential location
and job accessibility on commute duration (defined as the total travel time that a
worker spends on his/her round-trip commute); and structural equation models (SEM)
were applied to investigate the interrelationships among job accessibility and
commute and non-commute travel.
1
Particular attention was also paid to detecting
differences between new-town and urban-area commuters. Lastly, some policy
recommendations were proposed for regions with new towns.
What is missing in new-town research on job accessibility, non-commute time,
and work–life balance?
Job accessibility and commute time
Urban form has long been argued to influence people’s travel behaviors (e.g., travel
duration and modal choices; Cervero, 1989; Boarnet & Crane, 2001; Cervero &
Kockelman, 1997). Job accessibility has been highlighted as a core urban-structure
dimension (Shen, 1998; Handy & Niemeier, 1997; Wang, 2000). Job accessibility
refers to the ease of accessing employment opportunities through a land-use
transportation system (encapsulating two main dimensions: the spatial distribution
and number of job opportunities) and the transportation cost incurred in reaching them
(Geurs & van Wee, 2004; Batty, 2009; Merlin & Hu, 2017).
2
A considerable number of studies have investigated the influence of job
accessibility on people’s daily commute (Shen, 1998; Wang, 2000; Páez, Scott, &
Morency, 2012; Cheng & Bertolini, 2013; Wu et al., 2019). Primary focus has been
on the working poor’s difficulties in daily commutes. These studies found that, in
addition to poor access to jobs, the lack of transportation options may also play an
important role in the commute experience of the working poor (Kawabata, 2003;
Hess, 2005; Grengs, 2010; Hu, 2015; Kawabata & Shen, 2007; Jin & Paulsen, 2018;
Cui, Boisjoly, El-Geneidy, & Levinson, 2019). Such concerns for low-income groups
1
We also excluded mandatory trips (such as those related to education) from non-commute travel.
2
For reviews of the literature on job accessibility and its calculation, see Geurs and van Wee (2004)
and Páez et al. (2012).
6
were also revealed by recent empirical studies of Asian contexts, (e.g., Fan et al.,
2014; Zhao, 2015).
While commonly considered a location advantage, living close to job
opportunities does not always translate into shorter commutes. The literature has shed
light on the prevalence of “wasteful commutes,” where daily commute time markedly
exceeds the theoretically optimum commute time (Levine, 1998; Ma & Banister,
2006). Factors causing wasteful commute are diverse, ranging from a lack of suitable
jobs near home to personal preference (Mokhtarian, Salomon, & Redmond, 2001).
We contend that for some commuter groups (e.g., those who rely on public transit),
accessibility to employment still appears to play an important role in commute
experience (in part reducing the amount of times wasted during a commute) and is,
therefore, worth investigation (e.g., Hu & Wang, 2017).
Over the last few decades, Hong Kong has experienced rapid urban growth
and imbalanced urban development, aggravating the imbalance in job accessibility
across the city (Loo & Chow, 2010). In an empirical study, Kwok and Yeh (2004)
found higher job accessibility by public transportation than by cars between 1991 and
1996, although a larger increase was reported for accessibility by cars over the study
period. Using a gravity-based measure, He, Tao, Hou, and Jiang (2018) identified a
marked gap in accessibility by rail transit between urban areas and other parts of
Hong Kong. This study, however, did not consider commute experience.
Work–life balance: commute and non-commute activities
Work–life balance, the distribution of one’s time between work and non-work (e.g.,
family) activities in a relatively equitable manner, is a core component of maintaining
a healthy lifestyle and good quality of life (Greenhaus, Collins, & Shaw, 2003;
Kalliath & Brough, 2008). The inability to engage in non-work trips and activities due
to work-related commitments (e.g., long commutes and work hours) may threaten
one’s wellbeing (Wheatley, 2014). A considerable amount of research in the fields of
organization management and psychology has shown that a number of factors,
including flexible work schedules, location, and household responsibilities, can
significantly influence how people arrange their work and non-work activities, and
7
thus their perceived balance between work and personal/family life (e.g., Hill et al.,
2001; Greenhaus et al., 2003; He, 2013a; Sandow, 2019).
A growing body of transportation studies has considered work–life balance by
studying time-use patterns for daily travel and activities. Certain trade-offs exist
between time allocated to trip-making and associated activities (e.g., shorter travel
and longer activities) and between the time allocated to different activities (e.g., work
versus recreation). Some research has shed light on the trade-off between commute
and other activities (e.g., Chen and Mokhtarian, 2006; Dharmowijoyo, Susilo, &
Karlström, 2016; Solberg & Wong, 1992; Hjorthol & Vågane, 2014). For example,
Denstadli, Julsrud, and Christiansen (2017) revealed a negative effect of commute
time on commute satisfaction, which, in turn, affected satisfaction with work–life
balance. Sandow (2019) found a significant association between long-distance
commutes and separation rates among couples. Similarly, He (2013b) showed that the
work schedules and hours of parents, especially of mothers, affect their ability to
drive their children to school. Other research has addressed this issue by investigating
the effect of flexible work schedules (e.g., telecommuting) in reducing work-related
travel while increasing non-work trips (e.g., Zhu, 2012; Kim, Choo, & Mokhtarian,
2015; He & Hu, 2015). To summarize, we contend that workers with long commutes
because of little access to jobs may also face reduced time in other non-work
activities, possibly resulting in an imbalanced work–life relationship.
Are Hong Kong’s new towns self-contained?
Achieving urban development characterized by a balanced distribution of jobs and
housing as well as other public facilities has long been a key trend in city spatial
planning (Ogilvy, 1968). Some early seminal research examined levels of the self-
containment of new towns and the impact on daily commutes. Ogilvy (1968)
examined some new towns in the UK and found that higher job–housing balance
might still include considerable cross-movement (e.g., working elsewhere) and
possibly wasteful commutes. Cervero (1995a, 1995b) examined the level of self-
containment (i.e., local versus external commutes) of new towns across the UK, the
US, greater Paris, and Greater Stockholm. He found that more self-contained new
towns were indeed associated with shorter commutes. Some other cities have
demonstrated successful spatial planning practice in achieving self-contained new
8
towns such as Ørestad in Copenhagen (Knowles, 2012) and Singapore (Ng, 2018;
Goldblum, 2008). The main similarity between these cases appears to be a sufficient
supply of jobs and well-connected transit that can accommodate daily travel.
In Hong Kong, whether new towns have contributed to the establishment of
self-contained neighborhoods is subject to debate (Lau, 2010). After industrial
activities in Hong Kong (particularly the New Territories) were relocated to the Pearl
River Delta region, limited consideration has been given to nurturing the local
economy in the planning of new towns in a way that would encourage complementary
economic development (Ng, 2008). Thus far, the government has not implemented
any strategic spatial plans to develop the local economy in new towns, and the zoning
of industrial sites fails to enable their best spatial utilization.
Some researchers have examined the daily travel activity patterns of new town
residents in Hong Kong (usually residents of public housing estates or homeowners
who cannot afford to have properties in urban areas). For example, a route test (Lau,
2010) revealed that the average commute duration by public transportation from Tin
Shui Wai, a new town, to Hong Kong Island took as long as 80 minutes, because of
the spatial segregation of residents in remote new towns from their jobs and
employment opportunities (Forrest et al., 2004, Sui, 1995). Similarly, a recent study
found a level of forced mobility among new-town commuters (Tao, He, Kwan, &
Luo, 2020). Further, a long commute usually means less time to spend with family,
which may be especially important for workers with young children and/or disabled
or elderly people at home. Hence, assessing the spatial planning in Hong Kong from
the perspective of job accessibility, travel mobility, and work–life balance arguably
has implications for both Hong Kong and other cities facing economic restructuring.
Methods
Conceptual framework
The main analytical components of this study include comparing job accessibility and
commute and non-commute travel durations between new towns and urban areas.
Figure 1 depicts the conceptual framework indicating the key relationships tested in
this study. We expect new towns to be associated with longer commutes because of
their relatively remote locations. Next, we expect both job accessibility and living in
9
new towns to affect commute duration, but in opposite directions (negative and
positive, respectively). Finally, we expect longer commute duration to influence non-
commute travel duration negatively by reducing opportunities for non-commute travel
and activities. Although new towns and job accessibility may correlate with each
other to some extent, it is also necessary to include a separate variable for the former
to capture its fixed effect on commute duration. This is because of inherent
differences in local contexts (e.g., mixed land use, traffic conditions) between new
towns and urban areas.
Figure 1: Conceptual framework
Model for commute travel
The following methods were used in previous studies that modeled commute patterns
(e.g., Levinson, 1998; Manaugh, Miranda-Moreno, & El-Geneidy, 2010); we modeled
commute duration using multiple linear regression models. We estimated commute
duration to be the sum of the durations of a round-trip commute. Job accessibility and
residential location (particularly residing in a new town) were the main independent
variables of interest. Other independent variables included personal, household, and
neighborhood characteristics.
Model for non-commute travel
In this study, non-commute travel duration was employed as a proxy for work–life
balance, as it is significantly linked to the time allocated to non-work activities (Susilo
& Djist, 2009; Dharmowijoyo et al., 2016). Engaging in non-commute travel, in turn,
10
captures a key dimension of work–life balance (Wheatley, 2014; Hjorthol & Vågane,
2014).
Previous research (e.g., He, Cheung, & Tao, 2018; He & Hu, 2015; Chen &
Mokhtarian, 2006) often classified non-commute trips as either maintenance (e.g.,
going to a market or hospital) or discretionary (e.g., going to a playground or movie
theater) trips. In estimating non-commute travel duration, we selected travel time
between main bases (i.e., home and workplace) in proportion to time spent on
different activities between these bases (a similar method was adopted by Susilo &
Djist, 2009, 2010; Mao, Ettema, & Dijst, 2016). We conducted an analysis of both
overall non-work (discretionary) and purpose-specific (maintenance) travel time
(activity time was not included in the analysis).
To capture the influence of commute duration on non-commute travel
duration, we included job accessibility as a key influencing factor (as shown in Figure
1). The influences of sociodemographic and neighborhood characteristics on commute
and non-commute travel were also controlled for. Further, we included work hours as
an additional independent factor because they have important implications for the
time window and opportunities for non-work travel and activities (Fujii, Kitamura, &
Kishizawa, 1999; Hjorthol & Vågane, 2014). We used SEM as the main tool to fulfill
this analytic goal. SEM allows the simultaneous modeling of interactions between
multiple exogenous and endogenous variables. AMOS 24 software was used to
conduct SEM. Details of operationalizing SEM is provided in the Technical
Appendix.
Measurement of job accessibility
In line with Hansen’s (1959) definition of accessibility as “the potential of
opportunities for interaction,” we consider job accessibility as the spatial distribution
of potential job opportunities and the ease of reaching them through a given
transportation system. A variety of measurements have been operationalized to
capture job accessibility, including commute time (e.g., Guidon, Wicki, Bernauer, &
Axhausen, 2019), cumulative opportunities (e.g., Wang & Chen, 2015), and gravity-
based models (e.g., Guzman, Oviedo, & Rivera, 2017). Some studies have adopted
11
more advanced approaches, such as a doubly constrained model, to operationalize
accessibility (e.g., Cheng & Bertolini, 2013; Merlin & Hu, 2017).
In this study, we used the gravity-based model formulated by Shen (1998) to
measure job accessibility, as it best approximates our definition. Shen’s model is
largely in line with Hansen’s original model (1959), which considers the job
accessibility of a given location to be positively linked to the number of opportunities
available at other locations and to be negatively affected by the transportation cost of
reaching these locations. Shen’s model also improved on the original model by
considering the demand for job opportunities. That is, job seekers in other locations
who may compete for the same job opportunities reduce the availability of these
opportunities.
We estimated job accessibility at the street block (SB) level, the smallest
spatial unit of census in Hong Kong. After excluding open and other non-built-up
areas, the average size of SBs was 2.2–2.3 square kilometers in both 2001 and 2011.
For the details of job accessibility estimation, see the Technical Appendix. We used
the standardized values of accessibility in the final analysis.
Limitations
There are several limitations of the current study that future research may seek to
improve upon. First, it would be worthwhile to directly survey new-town commuters
on perceived levels of satisfaction with work–life balance. Second, we only
considered non-work trips at the individual level. Further research may probe deeper
into the intra-household joint activity–travel patterns, which may provide further
insight into the family life of commuters. Third, we focused on the influence of job
accessibility on time use for commute and non-commute travel. Residential location
choice and self-selection did not constitute part of the focus of the current study. In
the future, it may be worth considering these aspects to provide a more
comprehensive understanding of residential choice and work–life balance. Fourth,
while we employed travel survey data from separate time points, these data were also
drawn from separate samples. To overcome this limitation, collecting a longitudinal
survey over both weekdays and weekends on the same sample will be desirable for
12
future research. Last, future research may draw on more up-to-date data, should they
become available.
Study context and data
Hong Kong is the context for this study (Figure 2). Its total population is over 7
million. The region contains nine new towns spanning three generations (Civil
Engineering and Development Department, 2016). Full-scale development of the first
generation (Tsuen Wan, Sha Tin, and Tuen Mun) began in the early 1970s, the second
(Tai Po, Fanling/Sheung Shui, and Yuen Long) in the late 1970s, and the third
(Tseung Kwan O, Tin Shui Wai, and Tung Chung) in the 1980s and 1990s. As of
2016, over 3.4 million people (47% of the total population) resided in new towns
(Civil Engineering and Development Department, 2016). We used three sets of data:
TCS data, census data, and transportation network data from across Hong Kong. For
details of the data processing, see Technical Appendix.
Figure 2: Study context
The transportation system of Hong Kong consists of rail (known as Mass
Transit Railway [MTR]), bus transit, and other modes of transit (e.g., tram, ferry) and
13
a road network that is shared by buses and cars. In this study, we focused only on the
MTR, buses, and private cars. Around 70% of all daily mechanized trips are by public
transportation, including the MTR and buses, and around 12% by private vehicles
(Hong Kong Transport Department, 2014).
We summarized the commute patterns of public transportation based on 2011
TCS data (See Table 1). A considerable proportion (e.g., 40-60%) of both new-town
and urban-area commuters commute to urban areas (Kowloon and Hong Kong
Island). On average, the commute duration of new-town commuters was markedly
longer than that of their urban-area counterparts, except for Tsuen Wan and Tseung
Kwan O, which are located closer to urban areas. Further, the proportion of
commuters engaging in non-commute travel was slightly higher in Kowloon and
Hong Kong Island than in the new towns. Yet the total minutes of non-commute
travel was about the same for urban and new town residents.
We also estimated the independence index for new towns and other areas
across Hong Kong, which is the ratio of the number of internal commutes to that of
external commutes (Cervero, 1995a; see also Table 1). A higher independence index
indicates higher self-containment. Unsurprisingly, the independence indices in new
towns are markedly lower than in urban areas (e.g., lower than 0.1 versus 0.4 or
higher). Similar patterns largely persist across public-transportation and private-
vehicle commuters (see Appendices 1 and 2). This suggests that new towns are less
self-contained than urban areas. Compared to previous works on new towns, the
independence indices of Hong Kong’s new towns are lower than their UK
counterparts, which have an average independence index of 1.2 (Cervero, 1995a),
though they are on par with some new towns in Greater Stockholm area (between 0.1
and 0.2; Cervero, 1995b).
14
Table 1: Commuting patterns across Hong Kong (2011)
Home location
Workplace
(% of commuters)
Average
commute
duration
(min)
% of commuters
who took non-
commute trips
Average
non-commute
travel
duration
(min)
Independence
index
SNT
ONT
HKI
KLN
ROS
New town
(N=12,771)
Tsuen Wan (N=3,128)
28.32
10.20
21.74
36.22
3.52
94.84
5.08
36.42
0.21
Shatin (N=3,105)
18.52
16.78
20.00
41.29
3.41
103.89
4.32
31.23
0.16
Tuen Mun (N=1,411)
7.23
35.51
18.21
33.95
5.10
122.35
3.97
40.38
0.06
Tai Po (N=958)
9.81
28.50
16.28
40.81
4.59
114.31
4.59
27.74
0.08
Fanling/Sheung Shui (N=898)
8.02
35.75
14.92
33.74
7.57
124.89
5.12
31.70
0.06
Yuen Long (N=456)
8.33
33.11
17.54
31.80
9.21
109.37
3.73
37.20
0.05
Tseung Kwan O (N=1,823)
5.54
10.48
32.80
48.71
2.47
89.67
2.91
25.21
0.05
Tin Shui Wai (N=644)
0.47
41.61
18.17
32.45
7.30
128.34
5.12
39.10
0.00
Tung Chung (N=348)
5.75
16.67
19.25
28.45
29.89
104.60
2.01
27.30
0.05
Summary (N=12,771)
14.81
20.38
21.22
38.60
5.00
105.39
4.30
33.45
0.12
Urban area
(N=13,283)
HK Island (N=5,285)
n.a.
6.85
69.25
22.67
1.23
82.79
5.34
30.99
0.54
Kowloon (N=7,998)
n.a.
17.23
29.59
50.76
2.41
85.10
4.94
35.62
0.39
Summary (N=13,283)
n.a.
13.10
45.37
39.58
1.94
84.18
5.10
33.69
0.45
Rural area
(N=982)
n.a.
45.82
12.93
29.84
11.41
95.79
3.26
38.77
0.06
Total
(N=27,036)
6.99
17.73
32.79
38.76
3.73
94.62
4.65
33.72
0.28
Note: SNT (same new town); ONT (other new town); HKI (Hong Kong Island); KLN (Kowloon); ROS (rural and open space)
15
Results
Because of limited space, we report the results for only the public-transportation
commuters, only briefly mentioning some findings for the private-vehicle commuters
(with details provided in the appendices).
New towns have less job access than the older urban areas
First, we examined spatial patterns of job accessibility across Hong Kong (see figures
in Appendix 3). The revealed patterns show that job accessibility by public
transportation in 2011 (Appendix 3a) was higher in urban areas and lower in suburban
areas, reflecting a centralized job distribution pattern for commuters. Job accessibility
by private vehicles shows a similar pattern (Appendix 3b). A series of t-tests
confirmed lower job accessibility in new towns than in urban areas in both 2001 and
2011.
We also examined changes in job numbers and commuters in the nine new
towns and the other areas between 2001 and 2011 (see Appendix 4). In new towns,
the number of jobs increased by 76,743 and the number of commuters increased by
198,011. New towns closer to urban areas with mass transit, including Tseung Kwan
O (with industrial estates), Tsuen Wan (with industrial rezoning), and Shatin (with the
Science Park) saw the largest increase in job numbers. The number of commuters also
increased considerably in Tseung Kwan O, Tin Shui Wai, Tsuen Wan, and Tung
Chung, all of which had increases of 20,000 public-transportation commuters. In
contrast, urban areas saw relatively small changes in public-transportation (a decrease
of around 20,000) and private-vehicle (an increase of around 10,000) commuters.
TCS data shows subtle demographic and commute changes between 2002 and 2011
The descriptive statistics of the 2002 and 2011 TCS data are detailed in Appendix 5,
which shows relatively comparable sociodemographic compositions though with
some subtle differences. A higher proportion of the participants were from households
with dependent children in 2002 than in 2011; and household size was slightly smaller
in 2011 than in 2002.
A further examination indicates a noticeable decrease in average commute duration
for public-transportation commuters in rural areas between 2002 and 2011 (Appendix
16
6). However, changes for commuters in new towns and urban areas appear modest
(Appendix 6a). Compared to 2002, fewer commuters engage in non-commute (out-of-
home) activities on the surveyed date in 2011 (1,258 versus 1,695) despite a larger
sample being taken. However, among those who engaged in non-commute travel, the
average duration between 2002 and 2011 is similar for new town and urban dwellers
but increased for rural dwellers (Appendix 6b).
New town commuters experience longer commutes than urban and rural dwellers
Table 2 presents the modeling results for commute durations in 2002 and 2011,
respectively (for private-vehicle commuters, see Appendix 7). Two models were
estimated: the first grouped new towns together (with urban areas as the reference
group), and the second used separate dummy variables for different new towns and
urban areas (with Hong Kong Island as the reference group).
As expected, job accessibility had a strong negative effect on commute
duration in all models. Its effect also appears to be stronger for public-transportation
commutes than for private-vehicle commutes, especially in 2011. In line with the
literature (e.g., Kawabata & Shen, 2007; Grengs, 2010), this suggests that the
commute experience of public-transportation commuters was influenced significantly
more by the spatial distribution of jobs than was the commute experience of private-
vehicle commuters, despite improvements in the public-transportation network in the
preceding years.
The effect of new towns is highly significant across all models except for
Model 1 for private vehicles in 2002. The positive effect of new towns indicates that
new-town commuters, whether using public transportation or private vehicles, had
longer commutes overall than their urban-area peers (particularly so when compared
to Hong Kong Island). Kowloon also had longer commutes than Hong Kong Island.
In models using detailed location variables, public transport commuters from all new
towns, except for Tseung Kwan O and Tung Chung, had longer commutes in both
2002 and 2011. In rural areas, public transport commuters had longer commutes in
2002 but by 2011 they were shorter. This inconsistency may reflect changes in the
spatial distribution of commuters and jobs in rural areas. We also found the commute
17
distance (as Euclidean distance) of rural public-transportation commuters was on
average shorter than that of their new-town peers (20.5 km versus 23.6 km).
Among public-transportation commuters, older and female commuters had
shorter commute durations than their younger and male counterparts in basically all
models, which is in line with previous empirical evidence (e.g., Kim, Sang, Chun, &
Lee, 2012). This finding may be attributed to the household responsibilities of female
commuters and the relatively less active life stage of the elderly. Commuters from
public housing estates had longer commutes.
Table 2: Modeling results of commute duration for public-transportation commuters
*** p<0.01; ** p<0.05; * p<0.1
Variable
Coefficient estimates
2002
(N =18,568)
2011
(N =25,098)
Model 1
Model 2
Model 1
Model 2
Age
-0.015*
-0.018**
-0.023***
-0.025***
Female
-0.052***
-0.050***
-0.059***
-0.058***
Married
-0.014
-0.008
-0.015*
-0.011
Public housing estates
0.033***
0.049***
0.019***
0.031***
Household size
-0.010
-0.011
-0.005
-0.005
Having dependent children
0.002
0.001
-0.003
0.003
Having retired elderly
0.003
0.006
0.003
0.004
Having a domestic helper
0.005
0.008
0.011*
0.010
Standardized job
accessibility
-0.315***
-0.175***
-0.305***
-0.209***
Population density (ln)
0.007
-0.032***
-0.045***
-0.042***
Kowloon
0.106***
0.075***
0.081***
0.061***
Rural areas
0.025***
0.063***
-0.059***
-0.020**
New towns
0.166***
--
0.149***
--
Tsuen Wan
--
0.055***
--
0.086***
Shatin
--
0.165***
--
0.141***
Tuen Mun
--
0.143***
--
0.139***
Tai Po
--
0.124***
--
0.114***
Fanling/Sheung Shui
--
0.099***
--
0.118***
Yuen Long
--
0.076***
--
0.050***
Tseung Kwan O
--
0.028***
--
0.004
Tin Shui Wai
--
0.147***
--
0.104***
Tung Chung
--
0.035***
--
-0.001
Model summary
F
241.00***
170.65***
323.03***
230.49***
Adjusted R2
0.1439
0.1610
0.1430
0.1611
18
We also estimated job accessibility and modeled the commute duration for
commuters in five main industries across Hong Kong, which accounted for over 90
percent of the working population (see Appendix 8). The effects of new towns and job
accessibility on commute duration were mostly significant and in expected directions
across the five industries, except for an insignificant effect of new towns in 2002 (for
“transport, storage, and communications”). For private-vehicle commuters (Appendix
9), the effect of new towns was mostly negligible, while a significant effect of job
accessibility was found for some industries. Furthermore, the effect of job
accessibility on commute duration appears particularly strong for “financing,
insurance, real estate, and business” across all models. Further examination shows
that this may in part be because financing jobs are highly concentrated in urban areas
(particularly in the central business district on Hong Kong Island), while the
associated working population is spatially dispersed (see the figures in Appendix 10).
Longer commutes of new town residents leads to less time for non-commute travel
and poorer work-life balance
For this part of analysis, we focus only on those who made non-commute trips,
leaving us a much smaller sample size. Our SEM models follows the conceptual
framework illustrated in Figure 1, with correlations between job accessibility and
other sociodemographic and neighborhood variables were also added.
Figure 3 summarizes the results of the SEM models in 2002 and 2011,
respectively, in the form of path diagrams (for modeling results for private-vehicle
commuters, see Appendix 11). In these figures, the numbers above the arrows indicate
the standardized coefficients and the numbers in parentheses are the square multiple
correlations, which suggest the percentages of variance explained for commute and
non-commute travel durations. Acceptable model fit was achieved (in 2002: CFI =
0.971, RMSEA = 0.051, χ2/df = 4.331; in 2011: CFI = 0.991, RMSEA = 0.031, χ2/df
= 1.996). About 9%–10% of the variance in non-commute travel duration was
captured in the SEM models, which is not entirely unexpected given the less regular
nature of non-commute trips (e.g., He & Hu, 2015; Buliung, Roorda, & Remmel,
2008).
19
The modeling results suggest that the key relationships of interest (see Figure
1) are largely consistent in 2002 and 2011. For commute duration, the effects of job
accessibility and residing in new towns remained the strongest and the sign of
coefficients is all expected. In most cases, we found that non-commute travel duration
was strongly affected by commute duration and work hours, which largely confirmed
our expectation. Further, longer commute duration reduced non-commute travel
duration more significantly for public transport commuters than for private-vehicle
commuters, particularly in 2011 (see Appendix 11). This indicates that during the
study period public-transportation commuters were more disadvantaged. With respect
to sociodemographic characteristics, older people and women had shorter non-
commute travels, particularly among public-transportation commuters in 2002, but
less so in 2011. This may indicate changing social roles for commuters in these
groups over the decade (e.g., an increased number of older people and women are
engaging in the labor force). In both 2002 and 2011, of those commuting in private
vehicles, older people spent less time on non-commute trips than younger people.
We also estimated the indirect effect of new towns on non-commute travel
duration. For public-transportation commuters, the standardized indirect effect of
living in new towns on non-commute travel duration was -0.032 in 2002 and -0.039 in
2011, suggesting that they spent less time on non-commute travel than their urban-
area peers, all else being equal. For private-vehicle commuters, the indirect effect of
new towns on non-commute travel durations was -0.019 in 2002 and 0.006
(statistically insignificant) in 2011.
20
Figure 3: Modeling results of non-commute travel duration for public-transportation
commuters
Note:
*** p < 0.01; ** p < 0.05; * p < 0.1.
Numbers in parentheses refer to the percentages of variance explained.
Model fit for Model (a): N = 1,292, df = 18, CFI = 0.971, RMSEA = 0.051, χ2 = 77.957, p<0.001,
χ2/df = 4.331.
Model fit for Model (b): N = 1,026, df = 18, CFI = 0.991, RMSEA = 0.031, χ2 = 35.937, p<0.001,
χ2/df = 1.996.
To gain more insights, we split non-commute trips into discretionary and
maintenance categories and estimated separate models for the two (see Appendices 12
and 13). Acceptable model fits were achieved across most models, except for private
vehicles in 2011. Again, for public transport commuters, new town residence and
lower job access increased commute duration, which in turn reduced discretionary and
maintenance trip time in 2002 and 2011. In addition, the effect of commute duration
on discretionary trips appeared to be stronger in 2011 than in 2002, while the reverse
was true for maintenance trips. For private-vehicle commuters (see Appendices 14
and 15), weak relationships were found among new towns, commute duration, and
(both overall and category-specific) non-commute travel, especially in 2011. This
might be related to the expansion of the road network and the flexibility of owning a
car. We did not conduct an industry-specific analysis due to the small sample size of
the industries.
21
Discussion
Our analysis indicated that the level of self-containment of the new towns was
markedly lower than both the urban areas and some overseas counterparts (e.g, UK as
in Cervero, 1995a). On average, new towns were also associated with longer
commutes than urban areas. These findings confirm our speculation that, for new-
town commuters, larger spatial separation exists between their residential locations
and job opportunities than for urban-area residents. In some new towns (e.g., Tin Shui
Wai, Tseung Kwan O), there are signs of a narrowing gap in job accessibility,
particularly as a result of the increase in local job numbers and rail transit expansion.
Yet this appears to have changed the overall picture only marginally in terms of job
distribution across the city.
The modeling results further revealed that job accessibility played a more
important role in influencing commute duration for public-transportation commuters
than for private-vehicle commuters over the study period, suggesting that the
commute experience of the former group has largely relied on the connection and
quality of public transportation services. Residents of new towns had longer commute
times, confirming the disadvantage of new towns. Further, these findings appear to
hold across commuters of different industries, with a particularly strong effect
detected among commuters in the financing industry. SEM models offer insight into
the trade-off between commute and non-commute travel conditioned on job
accessibility and residential location. Specifically, time spent commuting and at work
reduced non-commute travel time (both overall and purpose-specific). Also, our
models provide evidence for the cascading effect of job accessibility on non-commute
time by prolonging commute duration. Last, new town public transit commuters
tended to have both longer commute times and lower job accessibility, indirectly
lowering time spent in non-commute travel. Taking these findings together, we
conclude that lower job accessibility coupled with longer commutes may have made it
more difficult for new-town commuters, especially those relying on public transport,
to engage in non-commute travel and activities on workdays than for urban-area
commuters. The lack of adequate non-commute travel implies that new town
22
commuters are more deprived of the right to a better balance between work and
quality of life compared to their counterparts who live in urban areas.
Policy recommendations
While this study focuses on new towns, our findings may be of interest and value to
scholars and practitioners focusing on other types of suburban areas (Forsyth, 2012)
across the world. Based on the analytical results, we propose some recommendations
that may help improve commute experience and work–life balance in new towns and
job-deficient suburbs. First, attempts should be made to create more self-contained
communities through balancing housing resources and job opportunities where
possible. More jobs from urban areas should be decentralized to the new towns, such
as, for example, relocating certain governmental offices to new towns and nurturing
local economies to satisfy people’s daily needs. However, job relation may not be
realistic for some industries (e.g., financing) that are highly subject to the effect of the
agglomeration economy. Given this, alternative strategies like flexible working
arrangements and telecommuting can be experimented with to help reduce commute
burdens (He, 2013a; He & Hu, 2015). Attention, however, should still be paid to
improving job accessibility by public transportation, particularly in some remote new
towns, through such means as more direct transit services to both urban areas and
other emerging job centers.
At a more local level, which may also resonate with the situation in many
other places such as master planned communities, urban renewal in new towns offers
some opportunities to promote the self-containment of local neighborhoods. As a
result of de-industrialization in new towns since the 1980s, some industrial buildings
(e.g., factories and warehouses) have been deserted, and their zoning codes have
become obsolete (Xian & Chen, 2015; Yeh & Xu, 2006). Transforming these
buildings for other uses should stimulate local employment and economic activities,
such as by providing rental space for small businesses (Xian & Chen, 2015). The
government could also consider subsidizing the rent in such underutilized industrial
buildings to make them more affordable for start-ups with budget constraints.
Furthermore, future new towns should be designed with a higher degree of flexibility
to allow them to be more adaptive and resilient when economic restructuring occurs.
In cases where a new town has failed to achieve self-containment, more research
23
would be needed to examine the reasons why and to reflect on the lessons learned.
Whether self-containment can be achieved for all new towns through spatial planning
is still debatable.
Our study provides some new evidence supporting the establishment of more
self-contained neighborhoods and the pursuit of higher job accessibility as a key
planning goal on the grounds of improving daily commutes and work–life balance.
Echoing Cervero’s (1995a) speculation, it appears that whether or not self-contained
new towns can help improve commutes and work–life balance also depends on the
transportation system and particularly, the role and configuration of public
transportation. This study shows that the commute experience of public-transportation
commuters relies more on their connection to jobs via transit services, which
influences their non-commute travel more strongly than it does that of private-vehicle
commuters. Further, some successful examples of new town planning, including
Singapore and Ørestad in Copenhagen, are characterized by highly integrated and
well-connected transit systems. Therefore, in cities where the establishment of new
towns or suburbs is underway, efforts should be made to ensure that reliable transit
connections between main residential clusters and employment centers will be in
place. Last, developing mixed communities would be a good strategy for stimulating a
community’s economy, which should be borne in mind when planning new towns and
suburban communities.
Conclusion
When the British government finally adopted the new town program in 1954, it was
supposed “to be continued, with annual reviews of the implications of general
industrial policy and the rate of housebuilding, preventing housing running ahead of
industrial growth, inducing firms to the development areas whenever possible and
ensuring that London new towns drew industry from Greater London” (Wannop,
1999, p. 214). To review the outcome of this lack of new town industrial and local
economy policies in the postcolonial context of Hong Kong, we conducted our
research mainly from the perspectives of job accessibility and commuter travel
patterns.
24
Accessibility to job opportunities has important implications for the
performance and social equity of a city. In Hong Kong, though it would be ideal for
people residing in new towns to stay in those towns to work, this has not been
achieved, largely because most manufacturing facilities were relocated in mainland
China or other regions of Asia where cheaper labor costs and more affordable rent
were available. This may have created a disparity in the daily commute experiences of
people living in new towns and their urban counterparts. Those who endure prolonged
commutes may also face higher stress in their daily lives because they have less time
for other non-work activities.
To enhance spatial planning in new towns or suburban areas, our study has
presented an in-depth investigation into the self-containment and job accessibility of
new towns and their effects on both commute and non-commute travel and duration –
one of the first, as far as we are aware, in the context of Hong Kong and broader new-
town literature. Our results confirm that living in new towns is associated with poorer
job–housing balance, lower job accessibility, and longer commute time, resulting in
less non-commute time that can be spent with family and friends or for self-
development. The empirical findings of this study to some extent confirm the
prevalence of spatial inequality experienced by new-town commuters and its potential
to upset their work–life balance.
The findings prove the importance of balanced job–housing distribution in
new towns or suburban areas. We reveal that some new towns have successfully
attracted more jobs than others for a variety of reasons, such as being closer to an
urban core, having a stronger initial industrial base, or simply having a longer history
of development. While planning efforts may take time to become effective, we should
aim to improve the transportation connection and accessibility between new towns
and employment centers in the short run. For more just and equitable cities, we, as
planners, should spare no effort to improve job accessibility, commute experience,
and work–life balance for our residents, whether they live in urban areas or new
towns.
Acknowledgements:
25
The authors thank the three anonymous reviewers and Editor Ann Forsyth for their
constructive comments and advice.
Funding details:
This work was supported by the General Research Fund (GRF) of the Hong Kong
Research Grants Council under Grants #CUHK14602017 and #CUHK14652516.
REFERENCES
Alonso, W. (1970a). The mirage of new towns. The Public Interest, 19, 3–17.
Alonso, W. (1970b). What are the new towns for? Urban Studies, 7, 37–55.
Batty, M. (2009). Accessibility: In search of a unified theory. Environment and
Planning B: Planning and Design, 36, 191–194.
Boarnet, M., & Crane, R. (2001). The influence of land use on travel behavior:
Specification and estimation strategies. Transportation Research Part A: Policy
and Practice, 35(9), 823–845.
Buliung, R. N., Roorda, M. J., & Remmel, T. K. (2008). Exploring spatial variety in
patterns of activity-travel behaviour: Initial results from the Toronto Travel-
Activity Panel Survey (TTAPS). Transportation, 35(6), 697–722.
Cervero, R. (1989). Jobs-housing balancing and regional mobility. Journal of the
American Planning Association, 55(2), 136–150.
Cervero, R. (1995a). Planned communities, self-containment and commuting: A
cross-national perspective. Urban Studies, 32(7), 1135–1161.
Cervero, R. (1995b). Sustainable new towns: Stockholm's rail-served satellites. Cities,
12(1), 41-51.
Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity,
and design. Transportation Research Part D: Transport and Environment, 2(3),
199–219.
Chen, C., & Mokhtarian, P. L. (2006). Tradeoffs between time allocations to
maintenance activities/travel and discretionary activities/travel. Transportation,
33(3), 223–240.
Cheng, J., & Bertolini, L. (2013). Measuring urban job accessibility with distance
decay, competition and diversity. Journal of Transport Geography, 30, 100–109.
Cui, B., Boisjoly, G., El-Geneidy, A., & Levinson, D. (2019). Accessibility and the
journey to work through the lens of equity. Journal of Transport Geography, 74,
269–277.
Denstadli, J. M., Julsrud, T. E., & Christiansen, P. (2017). Urban commuting–A threat
to the work-family balance? Journal of Transport Geography, 61, 87–94.
Dharmowijoyo, D. B., Susilo, Y. O., & Karlström, A. (2016). Relationships among
discretionary activity duration, its travel time spent and activity space indices in
the Jakarta Metropolitan Area, Indonesia. Journal of Transport Geography, 54,
148–160.
Fan, Y., Allen, R., & Sun, T. (2014). Spatial mismatch in Beijing, China: Implications
of job accessibility for Chinese low-wage workers. Habitat International, 44,
202–210.
Forrest, R., La Grange, A., & Yip, N. M. (2004). Hong Kong as a global city? Social
distance and spatial differentiation. Urban Studies, 41(1), 207–227.
Forsyth, A. (2012). Defining suburbs. Journal of Planning Literature, 27(3), 270-281.
Forsyth, A. & Peiser, R. (2019). The British new towns: Lessons for the world from
the new-town experiment. Town Planning Review, 90, 239–246.
26
Fujii, S., Kitamura, R., & Kishizawa, K. (1999). Analysis of individuals’ joint-activity
engagement using a model system of activity-travel behavior and time use.
Transportation Research Record: Journal of the Transportation Research Board,
1676, 11–19.
Geurs, K. T., & Van Wee, B. (2004). Accessibility evaluation of land-use and
transport strategies: Review and research directions. Journal of Transport
Geography, 12(2), 127–140.
Goldblum, C. (2008). Planning the world metropolis on an island-city scale: Urban
innovation as a constraint and tool for global change. In T.-C. Wong, B. Yuen, C.
& Goldblum (Eds.), Spatial planning for a sustainable Singapore (pp. 17–29).
Dordrecht: Springer.
Greenhaus, J. H., Collins, K. M., & Shaw, J. D. (2003). The relation between work–
family balance and quality of life. Journal of Vocational Behavior, 63(3), 510–
531.
Grengs, J. (2010). Job accessibility and the modal mismatch in Detroit. Journal of
Transport Geography, 18(1), 42–54.
Guidon, S., Wicki, M., Bernauer, T., & Axhausen, K. (2019). The social aspect of
residential location choice: On the trade-off between proximity to social contacts
and commuting. Journal of Transport Geography, 74, 333–340.
Guzman, L. A., Oviedo, D., & Rivera, C. (2017). Assessing equity in transport
accessibility to work and study: The Bogotá region. Journal of Transport
Geography, 58, 236–246.
Handy, S. L., & Niemeier, D. A. (1997). Measuring accessibility: An exploration of
issues and alternatives. Environment and Planning A, 29(7), 1175–1194.
Hansen, W. G. (1959). How accessibility shapes land use. Journal of the American
Institute of Planners, 25(2), 73–76.
He, S.Y. (2013a). Does flexitime affect departure time choice for morning home-
based commuting trips? Evidence from two regions in California, Transport
Policy, 25, 210-221.
He S.Y. (2013b). Will you escort your child to school? The effect of spatial and
temporal constraints of parental employment, Applied Geography, 42, 116-123.
He, S. Y. & Hu, L. (2015). Telecommuting, income, and out-of-home activities.
Travel Behaviour and Society, 2(3), 131–147.
He S.Y., Cheung Y.H.Y., Tao S. (2018). Travel mobility and social participation
among older people in a transit metropolis: A socio-spatial-temporal perspective,
Transportation Research Part A, 118, 608-626.
He, S. Y., Tao, S., Hou, Y., & Jiang, W. (2018). Mass Transit Railway, transit-
oriented development and spatial justice: the competition for prime residential
locations in Hong Kong since the 1980s. Town Planning Review, 89(5), 467–493.
Hess, D. B. (2005). Access to employment for adults in poverty in the Buffalo-
Niagara region. Urban Studies, 42(7), 1177–1200.
Hill, E. J., Hawkins, A. J., Ferris, M., & Weitzman, M. (2001). Finding an extra day a
week: The positive influence of perceived job flexibility on work and family life
balance. Family relations, 50(1), 49–58.
Hjorthol, R., & Vågane, L. (2014). Allocation of tasks, arrangement of working hours
and commuting in different Norwegian households. Journal of Transport
Geography, 35, 75–83.
Hong Kong Directory (2001). Hong Kong Directory (7th ed.). Hong Kong: Universal
Publications, Ltd. (in Chinese).
27
Hong Kong Transport Department (2014). Travel Characteristics Survey 2011 Final
Report, Hong Kong, Transport Department.
Howard, E. (1898). To-Morrow: A Peaceful Path in Real Reform. London: Routledge.
Hu, L. (2015). Job accessibility of the poor in Los Angeles: Has suburbanization
affected spatial mismatch? Journal of the American Planning Association, 81(1),
30–45.
Hu, L., & Wang, L. (2017). Housing location choices of the poor: Does access to jobs
matter? Housing Studies, 1–25.
Hui, E. C., & Lam, M. C. (2005). A study of commuting patterns of new town
residents in Hong Kong. Habitat International, 29(3), 421–437.
Jin, J., & Paulsen, K. (2018). Does accessibility matter? Understanding the effect of
job accessibility on labour market outcomes. Urban Studies, 55(1), 91–115.
Kalliath, T., & Brough, P. (2008). Work–life balance: A review of the meaning of the
balance construct. Journal of Management and Organization, 14(3), 323–327.
Kawabata, M. (2003). Job accessibility by travel mode in US metropolitan areas.
Theory and Applications of GIS, 11(2), 165–172.
Kawabata, M., & Shen, Q. (2007). Commuting inequality between cars and public
transit: The case of the San Francisco Bay Area, 1990–2000. Urban Studies,
44(9), 1759–1780.
Kim, C., Sang, S., Chun, Y., & Lee, W. (2012). Exploring urban commuting
imbalance by jobs and gender. Applied Geography, 32(2), 532–545.
Kim, S. N., Choo, S., & Mokhtarian, P. L. (2015). Home-based telecommuting and
intra-household interactions in work and non-work travel: A seemingly unrelated
censored regression approach. Transportation Research Part A: Policy and
Practice, 80, 197–214.
Knowles, R. D. (2012). Transit oriented development in Copenhagen, Denmark: From
the finger plan to Ørestad. Journal of Transport Geography, 22, 251–261.
Kwok, R. C., & Yeh, A. G. (2004). The use of modal accessibility gap as an indicator
for sustainable transport development. Environment and Planning A, 36(5), 921–
936.
Lau, J. C. Y. (2010). The influence of suburbanization on the access to employment
of workers in the new towns: A case study of Tin Shui Wai, Hong Kong. Habitat
International, 34(1), 38–45.
Lau, J. C. Y., & Chiu, C. C. (2013). Dual-track urbanization and co-location travel
behavior of migrant workers in new towns in Guangzhou, China. Cities, 30, 89-97.
Levine, J. (1998). Rethinking accessibility and jobs-housing balance. Journal of the
American Planning Association, 64(2), 133–149.
Levinson, D. M. (1998). Accessibility and the journey to work. Journal of Transport
Geography, 6(1), 11–21.
Loo, B. P., & Chow, A. S. (2011). Spatial restructuring to facilitate shorter
commuting: An example of the relocation of Hong Kong International
Airport. Urban Studies, 48(8), 1681–1694.
Ma, K. R., & Banister, D. (2006). Excess commuting: A critical review. Transport
reviews, 26(6), 749–767.
Manaugh, K., Miranda-Moreno, L. F., & El-Geneidy, A. M. (2010). The effect of
neighbourhood characteristics, accessibility, home–work location, and
demographics on commuting distances. Transportation, 37(4), 627–646.
Mao, Z., Ettema, D., & Dijst, M. (2016). Analysis of travel time and mode choice
shift for non-work stops in commuting: Case study of Beijing, China.
Transportation, 45(3), 751–766.
28
Merlin, L. A., & Hu, L. (2017). Does competition matter in measures of job
accessibility? Explaining employment in Los Angeles. Journal of Transport
Geography, 64, 77–88.
Mokhtarian, P. L., Salomon, I., & Redmond, L. S. (2001). Understanding the demand
for travel: It’s not purely ‘derived’. Innovation: The European Journal of Social
Science Research, 14(4), 355–380.
Ng, M. K. (2008). From government to governance? Politics of planning in the first
decade of the Hong Kong Special Administrative Region. Planning Theory and
Practice, 9(2), 165–185.
Ng, M. K. (2018). The art of planning theory and practice in Singapore. Planning
Theory & Practice, 19(3), 319–323.
Ogilvy, A. A. (1968). The self-contained new town. Town Planning Review, 39(1),
38–54.
Osborn F. J. & Whittick, A. (1977). New towns. London: Leonard Hill.
Páez, A., Scott, D. M., & Morency, C. (2012). Measuring accessibility: Positive and
normative implementations of various accessibility indicators. Journal of
Transport Geography, 25, 141–153.
Public Transport Atlas (2004). Public transport atlas (1st ed.). Hong Kong: Universal
Publications, Ltd. (in Chinese).
Public Transport Atlas (2010). Public transport atlas (6th ed.). Hong Kong: Universal
Publications, Ltd. (in Chinese).
Sandow, E. (2019). Til work do us part: The social fallacy of long-distance
commuting. In C. L. Scholten & T. Joelsson (Eds.), Integrating Gender into
Transport Planning (pp. 121–144). Cham: Palgrave Macmillan.
Shen, Q. (1998). Location characteristics of inner-city neighborhoods and
employment accessibility of low-wage workers. Environment and Planning B:
Planning and Design, 25(3), 345–365.
Shen, Q. (2000). Spatial and social dimensions of commuting. Journal of the
American Planning Association, 66(1), 68–82.
Solberg, E. J., & Wong, D.C. (1992). Family time use: Leisure, home production,
market work, and work related travel. Journal of Human Resources, 27(3), 485–
510.
Sui, D. Z. (1995). Spatial economic impacts of new town development in Hong Kong:
A GIS-based shift-share analysis. Socio-Economic Planning Sciences, 29(3), 227–
243.
Susilo, Y. O., & Dijst, M. (2009). How far is too far? Travel time ratios for activity
participation in the Netherlands. Transportation Research Record, 2134(1), 89–
98.
Tao, S., He, S.Y., Kwan, M. P., & Luo, S. (2020). Does low income translate into
lower mobility? An investigation of activity space in Hong Kong between 2002
and 2011. Journal of Transport Geography, 82, 102583
(https://doi.org/10.1016/j.jtrangeo.2019.102583).
Wang, F. (2000). Modeling commuting patterns in Chicago in a GIS environment: A
job accessibility perspective. The Professional Geographer, 52(1), 120–133.
Wannop, U. (1999). New towns. In B. Cullingworth (Ed.), British Planning: 50 Years
of Urban and Regional Policy (pp. 213–231). London: Athlone Press.
Wheatley, D. (2014). Travel-to-work and subjective well-being: A study of UK dual
career households. Journal of Transport Geography, 39, 187–196.
29
Wu, S. S., Zhuang, Y., Chen, J., Wang, W., Bai, Y., & Lo, S. M. (2019). Rethinking
bus-to-metro accessibility in new town development: Case studies in Shanghai.
Cities, 94, 211-224.
Xian, S., & Chen, H. (2015). Revitalisation of industrial buildings in Hong Kong:
New measures, new constraints? Habitat International, 47, 298–306.
Yang, X., Day, J. E., Langford, B. C., Cherry, C. R., Jones, L. R., Han, S. S., & Sun,
J. (2017). Commute responses to employment decentralization: Anticipated versus
actual mode choice behaviors of new town employees in Kunming, China.
Transportation Research Part D: Transport and Environment, 52, 454-470.
Yeh, A. G. O., & Xu, J. (2006). Turning of the dragon head: Changing role of Hong
Kong in the regional development of the Pearl River Delta. In A. G. O. Yeh, V.
Sit, G. Chen, & Y. Zhou (Eds.), Developing a competitive Pearl River Delta in
South China under one country-two systems (pp. 63–95). Hong Kong: Hong Kong
University Press.
Zhao, P. (2015). The determinants of the commuting burden of low-income workers:
Evidence from Beijing. Environment and Planning A: Economy and Space, 47(8),
1736–1755.
Zhu, P. (2012). Are telecommuting and personal travel complements or
substitutes? The Annals of Regional Science, 48(2), 619–639.
30
List of Appendices:
Technical Appendix
Appendix 1: Commuting patterns of public transportation in Hong Kong (2011)
Appendix 2: Commuting patterns of private vehicles in Hong Kong (2011)
Appendix 3: Standardized job accessibility in 2011
Appendix 4: Comparison of job numbers and commuters between 2001 and 2011
Appendix 5: Descriptive statistics of the samples
Appendix 6: Average commute and non-commute travel duration in 2002 and 2011
for public-transportation commuters
Appendix 7: Modeling results for commute duration for private-vehicle commuters
Appendix 8: Modeling results of commute duration for public-transportation
commuters by main industries
Appendix 9: Modeling results for commute duration for private-vehicle commuters of
different industries
Appendix 10: Spatial distribution of jobs and working population in the Financing
industry
Appendix 11: Modeling results for non-commute travel duration for private-vehicle
commuters
Appendix 12: Modeling results for discretionary trip duration for public-
transportation commuters
Appendix 13: Modeling results for maintenance trip duration for public-transportation
commuters
Appendix 14: Modeling results for discretionary trip duration for private-vehicle
commuters
Appendix 15: Modeling results for maintenance trip duration for private-vehicle
commuters
31
Technical Appendix
Data processing:
Over 70% of Hong Kong is open space or undeveloped, including country parks and
reservoirs. As these areas comprise a very small working population and a very small
number of jobs, we excluded these areas. Next, we generated SB centroids from the
remaining urban areas. We divided the working population and job numbers within
each TPU into corresponding SBs in proportion to their areas.
From the census data, data on the working population and job numbers were
extracted in conjunction with the transportation data to estimate job accessibility
across Hong Kong. The census data were provided at the tertiary planning unit (TPU)
level. To enable a more geographically detailed estimation of job accessibility, we
split the TPU-level census data into SBs using the area-weighted approach, as we also
did for the working population and number of jobs.
Since the census data only provided working population and job numbers for
2001 and 2011, we focused on the transportation networks in these years. For the road
and bus networks, we collected hardcopies of maps of the road network in 2001
(Hong Kong Directory, 2001) and bus service (lines and stops) in 2004 (Public
Transport Atlas, 2004) and 2010
3
(Public Transport Atlas, 2010) from Hong Kong’s
local atlas of transportation systems and manually digitized these maps. Then, we
integrated the digitized bus maps into the MTR networks (by establishing pedestrian
links between the two) to represent the public transportation networks in 2001 and
2011.
In the TCS, commuters (i.e., those actively employed, aged 18 years or above,
and having commuted on the day of the survey) constituted our main sample pool.
Close scrutiny of the data revealed some cases with wrongly coded commute records
(e.g., inconsistency in reported trip origin or destination) and other cases where
individuals travelled only to non-workplaces. We also removed cases of possible
incorrect geocodes and cases involving mixed use of public transportation and private
3
The 2004 and 2011 bus-network maps were the closest versions for 2001 and 2011, respectively, that
could be found at the time of this research.
32
vehicles, both of which are rare in Hong Kong. In addition, we focused on cases
where trip-making started and ended at home, which represents the most common
scenario. We ultimately retained 20,352 cases from 2002 and 27,036 cases from 2011.
Because a comprehensive data set of walk trips from 2011 was not available, we
omitted walk trips in the analysis, which may in part have reduced the observed
proportion of non-commute travel in the sample.
Modeling commute duration:
The following equation was used to specify the model for commute duration:
(1)
where C is commute duration, A is job accessibility, T is a dummy variable that
represents residing in the new towns, Xp, Xh, and Xn are personal, household, and
neighborhood characteristics, respectively, and is the error term.
Modeling non-commute travel duration:
The structural component of SEM can be expressed in the following form:
(2)
Where: is the vector of endogenous variables; is the vector of exogenous
variables; is the matrix of coefficients of the endogenous variables; is the matrix
of coefficients of exogenous variables; and is the matrix of residuals of the
endogenous variables.
In the current study, the relationships between non-commute travel durations
and other independent variables can be expressed as three separate equations
including Equation (1), the other two are
(3)
(4)
where NC is the non-commute travel duration and and are the error terms. The
rest of the symbols have the same meanings as in Equation 1.
Estimation of job accessibility:
33
We used Shen’s (1998) gravity-based model to measure job accessibility at the SB
level in 2001 and 2011. Note that information on jobs and the working population was
not available for 2002, which is why we estimated job accessibility in 2001 to obtain
an approximate estimation of the situation in 2002. The equations used to calculate
job accessibility by public transportation and private vehicles are as follows:
(5)
and
(6)
where
and
are job accessibility by, respectively, public transportation or
private vehicles in a given SB, with i as commute origin; is the number of jobs in
SB, with j as commute destination; is the number of workers living in any SB, with
k representing job seekers; is the proportion of workers who live in SB k and
commute by private vehicles; and
and
are the impedance functions
based on the travel costs of public transportation and private vehicles, respectively,
for commuting between SBs i and j (and between SB k and j for
and
).
For Hong Kong with N SBs in 2001 and 2011, i, j, k = 1, 2, …, N.
The impedance functions are specified as an exponential form with the base of natural
logarithms, that is,
and
. β is specified as the inverse of the
average journey time
4
by public transportation or private vehicles in Hong Kong, and
and
are the commute times of the two modes, which are estimated as the
shortest network-based travel time using the OD matrix module in ArcGIS.
4
According to the results of the travel characteristics survey (Transport Department, 2014), the average
travel times by public transportation and car were 43 and 24 minutes, respectively, in 2002, and 43 and
26 minutes, respectively, in 2011. These data were used to calculate job accessibility in 2001 and 2011.
34
Appendix 1: Commuting patterns by public transportation in Hong Kong (2011)
Home location
Workplace
(% of commuters)
Average
commute
duration (min)
% of commuters
who took non-
commute trips
Average
non-commute
travel duration
(min)
Independency
Index
SNT
ONT
HKI
KLN
ROS
New town
(N=11,846)
Tsuen Wan (N=2,964)
28.27
9.82
22.57
36.07
3.27
96.51
4.42
36.42
0.21
Shatin (N=2,877)
18.70
16.61
20.44
41.15
3.09
106.42
3.55
30.73
0.17
Tuen Mun (N=1,272)
5.58
36.32
18.87
34.75
4.48
127.04
3.46
38.19
0.05
Tai Po (N=821)
7.55
27.89
17.42
42.75
4.38
121.18
3.65
29.29
0.06
Fanling/Sheung Shui (N=830)
7.83
35.54
15.78
34.70
6.14
128.32
4.94
30.89
0.06
Yuen Long (N=402)
8.21
31.09
18.91
34.58
7.21
115.41
3.23
36.19
0.05
Tseung Kwan O (N=1,726)
5.45
9.91
33.89
48.38
2.38
90.49
2.49
24.33
0.05
Tin Shui Wai (N=613)
0.49
40.46
18.60
33.77
6.69
131.06
4.89
37.37
0.00
Tung Chung (N=341)
5.87
16.72
19.06
28.74
29.62
105.40
2.05
27.30
0.05
Summary (N=11,846)
14.55
19.89
22.04
38.94
4.58
107.94
3.72
33.02
0.12
Urban area
(N=12,567)
HK Island (N=4,943)
n.a.
6.37
69.76
22.64
1.23
84.12
4.75
30.75
0.53
Kowloon (N=7,624)
n.a.
16.75
30.25
50.77
2.23
86.27
4.45
35.64
0.40
Summary (N=12,567)
n.a.
12.67
45.79
39.71
1.84
85.42
4.57
33.64
0.45
Rural area
(N=685)
n.a.
47.88
10.66
30.66
10.80
102.97
1.61
30.02
0.05
Total
(N=25,098)
6.87
17.04
33.62
39.10
3.37
96.53
4.09
33.33
0.29
Note: SNT (same new town); ONT (other new town); HKI (Hong Kong Island); KLN (Kowloon); ROS (rural and open space).
35
Appendix 2: Commuting patterns by private vehicles in Hong Kong (2011)
Home location
Workplace
(% of commuters)
Average
commute
duration (min)
% of commuters
who took non-
commute trips
Average
non-commute
travel duration
(min)
Independency
Index
SNT
ONT
HKI
KLN
ROS
New town
(N=925)
Tsuen Wan (N=164)
29.27
17.07
6.71
39.02
7.93
64.70
17.07
36.39
0.17
Shatin (N=228)
16.23
18.86
14.47
42.98
7.46
72.04
14.04
32.81
0.13
Tuen Mun (N=139)
22.30
28.06
12.23
26.62
10.79
79.39
8.63
48.42
0.18
Tai Po (N=137)
23.36
32.12
9.49
29.20
5.84
73.15
10.22
24.43
0.20
Fanling/Sheung Shui (N=68)
10.29
38.24
4.41
22.06
25.00
83.09
7.35
38.40
0.07
Yuen Long (N=54)
9.26
48.15
7.41
11.11
24.07
64.43
7.41
40.50
0.05
Tseung Kwan O (N=97)
7.22
20.62
13.40
54.64
4.12
75.12
10.31
29.00
0.06
Tin Shui Wai (N=31)
0.00
64.52
9.68
6.45
19.35
74.39
9.68
56.33
0.00
Tung Chung (N=7)
0.00
14.29
28.57
14.29
42.86
65.71
0.00
0.00
0.00
Summary (N=925)
18.05
26.70
10.70
34.16
10.38
72.73
11.68
35.23
0.13
Urban area
(N=716)
HK Island (N=342)
n.a.
13.74
61.99
23.10
1.17
63.56
13.74
32.15
0.62
Kowloon (N=374)
n.a.
27.01
16.31
50.53
6.15
61.26
14.97
35.51
0.29
Summary (N=716)
n.a.
20.67
38.13
37.43
3.77
62.36
14.39
33.98
0.40
Rural area
(N=297)
n.a.
41.08
18.18
27.95
12.79
79.25
7.07
43.36
0.10
Total
(N=1,938)
8.62
26.68
21.98
34.42
8.31
69.90
11.97
35.41
0.23
Note: SNT (same new town); ONT (other new town); HKI (Hong Kong Island); KLN (Kowloon); ROS (rural and open space).
36
Appendix 3: Standardized job accessibility in 2011
37
Appendix 4: Comparison of job numbers and commuters between 2001 and 2011
Location
Job numbers
Commuters
Public transportation
Private vehicle
2001
2011
Change
(%)
2001
2011
Change
(%)
2001
2011
Change
(%)
New town
Tsuen Wan
253,622
266,613
12,991
(5.12%)
326,810
351,373
24,563
(7.52%)
26,412
20,704
-5,708
(-21.61%)
Shatin
105,153
118,250
13,097
(12.46%)
245,269
253,424
8,155
(3.32%)
24,896
23,726
-1,170
(-4.70%)
Tuen Mun
61,305
65,221
3,916
(6.39%)
195,465
203,935
8,470
(4.33%)
15,427
21,624
6,197
(40.17%)
Tai Po
41,487
51,022
9,535
(22.98%)
116,541
112,983
-3,558
(-3.05%)
11,127
15,616
4,489
(40.34%)
Fanling/Sheung Shui
30,155
31,650
1,495
(4.96%)
95,528
102,452
6,924
(7.25%)
6,072
7,014
942
(15.51%)
Yuen Long
32,434
38,535
6,101
(18.81%)
54,192
65,699
11,507
(21.23%)
4,133
6,122
1,989
(48.12%)
Tseung Kwan O
18,547
36,641
18,094
(97.56%)
121,403
177,191
55,788
(45.95%)
6,580
10,565
3,985
(60.56%)
Tin Shui Wai
8349
15,245
6,896
(82.60%)
75,717
124,096
48,379
(63.89%)
3,930
6,235
2,305
(58.65%)
Tung Chung
2,093
6,711
4,618
(220.64%)
9,341
34,366
25,025
(267.90%)
1,711
1,440
-2,71
(-15.84%)
Summary
553,145
629,888
76,743
(13.87%)
1,240,266
1,425,519
185,253
(14.94%)
100,288
113,046
12,758
(12.72%)
Urban area
HK Island
736,719
819,159
82,440
(11.19%)
575,563
524,481
-51,082
(-8.88%)
56,615
63,192
6,577
(11.62%)
Kowloon
863,474
999,431
135,957
(15.75%)
844,919
873,901
28,982
(3.43%)
45,210
49,175
3,965
(8.77%)
Summary
1,600,193
1,818,590
218,397
(13.65%)
1,420,482
1,398,382
-22,100
(1.56%)
101,825
112,367
10,542
(10.35%)
Rural area
72,604
91,199
18,595
155,912
182,014
26,102
33,914
44,113
10,199
38
(25.61%)
(16.74%)
(30.07%)
Data source: 2001 and 2011 Censuses
39
Appendix 5: Descriptive statistics of the samples
Characteristics
Year
2002
(N = 20,352)
2011
(N = 27,036)
Average (or %)
Average (or %)
Personal characteristics
Male
56.95%
55.25%
Female
43.05%
44.75%
Average age
37.43
40.33
Married
56.47%
55.81%
Single
43.53%
44.19%
Household characteristics
Living in public housing estates
52.08%
47.26%
Living in private housing estates
47.92%
53.74%
Average household size
3.51
3.32
Having dependent children at home
31.97%
26.53%
Having retired elderly at home
15.12%
16.44%
Having a domestic helper at home
9.26%
10.82%
Having at least one vehicle in the
household
15.97%
15.14%
Neighborhood characteristics
Urban area
48.77%
49.13%
New town
47.01%
47.24%
Rural area
4.22%
3.63%
Average population density ξ
51,574
53,355
Note:
Variables of household income were not included because more than 10% of the 2002
sample did not report their household income levels. However, we found a stronger
correlation between household income and housing type. Furthermore, the modeling
results that include the 2011 TCS data indicate that the inclusion of household income
did not contribute substantially to the explanatory power of the models. Hence, we did
not include the income variable in our final models. The average population density
(ξ) is calculated at the street block level.
40
Appendix 6: Average commute and non-commute travel duration in 2002 and 2011
for public-transportation commuters
41
Appendix 7: Modeling results for commute duration for private-vehicle commuters
*** p<0.01; ** p<0.05; * p<0.1
Variable
Coefficient estimates
2002
(N = 1,784)
2011
(N = 1,938)
Model 1
Model 2
Model 1
Model 2
Age
-0.068**
-0.069**
-0.008
-0.007
Female
-0.021
-0.024
-0.005
-0.008
Married
0.026
0.022
0.019
0.015
Public housing estates
0.004
0.005
0.068***
0.068***
Household size
-0.031
-0.030
-0.052*
-0.046*
Having dependent children
-0.003
-0.008
0.075***
0.070**
Having retired elderly
0.022
0.022
0.051**
0.049**
Having a domestic helper
-0.059**
-0.062**
-0.011
-0.012
Standardized job
accessibility
-0.217***
-0.244***
-0.126***
-0.081
Population density (natural
log transformed)
0.050
0.076**
0.058*
0.071*
Kowloon
0.022
0.024
-0.009
-0.022
Rural areas
0.017
0.017
0.150***
0.187***
New towns
0.047
--
0.079**
--
Tsuen Wan
--
0.001
--
0.001
Shatin
--
0.072**
--
0.078***
Tuen Mun
--
0.013
--
0.095**
Tai Po
--
0.065*
--
0.069**
Fanling/Sheung Shui
--
0.020
--
0.073**
Yuen Long
--
-0.058*
--
-0.007
Tseung Kwan O
--
-0.011
--
0.053**
Tin Shui Wai
--
0.001
--
0.001
Tung Chung
--
-0.050**
--
-0.007
Model summary
F
8.95***
6.83***
8.96***
6.44***
Adjusted R2
0.0548
0.0642
0.0507
0.0557
42
Appendix 8: Modeling results of commute duration for public-transportation commuters by main industries
2002
2011
Industry
Number of
observations
Adjusted
R2
Coefficient
of new
towns
Coefficient of
standardized
job
accessibility
Number of
observations
Adjusted
R2
Coefficient
of new
towns
Coefficient of
standardized
job
accessibility
Finance, insurance,
real estate, and
business services
3,250
0.2317
0.0683***
-0.409***
4,120
0.2489
0.131***
-0.382***
Manufacturing
2,177
0.1248
0.146***
-0.263***
845
0.1789
0.083**
-0.358***
Transportation,
storage, and
communications
1,833
0.1071
0.041
-0.316***
3,066
0.1066
0.116***
-0.245***
Wholesale, retail,
import/export trades,
accommodation, and
food services
4,990
0.1358
0.130***
-0.278***
7,606
0.1461
0.107***
-0.306***
Community, social,
and personal services
3,401
0.0944
0.0868***
-0.250***
6,698
0.1181
0.111***
-0.248***
*** p<0.01; ** p<0.05; * p<0.1
43
Appendix 9: Modeling results for commute duration for private-vehicle commuters of different industries
2002
2011
Number of
observations
Adjusted
R2
Coefficient
of new
towns
Coefficient of
standardized
job
accessibility
Number of
observations
Adjusted
R2
Coefficient
of new
towns
Coefficient of
standardized
job
accessibility
Finance, insurance,
real estate, and
business services
173
0.2378
-0.045
-0.517***
299
0.2109
-0.006
-0.392***
Manufacturing
233
0.0335
0.143*
-0.161**
80
0.1115
0.300**
0.032
Transportation,
storage and
communications
185
0.0270
0.007
-0.254**
245
0.0789
-0.029
-0.218**
Wholesale, retail,
import/export trades,
accommodation, and
food services
335
0.0689
-0.001
-0.263***
494
0.0721
0.059
-0.187***
Community, social,
and personal services
492
0.0409
0.001
-0.075
525
0.0079
0.053
-0.082
*** p<0.01; ** p<0.05; * p<0.1
44
Appendix 10: Spatial distribution of jobs and working population in the finance
industry
45
Appendix 11: Modeling results for non-commute travel duration for private-vehicle
commuters
Note:
*** p < 0.01; ** p < 0.05; * p < 0.1.
Numbers in parentheses refer to the percentage of variance explained.
Model fit for Model (a): N = 367, df = 18, CFI = 0.963, RMSEA = 0.056, χ2 = 38.864, p<0.001, χ2/df
= 2.159.
Model fit for Model (b): N = 232, df = 18, CFI = 0.934, RMSEA = 0.076, χ2 = 42.146, p<0.001, χ2/df
= 2.341.
46
Appendix 12: Modeling results of discretionary trip duration for public-transportation
commuters
Note:
*** p < 0.01; ** p < 0.05; * p < 0.1.
Numbers in parentheses refer to the percentage of variance explained.
Model fit for Model (a): N = 917, df = 18, CFI = 0.962, RMSEA = 0.058, χ2 = 74.173, p<0.001, χ2/df
= 4.121.
Model fit for Model (b): N = 643, df = 18, CFI = 0.985, RMSEA = 0.040, χ2 = 36.201, p<0.001, χ2/df
= 2.011.
47
Appendix 13: Modeling results of maintenance trip duration for public-transportation
commuters
Note:
*** p < 0.01; ** p < 0.05; * p < 0.1.
Numbers in parentheses refer to the percentage of variance explained.
Model fit for Model (a): N = 435, df = 18, CFI = 0.985, RMSEA = 0.037, χ2 = 28.806, p<0.001, χ2/df
= 1.600.
Model fit for Model (b): N = 421, df = 18, CFI = 0.990, RMSEA = 0.032, χ2 = 25.665, p<0.001, χ2/df
= 1.426.
48
Appendix 14: Modeling results for discretionary trip duration for private-vehicle
commuters
Note:
*** p < 0.01; ** p < 0.05; * p < 0.1.
Numbers in parentheses refer to the percentage of variance explained.
Model fit for Model (a): N = 250, df = 18, CFI = 0.983, RMSEA = 0.039, χ2 = 24.923, p<0.001, χ2/df
= 1.385.
Model fit for Model (b): N = 136, df = 18, CFI = 0.928, RMSEA = 0.073, χ2 = 31.082, p<0.001, χ2/df
= 1.727.
49
Appendix 15: Modeling results for maintenance trip duration for private-vehicle
commuters
Note:
*** p < 0.01; ** p < 0.05; * p < 0.1.
Numbers in parentheses refer to the percentage of variance explained.
Model fit for Model (a): N = 141, df = 18, CFI = 0.940, RMSEA = 0.069, χ2 = 30.106, p<0.001, χ2/df
= 1.673.
Model fit for Model (b): N = 112, df = 18, CFI = 0.844, RMSEA = 0.122, χ2 = 47.663, p<0.001, χ2/df
= 2.648.