Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Jin-Seok Hahn Ph.D., Seung-Young Kho Ph.D., Keechoo Choi Ph.D. & Dong-
Kyu Kim Ph.D.
To cite this article: Jin-Seok Hahn Ph.D., Seung-Young Kho Ph.D., Keechoo Choi Ph.D. & Dong-
Kyu Kim Ph.D. (2017): Sustainability Evaluation of Rapid Routes for Buses with a Network DEA
Model, International Journal of Sustainable Transportation, DOI: 10.1080/15568318.2017.1302022
Article views: 21
Download by: [The UC San Diego Library] Date: 14 March 2017, At: 18:59
ACCEPTED MANUSCRIPT
Sustainability Evaluation of Rapid Routes for Buses with a Network DEA Model
University
1
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
University
Abstract
This study establishes a network data envelopment analysis (DEA) model to evaluate the
sustainability of public transportation services targeting rapid routes for buses in the Seoul
from the both operators’ and users’ perspectives, this model produces results that reflect the
interaction of three sustainable transport service properties, i.e., efficiency, equity, and
environmental impacts. It is identified that the expansion of median bus lanes and the conversion
of conventional buses into compressed natural gas vehicles could improve the sustainability of
the public transportation services in the Seoul metropolitan area. Some limitations and future
Keywords
sustainability evaluation, rapid routes for buses, network data envelopment analysis (DEA)
2
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
1. INTRODUCTION
Since the concept of sustainable (or green) transportation has become a new paradigm that offers
efficient, equitable, and pro-environment transport services, it is expected that any sustainable
transport system should provide mobility and accessibility to all users in a safe and
environmentally-friendly mode of transport. With that in mind, many projects and/or initiatives
have been developed and implemented around the world with a reasonable set of sustainability
indicators. While there have been many studies related to the evaluation of the sustainability of
transportation services, most of them have focused on reviewing and arranging suitable indices
for their concerns. As mentioned by Cornel and Mirela (2008), the indicators used in traditional
models measured changes in one part as if they were entirely independent of changes in the other
parts. Sustainability indicators should reflect the interaction of three key factors of sustainable
transport service, i.e., efficiency, equity, and the environment (López and Monzón, 2010).
In the transportation sector, as asserted by Stag and Gifford (2005) in their article entitled
“Sustainable transportation and quality of life,” these kinds of interactions are very important,
and, especially, any evaluation of sustainability in the transportation sector also should consider
both the operators and the users. Therefore, an expanded scope of analysis is needed to switch
from external aspects, such as the transportation system, to internal aspects, such as the
transportation service.
(1) Evaluation of the sustainability of rapid routes for buses in the Seoul metropolitan area;
(2) Development of an approach to estimate the sustainability of rapid routes for buses using
3
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
This approach includes the provision of transport services and their usage because it
considers the detailed aspects of both operators and users. While evaluating the performance in
the transportation sector, it is worth noting that a transportation service cannot be stored, so the
output consumed (the final output), such as total riders, may vary considerably from the output
produced (the intermediate output), such as vehicle-km. However, as mentioned by Färe and
Grosskopf (2000), the traditional DEA model cannot consider the intermediate output. They also
showed the differences in the structures of the traditional DEA model and the network DEA
model. This is why we chose to use a network DEA model in this study. The research process of
2. LITERATURE REVIEW
Many studies have been conducted to evaluate the sustainability of the transportation sector. In
2007, Yu and Wen (2010) used the DEA model to evaluate the sustainability of the transportation
systems in 46 Chinese cities. They also used the Malmquist Productivity Index (MPI) to evaluate
changes in sustainability between 2006 and 2007, and the influential factors were found to be
GDP per capita, the scale of the cities, and the structure of industry.
Jeon et al. (2010) evaluated the sustainability of the Atlanta transportation system using
the multiple-criteria decision-making (MCDM) process. Yoshino et al. (2010) proposed the
environmental efficiency (EE) model, which expands the concept of the data envelopment
Shen et al. (2011) evaluated the productivity of road transportation in the 27 member
states of the European Union using the DEA model. They also measured the change in
4
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
productivity from 1995 to 2007 using the MPI. Li et al. (2013) evaluated the performance of bus
routes within a public transportation system using a revised DEA method and sensitivity analyses
of indexes.
Li et al. (2016a) assessed the efficiencies of seven bus companies in Nanjing City based
on 2009 and 2010 datasets using an enhanced DEA model with constrained fuzzy-Analytic
Hierarchy Process(AHP) cones. Li et al. (2016b) evaluated the efficiency of integrated transport
in China’s provinces over the period 1988-2009 based on the DEA model.
Fielding et al. (1985) proposed the performance evaluation for bus transit framework
which is composed of input, output, and consumption of three elements. These elements can be
combined into three generalized efficiency indicators, i.e., cost efficiency, cost effectiveness, and
service effectiveness. Hendrickson et al. (2006) evaluated the performance of U.S. transportation
service sectors using 1992 and 1997 benchmark input-output models. Potter (2007) used a
transportation system by estimating the CO2 emissions from vehicles used for personal
Crotte et al. (2009) used time series and panel co-integration techniques to estimate the
effects that fares, income, quality of service, and fuel prices have on the demand for the Mexico
City metro. Shiau and Jhang (2010) extended the framework proposed by Fielding et al. (1985)
to five generalized efficiency indicators to cope with transport sustainability measurement, i.e.,
cost efficiency, cost effectiveness, service effectiveness, service reduction, and service impact.
In addition to these earlier studies, many studies have considered the indices of
sustainable transportation (Kang and Kim, 2000; Hong, 2002; Lee, 2003; Hwang and Song,
5
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
2004; Kim, 2005; Kyeon, 2005). Kao (2014) reviewed some studies of the network DEA model
by examining the network DEA models used and the structures of the network system of the
problem being studied. Castillo and Pitfield (2010) suggested a framework for identifying and
selecting a small subset of sustainable transport indicators, ELASTIC. They demonstrated the
Tang and Lo (2008) reviewed the factors that contributed to the changes in Hong Kong’s
transportation policy and their successful implementation, and they emphasized the importance
of taking into account ongoing sustainability in transport policy. Their study was related to the
suggestion of a political plan for developing a sustainable transportation system, and many
studies have addressed the progress that has been made on this subject (Barker, 2005; Hidalgo
and Graftieaux, 2008; Boschmann and Kwan, 2008; Loo and Chow, 2008; Shay and Khattak,
2010; Wachs, 2010; Debnath et al., 2011; Buehler and Pucher, 2011; Dash and Balachandra,
2016; Li et al., 2016c; Piña and Martínez, 2016; Umer et al., 2016).
satisfaction, and behavioral intentions for bus passengers in Taiwan through a hierarchical linear
modeling technique. De Oña et al. (2016) proposed a structural equation model for investigating
on the relationship among some aspects influencing passengers’ behavioral intentions towards
the use of transit services. Many other studies also focus on investigation of quality in bus
service (de Oña and de Oña, 2015; Racehorse et al., 2015; Kurtulmuşoğlu et al., 2016; Wu et al.,
2016).
As mentioned above, most of the studies have focused on selecting evaluation indices
for urban transportation systems. However, Steg and Gifford (2005) proposed a different
6
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Kim (2008) and Jeon et al. (2010) conducted studies to consider the three aspects of
efficiency, equity, and the environment when evaluating sustainability. However, it was difficult
to ensure appropriate temporal and spatial transferability with the results of their studies, because
the former was based on expert surveys, and the latter was based on subjective weight values.
To overcome these limitations and to decrease the variability of the results, an objective
and robust statistical model is needed that can consider a variety of properties of a sustainable
transport service. Among them, we used three sustainability indicators, i.e., efficiency, equity,
and the environment, proposed by López and Monzón (2010) to evaluate the rapid routes for
buses from the perspectives of both operators and users. These indicators can evaluate the three
efficiency indicators of transport sustainability such as cost efficiency, service effectiveness, and
service impact among five efficiency indicators proposed by Shiau and Jhang (2010).
From the model side, only three studies considered the DEA model for the evaluation of
sustainability (Yu and Wen, 2010; Yoshino et al., 2010; Shen et al., 2011). These studies
evaluated both efficiency and environmental issues or only undesirable issues in their
sustainability analysis; equity issues were not addressed in these studies. Thus, further
consideration of the equity aspect is important to make the entire evaluation procedure more
7
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
3. THE MODEL
In this section, we describe the statistical data used to develop the network DEA model. A
multiple regression analysis was used to identify the relationship between inputs and outputs.
Based on the results of the analysis, the proposed network DEA model was formulated.
3.1 Data
The city of Seoul has an area of 605 km 2 and a population of more than 10 million. Five
hundred and ninety-three bus routes are being operated and bus transport constitutes more than
27% of all journeys. Among them, this study evaluated the sustainability of 110 rapid routes for
We classified outputs into operator outputs (intermediate outputs) and user outputs (final
outputs) based on the findings of Fielding et al. (1985). We also matched the final outputs with
three sustainable transport service properties based on the findings of Hahn et al. (2011a, 2011b).
Hahn et al. (2011a) constructed a network DEA model considering both desirable and
undesirable outputs. Also, Hahn et al. (2011b) performed an equity analysis of feeder bus routes
The final desirable outputs, i.e., total riders (TR) and equity estimates (EEs), reflected the
efficiency and the equity aspects of the rapid transit routes, and the final undesirable output, i.e.,
air pollution cost (APC), reflected the environmental aspect of rapid routes for buses, the
intermediate outputs, i.e., vehicle-km (VK) and travel speed (TS), the inputs, i.e., total
transportation costs (TTC), the ratio of stops of median bus lanes to all bus stops (MBL), and the
ratio of compressed natural gas vehicles to the total fleet of vehicles (CNG), and exogenous
8
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
factors, i.e., population (POP) and the number of registration vehicles (RV), were collected by
The data on bus stops were provided by the operation center called Transportation
Operation and Information Service (TOPIS), which is operated by the city of Seoul. We
computed the cost of the final undesirable output of air pollution (APC) using the emission
factors of air pollutants (CO, HC, NOX, PM, and CO2) provided by the guidelines published by
National Institute of Environment Research (NIER), which is under the umbrella of the Ministry
of Environment, 2007). We also computed the unit costs of each air pollutant, as suggested by
the guidelines and manuals published by the Korea Development Institute (Korea Development
Institute, 2008). Table 1 shows the unit costs of each air pollutant. Some other studies also have
provided the unit costs of each air pollutant (Markandya, 1998; European Commission, 1999;
Holland and Watkiss, 2000), but these unit costs for APC could not be used because they did not
We used the standard deviation suggested by Hahn et al. (2011b) to estimate the equity
m
2
EEi tij t / m 1 , (1)
j 1
where:
tij = the bus users against frequency in zone j through which route i passes
9
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
t = the average bus users against frequency in the overall zone j through which route i
passes
Table 2 shows the descriptive statistics of the data used in this study. Most previous studies that
have employed DEA model to evaluate the performance of but transit utilized labor, capital, and
fuel as inputs and the number of passengers and vehicle-km as outputs. In addition to these
Unlike the traditional DEA model, the network DEA model can consider internal or linking
activities of DMUs because the model reflects their disaggregated inputs and outputs (Tone and
Tsutsui, 2009). To identify the relationship between inputs and outputs, a multiple regression
analysis was used to modify the network DEA model, and the sustainability of the analysis target
was evaluated. The results of these relationships are summarized in Figure 2. In case of outputs
that are affected by exogenous factors (POP, RV, DV), we performed the reverse two-stage
method (Barnum et al., 2008) to eliminate these factors. Table 3 shows the results of the multiple
regression analysis. As can be observed, all of the coefficients are statistically significant at the
5% significance level, and all of the adjusted R-squares of the estimated regression models were
Based on the results provided in Table 3, the exogenous factors, DV and POP, were
correlated with the final undesirable output (EE) and the intermediate output (VK), respectively.
10
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
We used the reverse two-stage method to eliminate these factors, as expressed in Equations (2)
and (3).
6
EEadjust EE 1.065 10 DV (2)
The traditional DEA model cannot reflect the desirable input or undesirable output
because this model only allows the sustainability of DMUs to improve as input elements
decrease or output elements increase. Therefore, the data were transformed in this study, such
that MBL and CNG of the desirable inputs were increased, and the EE and APC of the final
undesirable outputs were decreased. These transformations did not influence the calculated
efficiency or inefficiency obtained from the DEA model (Seiford and Zhu, 2002). The
The proposed network DEA model is formulated in Equations (4) - (11) based on the network
DEA model suggested by Hahn et al. (2011a). This model can consider the output of efficiency
(TR), the output of equity (EE), and the output of the environment (APC) separately to estimate
the sustainability scores of each rapid routes for buses, and we can evaluate the sustainability of
the public transportation service using these scores. In addition, the sustainability scores
estimated by the proposed model can reflect internal or linking activities for the process of
producing DMUs, and this model considers the operator- and user-oriented outputs of the rapid
routes for buses sequentially, so we can estimate sustainability scores that represent the supply
11
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 3 shows the proposed network DEA model for evaluating the sustainability of the
rapid routes for buses. The proposed model differs from the model suggested by Hahn et al.
(2011a) in that it uses the air pollution costs of various air pollutants as an undesirable output and
the ratio of CNG vehicles to fleets of vehicles as an input that affects the undesirable output.
Furthermore, the correlations of inputs with outputs are different from those in Hahn et al.
Equation (4) is the objective function that minimizes the sustainable scores. The most
sustainable rapid routes for buses satisfies the condition that the value of is equal to 1.0.
Equation (5) is the input (TTC, MBLadjust, CNGadjust) constraint for intermediate (VKadjust, TS) or
final outputs (TR, EEadjust, APCadjust), and Equation (6) gives the output constraint for
intermediate outputs. Equation (7) is the intermediate output constraint for the final desirable
outputs (TR, EEadjust). These equations do not have the sustainability score k in the right-side
term. Equations (8) and (9) are output constraints for the final desirable and final undesirable
outputs (APCadjust), respectively. As shown in Equation (10), variable returns to scale are
assumed for all intermediate and final outputs. The intermediate and desirable final outputs are
assumed to exhibit strong disposability, while the undesirable outputs are assumed to exhibit
weak disposability. The traditional radial approach was used to measure the sustainability of the
rapid routes for buses. The DEA model can measure of efficiency with different characteristics,
i.e., radial and non-radial. In the radial approach, if a DMU has two inputs, the input-oriented
DEA model aims at obtaining the maximum rate of reduction with the same proportion. Seiford
and Zhu (2002) give radial measures which assume that the efficiency can be improved via
12
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Min k (4)
subject to
ji λ j
x ab b
xkiabθk , i, a, b, (5)
j 1
jm λ j
y ab a
ykm
ab
, m, a, b, (6)
j 1
jm λ j
y ab b
ykm
ab
, m, a, b, (7)
j 1
jf λ j
z ab a
zkfab , f , a, b, (8)
j 1
jf λ j
z ab a
zkfab , f , a, b, (9)
j 1
n
λcj 1, c, (10)
j 1
λcj 0, c, (11)
where:
13
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
jm = the amount of an intermediate output m of a bus rapid route j from node a to node
y ab
jf = the amount of a final output f of a bus rapid route j from node a to node b
z ab
4. RESULTS
The sustainability scores of 110 rapid routes for buses in the 2009 Seoul metropolitan area were
estimated by the proposed network DEA model, and they are shown in Table 5. The results
indicate that 36 routes were operated and managed sustainably (sustainability score of 1.0), and
the average of the sustainability scores estimated by the proposed model was approximately 0.6.
Route 720, for example, was the poorest route from the sustainability viewpoint, and a solution is
urgently required to improve sustainability. The standard deviation of the sustainability score was
0.303, indicating the existence of significant disparities in sustainability among the routes.
Table 6 shows the sustainability and efficiency scores of the analysis target, and the
equity and environmental scores were classified into seven Seoul districts and Gyeonggi
Province, which make up the Seoul Metropolitan Area. All areas were designated based on the
departure location of the bus route. District 2 in Seoul had the highest average sustainability and
environmental scores, whereas Gyeonggi Province had the highest average efficiency and equity
scores.
14
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Table 7 shows the input element rates that each route should reduce or increase to
improve its sustainability score. On average, TTC should be reduced by 44%, while MBL and
CNG should be increased by 220 and 118%, respectively. Additionally, Table 8 shows the
average rate of inputs that must either be increased or decreased to increase the sustainability of
the analysis target. These results showed that district 4 in Seoul had the highest reduction rate
according to the needs of TTC and the highest increase rate according to the needs of MBL,
whereas Gyeonggi Province had the highest rate of increase according to the needs of CNG.
We used a Tobit regression model (Tobin, 1958; Amemiya, 1973; Buckley and James, 1979;
Breen, 1996; Green, 1997) to investigate the influential determinant factors on sustainability
scores because DEA sustainability scores are censored above 1.0. The sustainability score
computed from the proposed network DEA model was used as the dependent variable, and the
Table 9 shows the result of the Tobit regression analysis using the sustainability score as
the dependent variable and the efficiency, equity, and environment scores as the independent
variables. Table 10 shows the results of the Tobit regression analysis using input factors as
independent variables. The results show a satisfactory level of statistical significance. The
influences of efficiency, equity, and environmental scores on the sustainability of rapid routes for
Therefore, a new method of operating that acknowledges the divergence of equity and
environment indices must be developed and implemented. Additionally, among the inputs that
15
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
influenced sustainability, TTC was the most influential factor, which means reasonable cost-
saving measures are necessary for the operators of the metropolitan rapid routes for buses.
to lay the groundwork to gather details and make precise evaluations about the current
transportation status. To accomplish such groundwork, this study, focusing on 110 rapid routes
for buses in the Seoul metropolitan area, collected data and formulated a network DEA model to
evaluate the sustainability of the Seoul BRT system and its services.
The proposed model reflects the relationships between outputs by sharing common
intermediate outputs. The model also considers sequentially the operator- and passenger-oriented
outputs in order to reflect the non-storable nature of the public transportation service, which
means that the service is not stored irrespective of whether or not it is used up. The proposed
model can provide the representative criteria required to assess the sustainability of rapid routes
for buses. In summary, 33% of the total analysis target reached the level of full sustainability,
and the average sustainability score was 0.628, meaning that there is still some room for
improvement in the future. A review of the average rate of increase and decrease of inputs to
meet sustainability targets showed that the TTC could be decreased by an average of 44%,
whereas MBL and CNG were required to increase by 220 and 118%, respectively.
After conducting a Tobit regression analysis to analyze the determinants for the
sustainability of the analysis target, the most significant determining factor on the sustainability
of DMUs was the environmental index, whereas the influence of the TTC was found to be the
16
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
highest of the inputs. Based on these results, an effective sustainability policy plan for the
poorest analysis target should be prepared to make the entire system even more sustainable. In
addition, a new operation method that considers equity and the environment is required,
diverging from the previous operation method, which emphasizes only efficiency. We also have
shown that, to enhance sustainability, reasonable cost-saving measures for the TTC must be
This study differs from previous studies in that it established a network DEA model to evaluate
the sustainability of the public transportation service, targeting rapid routes for buses. The
authors suggested an index to judge the sustainability of the analysis target by using the model.
With additional consideration of various inputs and outputs based on this empirical model, we
can take a more multifaceted approach to future policy planning for attaining a more sustainable
The proposed model and results can be improved through further research. This study
selected variables based on the findings of Fielding et al. (1985). However, the result of
sustainability scores can be biased depending on the variables that are used. To enhance the
accuracy and reliability of the result of sustainability scores, the optimal variables that can reflect
the sustainability of rapid routes for buses should be investigated. Furthermore, consideration of
the additional efficiency indicators which can evaluate the transport performance such as cost
effectiveness, service reduction will contribute to enhancing the reliability of the sustainability
scores.
17
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
The service quality is particularly one of most important factors to determine the
sustainability of bus service. To enhance the applicability of the developed model, therefore, it is
required to select appropriate factors for measuring the service quality and to include those in the
Tables 11 and 12 show the characteristics of the roads in the Seoul metropolitan area and
Gyeonggi Province, respectively. The characteristics of the roads in these two areas are different,
and road length per area, road length per capita, road length per automobile, and the number of
lanes, for example, affect the sustainability of buses. Therefore, the sustainability of buses can be
In order to expand the sustainability of rapid routes for buses based on the results of this
study, several actions can be implemented, e.g., 1) expansion of the median-exclusive bus-lane
system that guarantees a fast speed for buses, 2) establishment of a bus route network in which
arterial buses that connect suburban areas and downtown Seoul and feeder buses that link the
major subway stations or bus terminals in downtown Seoul complement each other to increase
the equity of bus users, and 3) a policy on the dissemination of CNG buses to improve the
Especially, the city of Seoul has a good level of bus transit service but moderate equity
imbalances have been detected. In order to enhance the sustainability of rapid routes for buses, in
some populous districts of the periphery could be adequate to check the current level of bus
service and adapt it to their actual level of need. This finding was confirmed by Jun et al. (2015).
They suggested that rapid routes for buses in the Seoul metropolitan area contributed to increase
18
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
in the probability of choosing transit over automobiles and the accessibility of rapid routes for
energy consumption from cars and increasing the use of walking and bicycles, for example, are
ACKNOWLEDGMENTS
This research was supported by the National Research Foundation of Korea (NRF) grant funded
by the Korean government (MSIP) (NRF-2010-0029446). The research was also funded by
Korea Environment Institute project (GP2015-01-02-02). The opinions expressed in the paper,
however, are solely those of the authors and do not necessarily reflect the opinions of the
respective agencies.
19
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
REFERENCES
Amemiya, T., 1973. Regression Analysis When the Dependent Variable Is Truncated Normal.
Econometrica 41:997-1016.
Barker, W.G., 2005. Can a Sustainable Transportation System Be Developed for San Antonio,
Barnum, D. T., Tadon, S., and McNeil, S. 2008. Comparing the Performance of Bus Routes after
Boschmann, E.E. and Kwan, M.P. 2008. Toward Socially Sustainable Urban Transportation:
Breen, R., 1996. Regression models; censored, sample selected, or truncated data. SAGE
Publications.
Buckley, J., and James, I., 1979. Linear regression with censored data. Biometrika 66:429-436.
Buehler, R. and Pucher, J., 2011. Sustainable Transport in Freiburg: Lessons from Germany’s
Castillo, H. and Pitfield, D. E., 2010. ELASTIC – A methodological framework for identifying
188.
Chen, H. K., 2016. Structural interrelationships of group service quality, customer satisfaction,
20
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Transportation 10(5):418-429.
Cornel, L., and L. Mirela, 2008. The Quantification of the Sustainable Development at Local
Crotte, A., Graham, D.J., and Noland, R.B., 2011. The Role of Metro Fares, Income, Metro
Quality of Service and Fuel Prices for Sustainable Transportation in Mexico City.
Dash, N., and Balachandra, P., 2016. Benchmarking urban sustainable efficiency: a case of
Debnath, A.K., Haque, M.M., Chin, H.C, and Yuen, B., 2011. Sustainable Urban Transport:
Board 2243:38-45.
De Oña, R. and de Oña, J., 2015. Analysis of transit quality of service through segmentation and
De Oña, J., de Oña, R., Eboli, L., and Mazzulla, G., 2016.Transit passengers’ behavioural
21
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Färe, R. and Grosskopf, S., 2000. Network DEA. Socio-Economic Planning Sciences 34:35-49.
Fielding, G. J., Babitsky, T. T., and Brenner, M. E., 1985. Performance Evaluation for Bus
Hahn, J. S., Kim, H. R., and Kho, S. Y.2011a, Analysis of the Efficiency of Seoul Arterial Bus
Routes and Its Determinant Factors. KSCE Journal of Civil Engineering 15 (6):1115-
1123.
Hahn, J. S., Kim, H. R., and Kho, S. Y. 2011b, Equity Analysis of Feeder Bus Routes in Seoul
Transportation 29 (2):15-24.
Hendrickson, C., Cicas, G., and Matthews, H.S., 2006. Transportation Sector and Supply Chain
1983:151-157.
Hidalgo, D. and Graftieaux, P., 2008. Bus Rapid Transit Systems in Latin America and Asia:
2072:77-88.
Holland, M. and Watkiss, P., 2000. Beta Version E1.02a Benefits Table Database: Estimates of
22
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Technology Centre.
Hong, G. S., 2002. Strategies for Sustainable Transportation. Korea Transport Institute.
Hwang, S. K., and Song, S. A., 2004. A Study for the Sustainable Urban Transport System
Jeon, C. M., Amekudzi, A. A., and Guensler, R. L., 2010. Evaluating Plan Alternatives for
Jun, M. J., Choi, K. C., Yu, J. W., and Chung, W. H., 2015. Bus Patronage Change after
Sustainable Bus Reform: A Nested Logit Approach with the Case of Seoul. International
Kang, S. M., and Kim, M. S., 2000. Environmental Indices and Development of Index System.
Kao, C., 2014. Network Data Envelopment Analysis: A Review. European Journal of
Kim, D. H., 2008. A Study on the Sustainability Evaluation Indicators and Method on Urban
Kim, J. S., 2005. A Study on the Evaluation in the Sustainable Transportation of the Seoul
Metropolitan Area by Comparing with Foreign Large Cities. Master’s Thesis, Graduate
23
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Korea Development Institute, 2008. A Study on Standard Guidelines for Pre-feasibility Study on
Road and Railway Projects (5th Edition), Korea Development Institute, Korea.
Kurtulmuşoğlu, F. B., Pakdil, F., and Atalay, K. D., 2016. Quality improvement strategies of
Area with Reflected Dwellers’ Intention. Master’s Thesis, Graduate School of Chung-Ang
University, Korea.
Lee, S. Y., 2003. A Study for the Sustainable Urban Transport System (Phase 1): Land
Development Pattern and Transportation Network System for Energy Saving. Korea
Transport Institute.
Li, J., Chen, X., Li, X., and Guo, X., 2013. Evaluation of Public Transportation Operation Based
on Data Envelopment Analysis. Procedia – Social and Behavioral Sciences 96: 148-155
Li, X., Liu, Y., Wang, Y., and Gao, Z., 2016a. Evaluating transit operator efficiency: An
enhanced DEA model with constrained fuzzy-AHP cones. Journal of Traffic and
Li, T., Cao, X., and Yang, W., 2016b. Measuring the Efficiency of Regional Integrated Transport
Engineering 44(1):23-34.
24
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Li, L., Lei, Y., Pan, D., and Si, C., 2016c. Research on Sustainable Development of Resource-
Based Cities Based on the DEA Approach: A Case Study of Jiaozuo, China.
Loo, B.P.Y. and Chow, A.S.Y., 2008. Changing Urban Form in Hong Kong: What Are the
Transportation 2:177-193.
López, E., and Monzón, A., 2010. Integration of Sustainability Issues in Strategic Transportation
Markandya, A., 1998. Economics of Greenhouse Gas Limitations: the indirect costs and benefits
Denmark.
Marsden, G., Kimble, M., Nellthorp, J., and Kelly, C., 2010. Sustainability Assessment: The
Ministry of Land, Infrastructure and Transport, 2015. Statistical Yearbook of MOLIT 2015,
National Institute of Environment Research, 2007. Air Pollutants Emission Estimation Method,
Nicolas, J. P., Pochet, P., and Poimboeuf, H., 2003. Towards Sustainable Mobility Indicators:
25
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Piña, W.H.A. and Martínez, C.I.P., 2016. Development and Urban Sustainability: An Analysis of
Racehorse, V. J., Zhang, G., Sussman, A., Jian, A., and Parker, T. 2015. Bus rapid transit system
deployment for high quality and cost-effective transit service: a comprehensive review
Shay, E. and Khattak, A.J., 2010. Toward Sustainable Transport: Conventional and Disruptive
4:14-40.
Shen, Y., Ruan, D., Hermans, E., Brijs, T., Wets, G., and Vanhoof, K., 2011. Sustainable Road
2242:37-44.
Shiau, T. A. and Jhang, J. S., 2010. An integration model of DEA and RST for measuring
Shin, S. I., Jang, Y. M., Kim, S. G., and Kim, C. S., 2005. Urban Accessibility Index for
26
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Transportation. 23(8):31-42.
Steg, L. and R. Gifford, 2005. Sustainable Transportation and Quality of Life. Journal of
Tang, S. and Lo, H. K., 2008. The Impact of Public Transport Policy on the Viability and
Tobin, J., 1958. Estimation of Relationship for Limited Dependent Variables. Econometrica
26:24-36.
Tone, K. and M. Tsutsui, 2009. Network DEA: A Slacks-based Measure Approach. European
Umer, A., Hewage, K., Haider, H., and Sadiq, R., 2016. Sustainability assessment of roadway
Wachs, M., 2010. Transportation Policy, Poverty, and Sustainability. Journal of the
Wu, J., Yang, M., Rasouli, S., and Xu, C. C., 2016. Exploring Passenger Assessments of Bus
Yoshino, D., Fujiwara, A., and Zhang, J., 2010. Environmental Efficiency Model Based on Data
27
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Yu, Y. and Z. Wen, 2010. Evaluating China’s Urban Environmental Sustainability with Data
28
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
CO HC NOX PM CO2
29
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Standard
Production factors/DMUs characteristics Data Max. Min. Mean
deviation
TR 114 3 45 24
Final Desirable
EE 0.037 0.005 0.020 0.007
outputs
Undesirable APC 10,961 59 2,393 2,242
DV 1 0 1 0
OB 52 2 18 9
DMUs characteristics LR 95 7 44 21
FR 313 43 170 56
MBL: the ratio of stops of median bus lanes to all bus stops (%)
30
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
CNG: the ratio of compressed natural gas (CNG) vehicles to fleet of vehicles (%)
31
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Dependent Independent
Production factors Coefficients t-statistics p-value R2
variable variables
32
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Standard
Production Factors Data Maximum Minimum Mean
Deviation
33
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
DMU Score DMU Score DMU Score DMU Score DMU Score
101 0.270 149 1.000 340 1.000 571 0.538 721 0.345
102 1.000 151 0.421 341 1.000 602 0.932 730 0.476
103 1.000 152 0.402 362 0.326 603 0.276 750A 0.644
104 0.230 153 1.000 363 0.325 604 1.000 750B 1.000
105 1.000 162 0.667 401 0.455 605 0.358 751 0.318
106 1.000 163 1.000 402 0.224 606 0.279 752 0.422
107 1.000 171 1.000 406 0.265 607 0.506 753 0.212
108 0.414 172 0.625 407 0.353 640 0.241 760 0.374
109 0.210 201 1.000 408 0.358 641 0.434 9401 0.473
110A 0.714 202 0.833 410 0.426 642 1.000 9403 0.389
110B 1.000 240 1.000 420 0.280 643 0.570 9404 0.489
120 0.748 241A 1.000 440 0.323 650 0.313 9408 0.416
130 1.000 241B 1.000 441 0.329 651 1.000 9409 0.929
141 0.454 261 0.383 461 0.625 652 1.000 9503 1.000
142 0.532 262 1.000 462 0.584 661 0.263 9701 0.547
143 0.909 263 0.696 472 0.404 702A 1.000 9703 0.402
144 0.436 271 0.351 501 1.000 702B 0.670 9707 1.000
145 0.437 272 0.342 503 0.253 703 0.431 9708 1.000
146A 1.000 301 0.744 504 0.328 704 1.000 9709 1.000
146B 1.000 302 0.313 505 0.380 706 1.000 9710 0.908
34
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
147 0.269 303 1.000 506 0.240 708 1.000 9711 0.766
148 0.317 320 0.289 540 0.480 720 0.208 9714 1.000
# of
Max. 1.000 Min. 0.208 Ave. 0.628 S. D. 0.303 36
“1.0”
35
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
TABLE 6 Average Scores Classified into Seven Seoul Districts and Gyeonggi Province
36
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
37
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
147 -73 1244 138 462 -42 52 1 9401 -53 440 113
38
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Inputs (%)
Region District
TTC MBL CNG
1 -39 137 12
2 -40 122 4
3 -42 374 5
5 -61 202 29
6 -51 182 9
7 -44 242 17
39
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
* p<0.05, ** p<0.01
40
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
* p<0.05, ** p<0.01
41
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
42
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Paved
Non- Un
Region Length Sub Over
2 4 6 8 paved
10 opened
total Lanes Lanes Lanes Lanes
Lanes
43
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
44
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
45
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Eq. Eq.
Node 02 (5) Node 12 (6) Node 22
Desirable Input1 Intermediate Output2 Final Desirable Output2 Eq.
(MBLadjust) Eq. (TS) Eq. (EEadjust) (8)
(5) (7)
Eq. Eq.
Node 03 (7) Node 23 (9)
Desirable Input2 Final Undesirable Output
(CNGadjust) Eq. (APCadjust)
(5)
46
ACCEPTED MANUSCRIPT