You are on page 1of 16

Research Article

Transportation Research Record


2021, Vol. 2675(3) 17–32
Ó National Academy of Sciences:
Efficiency Based Evaluation of Public Transportation Research Board 2020
Article reuse guidelines:
Transport and Paratransit Systems with sagepub.com/journals-permissions
DOI: 10.1177/0361198120980322

a View to Integrating Transportation journals.sagepub.com/home/trr

Nandan H. Dawda1, Hardik Gajera2, Gaurang J. Joshi1,


Shriniwas S. Arkatkar1, and Sanjay Kumar M. Dave2

Abstract
Rapid urbanization, increasing population, and a booming economy have stimulated growth in Indian cites, resulting in higher
levels of dependence on private vehicles. This has raised serious issues over the sustainability of transport infrastructure in
cities. Public transportation (PT) systems can help achieve sustainable transportation. But they face competition from inter-
mediate public transport (IPT) or Paratransit. The existing PT and IPT systems, therefore, need to re-plan their operations to
complement each other and deliver a wider network of services. To achieve this, an evaluation of both PT and IPT systems at
microscopic levels is needed to facilitate their possible integration in a coordinated multimodal transportation system. The
present paper attempts to evaluate the public and paratransit systems in the city of Surat in the Indian state of Gujarat using
a data envelopment analysis (DEA) technique in DEAP software at the individual and system level. The operations, route
design, and cost efficiency of both systems are evaluated to understand how well the existing set-up caters to demand. The
arithmetic mean of operations, route design, and cost efficiencies of 26 routes of PT system were worked out to be 0.89,
0.94, and 0.69, while that of the 13 routes of the IPT system were 0.92, 0.97, and 0.88 respectively. The system investigation
shows that analytically, only 8% of IPT routes were inefficient in all the three aspects, compared with 42% of PT routes. The
performance of each route was analyzed, appropriate suggestions made, and the potential of these systems for designing an
integrated transport system highlighted.

Rapid urbanization, increases in population, and thriving policies has undergone a paradigm shift, with more
economic growth are stimulating growth in Indian cites. emphasis laid on the movement of passengers instead of
It is estimated that the urban population of India will rise vehicles (5). It is well known that public transportation
to 40% of the country’s total population by 2030 from systems can mitigate the dis-benefits associated with pri-
the current 30% (1). At present the country has approxi- vate vehicles like cars and two-wheelers and, therefore,
mately 458 cities with a population of more than 0.1 mil- contribute toward sustainability goals (6). To encourage
lion inhabitants (2). These urban areas contribute nearly and promote public transportation, various policies such
63% of India’s Gross Domestic Product (GDP); this is as the Smart Cities Mission (89 cities), the Atal Mission
likely to increase to 75% by 2030. If these areas fail to for Rejuvenation and Urban Transformation
function efficiently, they have the potential to constrain (AMRUT), the Faster Adoption and Manufacturing of
India’s economic growth (1). The rapid development of Electric Vehicles (FAME), the Green Urban Transport
Indian cities in conjunction with improvements in the Scheme (GUTS), and Move-in India (MII) have been
quality of life have increased the affordability of private introduced by the government over the last decade (7).
and IPT modes of transport. This can be seen in the form
of increased transport externalities such as traffic conges-
tion, delays, more accidents, and poor air quality. This 1
Department of Civil Engineering, Sardar Vallabhbhai National Institute of
has ultimately raised questions over the sustainability of Technology, Surat, Gujarat, India
2
the transportation infrastructure (3, 4). Department of Civil Engineering, The Maharaja Sayajirao University of
Baroda, Vadodara, Gujarat, India
Traditionally, transport policies were geared toward
balancing demand with supply. However, in response to Corresponding Author:
calls for sustainability, the outlook of transportation Shriniwas S. Arkatkar, sarkatkar@gmail.com
18 Transportation Research Record 2675(3)

Conventionally, public transport (PT) in Indian cities Capacity and Quality Service Manual (TCQSM) is the
is provided by bus-based systems such as city buses, bus widely accepted standard available for measuring the per-
rapid transit (BRT) systems, rail-based systems such as formance of transit services. A comparison of public
metro rail, suburban rail, tram systems, and privately transport with the established standard fails to provide
operated shared mobility in the form of paratransit or in-depth insights in relation to suitable policy alternatives
intermediate public transport (IPT) systems (8). Unlike to make the system more efficient (14). Against this back-
in developed countries, the evolution of the public trans- drop, the second approach (calculating relative efficiency
portation systems in India is found to be largely unplanned within routes) seems to be more effective in evaluating
and unrelated to the size and population of a given city. the public transportation system. Several methods are
This can be seen in the non-uniform distribution of transit available to measure and assess the performance of pub-
ridership. Presently, only 63 of the 458 Indian cities with lic transportation systems. These methods are designated
more than 0.1 million population have adequate formal city either as parametric or non-parametric (13). Most
bus systems. Out of these 63 cities, only 15 have a rail-based researchers (15–18) have evaluated the system by para-
or bus-based system as a mass transit system (2). The aver- metric tests. However, assumptions in relation to the
age modal share of public transportation systems in the functional form of the production or cost function are a
case of Tiers I and II together, and Tier III cities is approxi- major drawback to the parametric approach. This has
mately 33% and 4%, respectively. Looking at diversity, motivated researchers to adopt non-parametric methods.
bus modal share in these cities varies from 1% in Surat to Among these methods, various techniques like data
43% in Bangalore (8). envelopment analysis (DEA) (19, 20), Stochastic Frontier
Even after the implementation of different policies to Analysis (SFA) (21, 22), mixed DEA-SFA (23), and so
promote public transportation systems, the mode shares forth are proposed. It is well known that DEA has been
of these systems in small and medium-sized cities, that is, widely used to measure the efficiencies and effectiveness
cities with populations less than 1 million, is found to be of public transit systems, especially Bus Rapid Transit
dismally low (3). In such cities, paratransit modes form Systems (BRTS) (24).
the major mode of transportation alongside private The application of DEA for measuring the perfor-
modes. For the majority of Indian cities, the modal share mance of transit services can be categorized into four
of paratransit is observed to be nine times more than that stages: bus stop, route, corridor, and system. Researchers
of public transport which shows the user-dependence on (11, 25–30) have estimated efficiency scores of the vari-
paratransit systems (8). As a result, transportation modes ous transit system operating agencies as a whole using
face contradicting and competing objectives. To attain conventional DEA models while (31–34) evaluate the
sustainability, the bus and paratransit systems in cities, efficiency of the sub-units/routes/corridor of transit sys-
therefore, need to plan their operations such that they tems, while (35) evaluated the efficiency of the bus stop/
complement each other and deliver a wider network of terminal or interchange.
services. To achieve this, it is necessary to evaluate both DEA is a non-parametric approach and a linear pro-
public and paratransit systems at the microscopic levels gramming technique to measure relative efficiencies of a
to facilitate their possible integration into a coordinated set of peer units called decision-making units (DMUs).
multimodal transportation system. The basic assumption is that each DMU requires certain
Furthermore, to evaluate the existing condition of the inputs to produce its desirable output. A DMU’s perfor-
public transport of a city, performance measures are an mance is directly compared against the best DMU avail-
essential tool (9) for understanding the users’ perceptions able. Furthermore, inputs and outputs may have very
of the transit service, the business perspective of the oper- different units. This has become a popular and powerful
ator, the role of the transit service (9), and how efficient approach largely because of its ability to model multiple
and effective the system is (10, 11). Effectiveness implies input and output relationships. Initially, it was developed
the provision of bus services with minimum travel cost based on the original work of Farrel in 1957 and was
and time, whereas efficiency deals with the delivery of the later popularized by Charnes, Cooper, and Rhodes as
services at minimum operating cost without impeding the the CCR model in 1974. Later, Banker developed an effi-
daily travel of the people (12). Several attempts have been ciency frontier structured by both constant and decrease
made to develop performance measures for transit ser- returns to scale. It is used to empirically measure the pro-
vices based on different aspects of passengers, operating ductive efficiency of DMUs by comparing them to the
agencies, and communities. best practice of a DMU or combination of DMUs. This
Traditionally, the performance of the system can be model is called the Bankers Charnes and Cooper (BCC)
assessed by either of two methods: (a) comparison to model (36). Based on its simplicity and robustness, the
standards or well established benchmarks; or (b) calculat- present paper adopted the DEA technique for the effi-
ing relative efficiency within the routes (13). The Transit ciency measurement of the system.
Dawda et al 19

The comprehensive review related to route evaluation


of transit services reveals that parameters such as trans-
fers between routes, number of stops, service reliability,
waiting time, travel time, number of vehicles, operational
and maintenance costs, and fuel costs are mostly used as
an input parameter while vehicle-kilometer, passenger-
kilometer, transit ridership, vehicle speed, revenue, seat
availability rate, and route directness are the common
output parameters. This input-output matrix is used to
measure several efficiencies such as operational, schedule
design, service and delivery, cost, route design, and com-
fort and safety, and so forth (37). The present study,
therefore, uses DEA analysis for measuring the perfor-
mance of both public and paratransit systems at an indi-
vidual system level. In the case of paratransit, which is
thought of as an ‘‘Informal transport system’’ in develop-
ing countries, the evaluation of systems is carried out
either by assessing their physical and operational charac-
teristics, evaluating the perception of paratransit users,
or by measuring the service quality attributes of para-
transit users (38–42).
The detailed review of the performance of public and
paratransit service reveals that there is limited reporting
on the performance evaluation of IPT and city bus sys-
tems operating in mixed traffic conditions using DEA Figure 1. Flow chart of analysis.
Note: IPT = intermediate public transport; PT = public transportation
analysis. No such attempts have been made to evaluate
systems; DEA = data envelopment analysis; BCC = Bankers Charnes and
the paratransit as the whole system such as the formal Cooper model.
bus transport system. Also, these systems were evaluated
independently without discovering their potential to form
an integrated transport system. Methodology
With this motivation, the present paper attempts
to evaluate the public and paratransit system of the The overall methodology used for analyzing the perfor-
city of Surat in the state of Gujarat in India using a DEA mance of PT and IPT system is shown in Figure 1.
technique at the individual and system levels. As an
important outcome, the study suggests some policy inter-
Identification of Indicators and Model
ventions for possible integration under a coordinated
multimodal system perspective as a goal to achieve sus-
Specification
tainability. Operational efficiency, route design effi- For the present study, each PT and IPT route is treated
ciency, and the cost efficiency of both systems were as a DMU and six BCC-DEA models are developed to
evaluated simultaneously to understand how well the evaluate their route, cost, and operational efficiencies.
existing setup is catering to the present travel demand. The route efficiency gives an idea of the supply and
The performance of each route within the system was demand relationship. The cost efficiency represents the
analyzed and appropriate suggestions were made. Lastly, economic performance at the route level and provides a
the potential of these systems for designing the proposed comprehensive assessment of ridership productivity and
integrated transport system was assessed. The rest of the financial performance. Operational efficiency provides
paper is structured as follows. The next section explains information about how effective the present schedule is.
the methodology adopted for the study followed by the Table 1 (Column 1) shows the measures used for the
identification of indicators and model specifications. We identification of efficiencies of PT and IPT routes.
then illustrate the study area and data collection. Route Moreover, associated input and output parameters,
evaluation of PT and IPT systems for single and multiple along with their units, are given in Columns 2, 3, 4, and
efficiency measures using the DEA technique is then 5 of Table 1, respectively. Finally, Column 6 shows the
explained. This is followed by a comparative analysis of orientation of the model.
PT and IPT system. We conclude with findings, policy The input and output parameters used can be
interventions, and concluding remarks. explained briefly as:
20 Transportation Research Record 2675(3)

Model orientation (6)


Route Length: the total distance between the origin
and destination of the route. It was derived for all the

Output oriented
Input oriented

Input oriented
DMUs.
Number of stops: this parameter can be used to show
the total number of places from where passengers can
enter or exit the system. The number of stops on a
route plays a significant role in the performance of a
IPT model

route as at every stop the vehicle stops, causing delay

passenger); 1/route directness


and increased travel time.

(travel time by private mode/


travel time by transit or IPT)
Output parameters (5)

Number of vehicles: represents the total number of


Ridership (passengers/day);

Seat availability rate (seats/


Revenue generated (INR)

transit units/auto-rickshaws (IPT units) serving on a


vehicle-km (km/day)

particular route. For the PT system, the data were


taken from the operator while for the IPT system, the
total IPT units on each route were determined by
extracting 16 h of videography survey carried out at
designated IPT stops on the route.
Ridership: defined as the total number of passengers
using the service per day (passengers/day). For the PT
system, automatic passenger count (APC) data were
maintenance cost (INR)
Route length (km); no. of

obtained from the operator. In the case of the IPT


stops (numbers); no. of

No. of stops (numbers);


Input parameters (4)

system, a revealed preference survey of operators was


IPT units (numbers)

average frequency

conducted to estimate the ridership.


operational and
Fuel cost (INR);

Vehicle-km: describes the supply of transit services. It


is defined as the sum of the distance traveled by all
vehicles running on a particular route. For both
(TU/h)
City bus model

modes, it can be obtained by the product of multiply-


ing numbers of transit units/auto-rickshaw by the
numbers of trips in total route length.
Operational and maintenance cost (O&M): includes
Revenue generated (INR)
Output parameters (3)

(travel time by private


day); vehicle-km (km/

fuel cost, operator’s salary (in the case of PT systems),


Ridership (passengers/

mode/travel time by
1/route directness
Seat availability rate
(seats/passenger);

and the maintenance cost. In the case of a PT system,


transit or IPT)

the cost of the salaries paid to the drivers and ticket


collectors employed for each bus on a particular route
Note: INR = Indian rupees; IPT = intermediate public transport; TU = transit units.

and the total cost of fuel for all transit units operating
on a particular route were obtained from the city bus
day)

operator. For the IPT system, the stratified random


sampling technique was adopted to carry out the
revealed preference survey of IPT operators at specific
Table 1. Description of the Model for IPT and City Bus

maintenance cost (INR)


Route length (km); no. of
stops (numbers); no. of

(TU/h); peak frequency


transit units (numbers)

No. of stops (numbers);


Input parameters (2)

stops on each route to derive the maintenance and


off-peak frequency

fuel cost. The number of IPT operators to be sur-


operational and

veyed on each route was decided based on the number


Fuel cost (INR);

of IPT units running on it. Considering the possibility


of exaggeration in the responses of IPT operators,
(TU/h)

instead of asking discrete values, a specific range of


maintenance and fuel costs was proposed to them.
The proposed ranges of O&M were derived from a
pilot survey of IPT operators. Based on the survey,
Cost efficiency
measure (1)

the single value of total maintenance and fuel cost for


Route Design
Efficiency

Operational
efficiency

efficiency

the entire route was calculated using the weighted


average method.
Revenue generated: the total income generated on a
particular route. The passenger demand data and fare
structure are combined to determine the total revenue
Sr. no.

generated for the PT system while, in the case of an


1

3
Dawda et al 21

IPT system, the revenue was obtained using a


revealed preference survey of IPT operators.
Frequency: defined as the number of transit units
passing through a given point on a transit route in
one direction per hour. For the evaluation, three
types of frequency values—peak-hour frequency,
average frequency, and off-peak hour frequency—
were used as inputs.
Seat availability rate: the ratio of the total number of
seats available to the total number of passengers tra-
veling during peak hours.
Route directness: estimated as the ratio of the travel
time by city bus or IPT to the travel time for the same
route in private mode (automobile). The smaller the
ratio, the better the service. However, in the present
study the inverse of this parameter is used to assist
the DEA model. The value of ‘1/route directness’ is,
therefore, used: the greater the value, the better the
performance of the system. Figure 2. IPT and PT network of Surat city.
Note: IPT = intermediate public transport; PT = public transportation
systems.
Study Area
Surat is one of the fastest-growing cities in India. It is with the existing city bus network. The unplanned net-
located on the west coast. It has been experiencing rapid work is so well established that the IPT units serve pas-
growth in population with a growth rate of 55.29% over
sengers at predefined stops. Thus, instead of acting as a
the last 10 years (43). The average annual growth rate is
feeder network, the IPT network coincides with the pub-
4.5%. The population in 2011 was 4.5 million with a
lic transport network. Figure 2 shows the IPT and PT
density of 13,680 persons/km2. At present, the vehicular
routes of the study area.
population in Surat city is 2.4 million with an annual
For the current study, 26 operational city bus routes
growth rate of 35% (44). About 90% of all vehicles are
and 13 selected IPT routes (which overlap with the routes
private such as two-wheelers and four-wheelers. The
of PT system) were evaluated in all the three aspects of
public transport mode share is less than 1%, against the
cost, operations, and route design. The data collection
desired share of 40% to 60% (45).
method is presented below.
Surat’s present public transportation system consists
of 37 routes (28 routes of city bus services and nine
routes of the BRT system) with a fleet size of 358 buses
Data Collection
(238 minibuses running on city bus services and 120 stan-
dard buses plying the BRT corridor). The total network To quantify the inputs, the operational schedule of buses,
coverage of the city bus system was calculated by assum- automatic vehicle count data, infrastructure details, and
ing a buffer of 500 m along the routes using Arc-GIS fare matrices were collected from transit operators.
software. It was observed to be 276.13 km2 (46). The city Further, data were used to decode the input parameters:
has also had IPT service for some decades. Presently, vehicle-kilometer, frequency, number of buses on each
there are about 89,000 registered auto-rickshaws/IPT route, route length, number of stops, daily transit rider-
units with an 11% mode share (44). IPT operates in two ship, salary paid to the operators, seat availability rate,
different ways in the city: (i) fixed routes with fixed fares and fuel cost.
based on distance; and (ii) flexible routes with varying For the IPT system, a survey of operators was carried
fares based on distance as well as the time of the day. out on each route to obtain the parameters such as
The share of IPT units operating on fixed routes is more monthly income, maintenance and fuel cost, vehicle-km,
dominant compared with the flexible route services. The number of stops, and route length. Videography surveys
absence of an efficient bus service has resulted in the shift were carried at each predefined fixed stop on all the
of trips from buses to IPTs on point-to-point services. routes for one day (16 h) to obtain data including the
Around 0.77 million passenger trips are undertaken by number of auto-rickshaws/IPT units, frequency, and pas-
IPT (47). Presently, the IPT network, which has evolved sengers. To determine the route directness for both sys-
in response to demand over the years, is overlapping tems, a performance box survey was carried out on all
22 Transportation Research Record 2675(3)

the routes for both morning and evening peak hours. Xij = the ith input for DMUj,
The input and output data for all the selected routes are Yrj = the rth output for DMUj,
shown in Table 2. xio = amount of input i used by DMUo,
yro = amount of output j produced by DMUo,
lj = the nonnegative scalars (weights) for DMUj, and
Route Evaluation of IPT and PT Systems u : = the optimal output level.
using DEA BCC Model The above equations are theoretical forms of input
The DEA model can be either input oriented or output and output-oriented DEA BCC models. The DEA tech-
oriented. The input-oriented model provides information nique determines the efficiency scores of a single route by
about excess supply compared with the best frontier. The comparing its performance with the most efficient route
main aim of the input-oriented model is to minimize of the system. In this method, each route is denoted as
inputs whereas the output-oriented model determines the DMU and the term DMUo is used to represent the
potential of inefficient routes to maximize the output. DMU under evaluation. The term xio and yro are the ith
BCC model is adopted in the present study to determine input and rth output of the DMUo. The DEA model
the cost, route design, and operational efficiencies. The attempts to create the most efficient DMU for a given
reason for choosing the BCC model along with the CCR DMUo, whose productivity is dependent on the linear
model is that it makes the variable returns to scale (VRS) combination of frontier DMUs’ input and output.
assumption, which means that efficiency may increase or The objective function 1) attempts to maximize the
decrease with a change of size in input or output. The level of output for the DMU under consideration.
VRS assumption is more appropriate for transit lines Constraint 2) sets the limit of input to no more than the
with higher variability of productivity. Mathematically, observed amount of input. Constraint 3) specifies the
the BCC model can be expressed as follows: output level is at least as good as that of DMUo.
Ps Constraint 4) ensures that the DEA model’s VRS status
ur Yrj which shows the sum of all scalars is equal to unity.
Minimize uo = Pmr = 1 ð1Þ
i = 1 vm Xij
Constraint 5) describes non-negativity restriction for lj .
As per the VRS method, u = 1 is always a feasible
Subjected to constraints solution to the DEA model. If u = 1 turns out to be the
Xm optimal solution, then DMUo is efficient (as the model is
i=1
lj xij ł uxio i = 1, 2, . . . :m; ð2Þ unable to find a virtual DMU with a higher level of out-
X
s put with a constant level of current input). If the optimal
lj yrj ø yro r = 1, 2, . . . :::, s; ð3Þ solution u is greater than 1, then the DEA model has
r=1 identified a DMU that can perform better than DMUo
X26 which indicates that DMUo is inefficient. An efficiency
j=1
lj = 1, ð4Þ score (u) between 0.6 and 1 means fairly efficient. An effi-
ciency score (u) less than 0.6 indicates an inefficient sys-
lj ø 0 j = 1, 2, . . . ::, 26 ð5Þ
Ps tem. A scale to classify the efficiency score was used,
ur Yrj according to Lao and Liu (48). The ranks were obtained
Maximize uo = Pmr = 1 ð6Þ
i = 1 m Xij
v by considering efficiency scores. For routes having effi-
ciency score ‘‘unity,’’ the ranking was done based on the
Subjected to constraints peer counts (this indicates the number of times an effi-
Xm cient route acts as a reference for the inefficient routes).
i=1
lj xij ł xio i = 1, 2, . . . :m; ð7Þ Table 3 shows the efficiency scores of each route in all
Xs three aspects. To evaluate the internal performance of
r=1
lj yrj ø uyro r = 1, 2, . . . :::, s; ð8Þ the system, it becomes necessary to understand how the
X26 efficiency scores vary for each route. The evaluation of
l = 1,
j=1 j
ð9Þ each route based on a single efficiency measure for both
systems is explained below.
lj ø 0 j = 1, 2, . . . ::, 26 ð10Þ
The public transport system operates with a service
where motive while the paratransit system operates with a profit
ur = decision variable representing the weight for output r, motive. As an outcome, the public transit system runs
vm = decision variable representing the weight for input m, throughout the city to make it more accessible as well as
J = index of decision-making units, j = 1, . . ., n, available. Many of the routes would, therefore, have
I = index of input, i = 1, . . ., m, lower efficiency scores and can be termed as inefficient
R = index of output, r = 1, . . ., s, with reference to better performing routes. If the routes
Dawda et al 23

Table 2. Input and Output Data of PT and IPT System

Intermediate public transport


Inputs Outputs
* * * ** ** * * *
Route no. I-1 I-2 I-3 I-4 I-5 I-8 O-1 O-2 O-3** O-4* O-5

IPT-1 4 6 1199 0.117 0.259 75 47960 57552 1.799 0.6 0.74


IPT-2 5 8 1244 0.283 0.181 78 87080 139328 0.933 0.6 0.76
IPT-3 8 8 1435 0.263 0.224 90 86100 129150 1.375 0.6 0.68
IPT-4 3 4 1412 0.115 0.278 88 84720 56480 1.747 0.5 0.72
IPT-5 2 3 1981 0.081 0.534 124 63392 39620 3.677 0.75 0.67
IPT-6 4 6 1907 0.279 0.297 119 114420 137304 1.692 0.6 0.74
IPT-7 3 5 1607 0.118 0.386 100 38568 57852 2.32 1 0.76
IPT-8 3 4 971 0.079 0.18 61 38840 38840 1.112 0.75 0.47
IPT-9 2 2 1246 0.041 0.307 78 39872 19936 1.76 0.75 0.70
IPT-10 4 5 1500 0.152 0.297 94 45000 75000 1.568 1 0.72
IPT-11 4 4 2363 0.168 0.412 148 94520 82705 3.249 0.75 0.89
IPT-12 4 4 1368 0.148 0.157 86 95760 72778 1.177 0.6 0.59
IPT-13 3 4 2350 0.229 0.351 147 84600 112800 2.291 1 1.02

City bus transport system

Inputs Outputs
* * * * * * * * *
Route no. I-1 I-2 I-3 I-4 I-5 I-6 I-7 O-1 O-2 O-3* O-4* O-5

3 25 13.3 12 6959 11333 5 6 24933 1824 25375 1.16 1.05


102 32 7.7 6 5125 8267 3 5 9166 1334 10786 4.37 0.81
104 22 9.5 9 6378 8933 4 6 25247 1644 29697 1.87 1.17
105 28 12.3 9 6576 9733 4 6 6140 1705 7229 7.98 0.64
106 28 13.3 12 9488 10000 4 5 15071 1824 17726 1.68 0.85
107 37 14.3 14 12474 14267 5 8 62083 3249 73020 0.98 0.70
108 35 19.9 9 8943 10667 3 4 33713 2198 39659 0.89 0.66
109 21 15 9 7789 8800 4 5 28563 1986 33600 1.38 0.56
112 23 10.7 9 6385 9067 3 6 26258 1761 30891 1.83 0.47
116 33 13.7 10 8137 11067 4 5 20140 1987 23695 2.05 0.90
117 32 11 7 5409 8933 2 4 15491 1458 18220 2.08 0.88
118 29 15 9 8067 9867 3 4 27659 2095 32538 1.13 0.78
126 29 10.2 9 6913 9867 4 6 20919 1752 24609 2.29 1.00
153 18 6.5 5 3237 5733 3 4 7162 814 8431 4.57 0.69
202 26 8.2 7 4808 8133 4 6 14660 1215 17249 3.30 1.06
204 29 10.3 8 6120 9200 4 5 15130 1509 17800 2.58 1.10
205 58 14 12 11150 15733 3 6 26326 2908 30965 1.80 0.89
206 31 14 4 4036 6800 2 3 4885 1073 5747 4.18 0.78
209 27 9.9 9 7201 9600 4 6 16661 1798 19604 2.84 1.28
212 25 4.6 10 6785 10000 4 6 23160 1824 27244 2.07 1.00
254 39 14.6 9 6424 11200 3 4 14834 1647 17455 2.09 0.68
402 26 11.5 9 6009 9467 4 5 17182 1604 20213 2.36 0.84
403 19 6.9 5 3964 5867 4 5 9727 994 11445 4.13 0.41
410 42 19.9 6 4851 9600 2 3 4387 1274 5163 4.40 0.70
506 33 15 4 3008 7067 2 3 9166 745 10786 2.16 0.76
658 23 17.2 9 7127 9067 3 4 20587 1888 24214 1.54 0.70

Note: IPT = intermediate public transport; PT = public transportation systems.


*
I-1, number of stops; I-2, route length; I-3, number of auto-rickshaws/number of transit units; I-4, fuel cost; I-5, operation and maintenance cost; I-6, off-
peak frequency; I-7, peak hour frequency; I-8, average frequency; O-1, PAX/day; O-2, vehicle-km; O-3, revenue; O-4, seat availability rate; O-5, route
directness.
**
Values in millions (INR).
24 Transportation Research Record 2675(3)

Table 3. Route Design, Cost, and Operational Efficiency of PT and IPT Systems

PT models
Route design efficiency Cost efficiency Operational efficiency
Route number u1 u2 Rank u1 u2 Rank u1 u2 Rank

3 0.804 0.867 20 0.623 0.69 15 0.85 0.892 15


102 0.913 1 6 0.36 0.456 20 0.852 0.862 17
104 0.842 1 8 0.795 0.953 3 1 1 3
105 0.803 0.863 23 0.188 0.212 25 1 1 6
106 0.804 0.867 20 0.346 0.42 23 0.765 0.84 20
107 1 1 1 1 1 13 0.426 0.559 26
108 1 1 7 0.758 0.849 8 0.667 0.805 23
109 1 1 3 0.746 1 6 0.555 0.857 18
112 0.856 0.982 13 0.826 0.98 5 0.523 0.919 12
116 0.839 0.871 19 0.497 0.532 21 0.782 0.815 22
117 0.855 0.913 16 0.575 0.686 10 1 1 4
118 0.979 1 10 0.689 0.797 7 0.829 0.893 14
126 0.828 0.896 17 0.608 0.683 17 0.772 0.826 21
153 0.685 1 11 0.445 1 2 1 1 1
202 0.731 0.874 18 0.613 0.79 12 0.899 0.905 13
204 0.793 0.867 22 0.497 0.581 16 0.983 0.984 9
205 1 1 9 0.474 0.481 22 0.674 0.683 25
206 1 1 4 0.243 0.387 18 1 1 2
209 0.854 0.931 15 0.465 0.535 14 1 1 5
212 1 1 5 0.686 0.765 9 0.814 0.875 16
254 0.754 0.784 26 0.464 0.525 24 0.666 0.769 24
402 0.76 0.849 24 0.575 0.662 19 0.773 0.857 19
403 0.831 1 2 0.493 1 4 0.788 0.947 10
410 0.806 0.813 25 0.182 0.225 26 1 1 7
506 0.744 1 12 0.613 1 1 0.974 1 8
658 0.917 0.945 14 0.58 0.699 11 0.784 0.933 11

IPT models
Route design efficiency Cost efficiency Operational efficiency
Route number u1 u2 Rank u1 u2 Rank u1 u2 Rank

IPT-1 0.581 0.853 13 0.88 0.963 6 1 1 2


IPT-2 1 1 1 0.653 0.678 13 0.987 1 6
IPT-3 0.925 0.945 10 0.78 0.8 10 0.782 0.797 10
IPT-4 1 1 8 0.797 0.871 7 0.885 0.895 9
IPT-5 1 1 7 1 1 4 1 1 5
IPT-6 1 1 2 0.723 0.731 11 0.669 0.687 13
IPT-7 0.552 0.893 11 0.804 0.829 9 0.985 1 7
IPT-8 0.58 1 5 0.783 1 3 1 1 4
IPT-9 0.769 1 4 0.951 1 5 1 1 1
IPT-10 0.673 0.881 12 0.67 0.684 12 0.942 1 8
IPT-11 1 1 9 1 1 1 0.67 0.788 12
IPT-12 1 1 3 0.953 1 2 0.749 0.792 11
IPT-13 1 1 6 0.83 0.834 8 0.971 1 3

Note: IPT = intermediate public transport; PT = public transportation systems. Digits in bold represent the rank of each PT & IPT route under evaluation.

are found to be inefficient, this means that it consumes routes, numbers 112, 117, 209, and 658, had efficiency
more input compared with frontier routes having a simi- scores close to unity while the remaining 10 routes dis-
lar output. played low-efficiency scores between 0.6 and 1 which
indicates the need for some policy interventions. On the
evaluation of these routes in detail, the two main causes
Route Evaluation Based on Single Efficiency Measure of route inefficiency were identified as: (i) the number of
Route Design efficiency: For the PT system, 12 of the 26 buses plying those routes was higher compared with an
PT routes were found to be fully efficient. The four average number of buses operating on more efficient
Dawda et al 25

routes, suggesting a need to shift the buses from these routes (Route nos. 1, 2, 5, 7, 8, 9, 10, 13) should be done
routes to other profitable routes based on passenger (54–56).
demand and; (ii) the length of these routes was much In subsequent sub-sections, route evaluation based on
longer compared with route length on other profitable multiple efficiency measures is presented comprehensively.
routes. Thus, to address this issue, it is recommended
that the route length be reduced based on the passenger
demand profile of bus stops. In the case of the IPT sys-
Route Evaluation Based on Multiple Efficiency
tem, the efficiency scores of nine routes out of a total of Measure to Integrate PT and IPT System
13 were unity, and four routes had efficiency scores near The discussion above is based on the particular efficiency
unity. Therefore, no specific recommendation is sug- measure taken into consideration, that is, cost or opera-
gested for the IPT system. tional efficiency, or route design efficiency. However, to
Cost efficiency: The cost efficiency model shows that integrate PT and IPT, it becomes essential to understand
the PT system is more inconsistent as only five routes the performance of both systems with all possible scenar-
were found to be efficient. Eleven routes with an effi- ios of efficiency. In that context, therefore, the routes are
ciency score of more than 0.6 can be made efficient by summarized into eight different cases in the form of pos-
increasing the use of existing transit units. If the number sible combinations based on the efficiency measures
of trips made by a bus per day is increased, the frequency shown in Table 4. These eight combinations of efficiency
of the services will increase, and the waiting time will be measures have taken into consideration of all possible
reduced at each bus stop en route. This will result in scenarios in practice. The matrix consists of eight rows
increased revenue generation. But because 10 routes had and six columns. Columns 2, 3, and 4 of Table 4 show
a score of less than 0.6 reveals a huge imbalance between efficiency measures. Column 5 gives the findings and
the cost incurred and revenue generated. For these Column 6 details the recommendations for each scenario
routes, the allocation of the number of buses and trips of a combination of efficiency measures. The different
made by each bus needs to be analyzed so that the cost rows of Table 4 show the different cases for which a
efficiency can be enhanced. For the IPT system, five combination of efficiency measures is evaluated.
routes had a score of unity. Also, the other routes exhib- Discussing the extreme cases, 8% PT and 23% IPT
ited higher efficiency scores compared with the PT sys- routes were efficient in all aspects, while 42% PT and
tem. Specifically, routes 2, 3, 4, 6, 7, 10, and 13 which 8% IPT routes were inefficient in all aspects. This estab-
operate in the Central Business District (CBD) area had lishes that passengers are choosing IPT services as a pre-
lower efficiency scores. The IPT units of these routes dominant mode on routes which are also served by PT.
operate in a congested stream with low speed, resulting One of the strategies to make PT operations more effi-
in higher fuel costs which cause inefficiency. cient would be to review the quality of PT services from
Operational efficiency: In the case of the PT system, a rider’s perspective and propose an integrated network
seven routes were found to be fully efficient. Route 107 of PT and IPT for the entire city. Eventually, it may
has the lowest operational efficiency score of 0.559 but is result in a greater role for IPT in feeding the bus service
cost-effective (efficiency score of 1) and route design is rather than competing with the transit (68, 69).
also equally efficient (efficiency score of 1). Inefficient
operation of routes is mainly because the frequency of
Comparison of Efficiency across PT
buses in peak and off-peak periods is quite high com-
pared with passenger demand. For all 19 inefficient and IPT Systems
routes, except routes 107 and 108, supply (seats offered) Furthermore, to compare the PT and IPT systems with
was observed to be more than double the actual demand each other, the radar diagram was plotted demonstrating
(passenger demand). Based on this relationship, it is sug- route, operation, and cost efficiency scores as shown in
gested that the reduction of peak hour frequency/increase Figure 3, a and b. The value inside the bracket represents
in headway would reduce the gap between demand and the rank of the route based on all three aspects.
supply and ultimately make these routes more efficient. The radar diagram shows the values of efficiency from
(49–53) In the case of IPT, eight routes on the system 0 to 1 originating from the innermost concentric circle to
had efficiency scores of unity and five routes scored in the outermost concentric circle. Figure 3, a and b, reveals
the range of 0.6–1. It was observed that IPT units are that the IPT system performs better than the PT system
available to the users in a range of 24 s (Route no. IPT in all aspects as values closer to unity are seen more in
11) to 59 s (Route no. IPT 8) which indicates the higher the case of the IPT system. The arithmetic means of oper-
availability of the IPT units on all the inefficient routes. ation, route design, and cost efficiency scores for the PT
Therefore, it is suggested that shifting the IPT units from system are 0.89, 0.94, and 0.69 whereas, for the IPT sys-
inefficient routes (Route nos. 3, 4, 6, 11, 12) to efficient tem, they are 0.92, 0.97, and 0.88 respectively. This
26 Transportation Research Record 2675(3)

Table 4. Eight Cases of Combination of Efficiency Measure

Operational Cost Route design Findings Recommendations

Case I:  The exact tradeoff has occurred  Policy interventions like


Only operationally P O O between frequency and seat interlining of transit units
efficient availability, the number of stops, and (57), skip stops operations
route directness. of bus services (58, 59) and
 Inequity between transit units and designing of truncated
ridership, route length, and vehicle- routes (60) are suggested.
kilometer.
Case II:  The number of transit units and route  Actions like an increment in
Operationally O P P lengths are provided suitably in the the headway of buses,
inefficient context of observed ridership and timetable development
required vehicle -km. incorporating the passenger
 Excess frequency and number of stops demand variations (61), and
are witnessed with reference to transit signal priority (62)
passenger demand and route can make the system more
directness. effective.
Case III:  The cost incurred for maintaining and  Actions like route
Only cost-effective O P O operating the transit unit is closely rationalization (63),
equal to the revenue generated. reduction in allocated transit
 Need to optimize the input and units, and provision of
output parameters involved in accelerated transit services
operations and route design. between high demand
origins and destinations (58,
64) are recommended.
Case IV:  This reflects an imbalance between the  Plans like provision of night
Inefficient in P O P revenue generated and the cost haulage services (46), and
cost prospective incurred by the operator on transit revising vehicle and crew
units. scheduling (65) are advised.
 All the inputs that is, number of stops,
frequency, route length are very well
adjusted in comparison to outputs:
passenger demand, route directness,
and vehicle-km.
Case V:  Optimal relations between input-  The reduction of peak and
Effective in route O O P output parameters, that is, transit off-peak hours frequency
design only units and ridership, route length, and would increase efficiency
vehicle-kilometer is achieved. (66).
 The inappropriate tradeoff has
occurred between frequency and seat
availability, the number of stops, and
route directness.
Case VI:  Frequency and number of stops are  Maximizing the use of transit
Inefficient in route P P O optimized but the route length and units shall result in a hike in
design only number of transit units need to be efficiency (67).
reduced.
Case VII:  The routes following, in these cases  These are the best
Efficient in all aspects P P P are the best performing and they can performing routes.
be considered as an ideal route in Therefore, no modifications
relation to input allocation and output are required to be made.
generation. For any system DEA
selects efficient routes as a frontier
DMU.
Case VIII:  These are the routes that reduce the  It is recommended that all
Efficient in no aspects O O O overall efficiency of the system. The supply and demand
operators of transit services need to parameters be audited and
prepare an optimal mix of the supply modified considering the
and demand to increase efficiency. frontier DMUs’ supply-
demand characteristics.

Note: ‘‘P’’ indicates efficient and ‘‘O’’ indicates inefficient. DEA = data envelopment analysis; DMU = decision-making units.
Dawda et al 27

Figure 3. (a) Efficiency scores and ranks of routes of PT system, and (b) efficiency scores and ranks of routes of the IPT system.
Note: IPT = intermediate public transport; PT = public transportation systems.

indicates that both systems provide shared mobility ser- Major Findings
vices, while still performing various roles in addressing to
The microscopic evaluation of both systems at the route
the city’s mobility needs. The major reason for the varia-
level draws our attention to the following findings. To
tion of system efficiency is that PT services are run to
get a better idea of the performance of both systems, the
provide mobility to the maximum number of citizens in
average of all three efficiency scores of the PT and IPT
the city with no profit motive, whereas the IPT services
routes are written in brackets along with the route
operate with a commercial motive of maximizing revenue
number.
for the operations.
The PT system operates in the city at both peak and  The present study reveals that the IPT system is
off-peak hours and thus performs poorly in operational
more efficient compared with the PT system as
efficiency indicators. By contrast, the IPT systems run
most of the routes were efficient according from
on high passenger demand corridors and can vary their
both passengers’ and operators’ perspectives. As a
routes dynamically (in response to demand). Because of result, the IPT system is competitive with the PT
this added advantage, they can serve more passengers system. It is, therefore, recommended that the IPT
and generate higher revenues, giving them greater opera- network be designed in congruence with the exist-
tional efficiency. Additionally, fares on the IPT system ing PT system in such a way that paratransit aug-
were higher compared with the PT system. The average ments passenger demand in peak hours.
fare charged per kilometer by IPT mode was found to be  Two routes of PT (Route nos. 106 [0.709] and 206
Rs.2.62/km while in the case of PT mode, it was Rs.1.1/ [0.795]) and IPT (Route nos. 3 [0.84] and 12 [0.93])
km. As a result, the average ratio of revenue generated cater to passenger demand in the southwest part
to O&M cost per day estimated for both the PT and IPT of the city. It is observed that because of shorter
system was 2.15 and 14.63 respectively. This means that route lengths, fewer stops, and the higher fre-
the revenue generated compared with maintenance and quency provided by IPT system, passengers pre-
fuel cost for IPT as a whole is much higher than for the ferred to travel by IPT rather than by PT. Because
PT system. This makes IPT a cost-effective service com- of this, the total demand of PT routes is less com-
pared with the PT system. Even from the users’ perspec- pared with IPT routes plying the same stretch of
tive, the provision of a high frequency of services by the road. As a result, the efficiency scores for PT
IPT system makes it effective in relation to route design. routes were found to be less compared with IPT
The major shortcoming in the present transport sys- routes.
tem is that both systems operate independently and, as a  Five PT routes (Route nos. 102 [0.77], 112 [0.96],
result, IPT creates unhealthy competition with formal 212 [0.88], 202 [0.85], and 118 [0.89]) along with
public transport. Both PT and IPT systems are crucial in one IPT route (Route no.10) serve the northern
developing sustainable urban mobility services in Indian part of the city. The detailed assessment of these
cities. It is, therefore, necessary to understand how both routes shows that, out of 31.9 km of total route
systems cater to passenger demand. length, 20 km has more than two routes
28 Transportation Research Record 2675(3)

overlapping. As a result, these routes struggle for The Operational Integration of Paratransit and City
passenger demand and create internal competition Bus Routes in Relation to Timetable Development
for each other. This causes varying efficiency
The policy of integrating IPT with PT is necessary to
scores for all the five PT routes mentioned above.
 optimize the operations of both modes. Based on the case
Two PT routes (Route nos. 107 [0.85] and 108
study of countries such as Thailand and Bangladesh,
[0.88]) and IPT (Route nos. 1 [0.93] and 7 [0.90])
integrating the routes of both systems based on the pas-
provide services to the northwestern part of the
senger demand requirements and considering IPT routes
city. Despite a 100% overlap of the IPT routes on
while scheduling the PT system could help increase the
PT routes, both systems works well and can sat-
efficiency of both the systems. The detailed investigation
isfy the urban travel demand in these areas. Still
of origin and destination patterns of urban travelers and
the efficiency scores of IPT routes were found to
the availability of the IPT mode as a feeder system at
be higher than those of the PT routes.
 One PT route (Route no. 3 [0.816]) is a combina- major locations needs to be carried out to integrate the
tion of six IPT routes (Route nos. 2 [0.89], 4 PT as well as the IPT systems (68, 70, 71).
[0.922], 5 [1], 6 [0.806], 8 [1], and 12 [0.93]) of
shorter trip lengths along the same stretch of road
serving passengers in the central part of the city.
Development of an Integrated Information System (IIS)
Despite such an overlap, each of the routes caters The inclusion of the formalized fixed IPT routes should
to significant passenger demand compared with be incorporated into the existing city bus application.
seats supplied and has higher efficiency scores. Transit maps, trip planners, real-time information of
 There are radial routes serving the different parts upcoming transits, fare by each mode of travel, and alter-
of the city in all directions: PT route no. 109 native modes of travel are the type of information that
(0.952) and IPT route nos. 5 (1) and 13 (0.94) in should be made available in mobile applications to facili-
the east; PT route no. 116 (0.73) and IPT route tate trip planning of urban commuters. ‘‘Hong Kong e-
no. 2 (1) in the south; PT route no. 209 (0.82) and Transport’’ in Hong Kong, ‘‘TRANS Link’’ in Australia,
IPT route no. 9 (1) in the southeast; PT route no. and ‘‘TFL Journey planner’’ in London are examples of
153 (1) and IPT route no. 11 (0.92) in the north- some successful mobile applications for public transport
east. These run parallel in such a way that IPT trip planning.
routes operate up to a certain length, whereas PT
routes provide connectivity to the outer suburban
areas, meaning that both PT and IPT routes cater Formulation of an Urban Metropolitan
to significant passenger demand. Transport Authority
The creation of the Urban Metropolitan Transport
Authority (UMTA) will help the city to cater to the
Policy Interventions mobility needs of its citizens more efficiently. It will inte-
Over time, the PT and IPT systems in the city have grate all modes at functional, financial, social, and policy
evolved in such a way that they either complement or levels and regulate the operations of all the shared trans-
compete with each other in certain areas of the city. port systems to attain sustainability (72, 73). The suc-
Based on the evaluation of these systems using both sin- cessful functioning of UMTAs has been seen in cities
gle and multiple efficiency measures, along with the eva- such as London, Singapore, and Hong Kong.
luation within and across the systems, the following
policy interventions are recommended.
Designing Interchanges and Multimodal Hubs
The development of physical interchanges such as trans-
Formalization of the Paratransit System port hubs and center points with embedded exchange
The IPT system has evolved as a dominant mode of facilities such as an islands, interchanges, and covert
urban transport in Indian cities, generating ample walkways is recommended at important locations of the
employment opportunities. This mode focuses on the city. This will make it easier for passengers to transfer
individual advantage catering to the high demand of pas- from one mode to another and provide accessibility as
senger growth. Based on lessons learned from developing well as connectivity. King’s Cross Station in London,
countries such as Turkey and some Africa countries, a Union Station in Los Angeles, Penn Station in New
regulatory framework in relation to routes of operation York City, and Finch Station in Toronto are the exam-
is required to formalize the IPT system (70). ples of some successful multimodal hubs worldwide.
Dawda et al 29

Introduction of an Integrated Fare System Author Contributions


Fare integration is a unique feature of multimodal trans- The authors confirm contribution to the paper as follows: study
port systems. This suggests the need to develop a unique conception and design: Dawda and Joshi; data collection:
fare payment e-mobility card such as the Oyster Card in Dawda and Gajera; analysis and interpretation of results:
Arkatkar, Dawda, Dave; draft manuscript preparation: Dawda,
London, the Octopus Card in Hong Kong, and the EZ
Gajera, Joshi, Arkatkar, and Dave. All authors reviewed the
Card in Singapore. The past literature reveals that an results and approved the final version of the manuscript.
integrated fare collection system results in an increase in
passenger trips and reduces travel time (74).
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
Concluding Remarks respect to the research, authorship, and/or publication of this
article.
In most Indian cities, the two important modes of urban
transportation are PT and IPT systems. However, inte-
gration between them is non-existent across different Funding
routes which leads to conflict and the inefficiency of the The author(s) received no financial support for the research,
PT system. To understand and integrate these two modes authorship, and/or publication of this article.
and their distribution across different routes in a city,
proper evaluation of the modes needs to be conducted. References
The present study sought to conduct a microscopic eva-
1. Agarwal, D. O. P., M. S. Rathi, M. K. Kalra, M. M.
luation of these modes. To evaluate the existing PT and Gupta, M. S. Pal, M. A. Lakshmi, M. S. Bhattacharya,
IPT system, three efficiency measures—route, cost, and and M. A. Mishra. Review of Urban Transport in India.
operation—were analyzed separately. Further, they were Institute of Urban Transport, Delhi, 2014.
analyzed together with eight scenarios. Finally, efficiency 2. Bhandari, A., S. Juyal, H. Maini, A. Saxena, and , A. Sri-
across PT and IPT systems was also probed with the help vastava. Moving Forward Together: Enabling Shared Mobi-
of a radar diagram. Some of the conclusions based on lity in India. NITI Aayog, 2018.
the present study are given here: 3. Singh, S. K. Review of Urban Transportation in India.
Journal of Public Transportation, Vol. 8, No. 1, 2005,
pp. 79–97.
 In response to the absence of efficiency evaluation 4. Adhikaria, A., S. Basu, I. Biswas, A. Banerjee, and P. P.
models for PT and IPT systems together, the pres- Senguptae. A Route Efficiency Analysis using Shannon
ent paper proposes a framework for evaluating Entropy-Based Modified DEA Method and Route Char-
both systems using DEA techniques. The pro- acteristics Investigation for Urban Bus Transport in India.
posed methodology can be used to evaluate the Infor: Information Systems and Operational Research, Vol.
efficiency of the multiple modes of public trans- 56, No. 3, 2018, pp. 332–359.
port (metro, city bus, BRTS, tram, LRT, BRT, 5. NUTP. National Urban Transport Policy, 2014. Ministry
and IPT) together. It will help operators as well as of Housing and Urban Affairs, New Delhi, 2014.
city planners to understand how their public tran- 6. Hendy, P., and A. Flausch. Public Transport Trends. UTPI
sit is running and offer possible solutions for cre- Awards, Milan, 2015.
7. Flausch A. India Should Focus on Public Transport in Cities
ating an integrated transport system. This
in Coming 10 Years - Secretary General at UTTP Brussels.
framework was demonstrated using a case study UITP, India, 2017.
in Surat where the IPT and PT networks overlap 8. Gadepalli, R., C. Fabianski, J. Pourbaix, and J. Singh.
significantly. Regulatory Frameworks for Integrated Shared Mobility
 Route evaluation based on single efficiency mea- Governance in India. UITP, India, 2018.
sures may not be accurate because other measures 9. Eboli, L., and G. Mazzulla. A Methodology for Evaluat-
will be neglected. However, when all efficiency ing Transit Service Quality Based on Subjective and Objec-
measures are taken into consideration, route effi- tive Measures from the Passenger’s Point of View.
ciency can be analyzed more comprehensively. Transport Policy, Vol. 18, 2011, pp. 172–181. https:
Therefore, eight different scenarios are given in //doi.org/10.1016/j.tranpol.2010.07.007.
the present study, taking into account all possible 10. Kramer, F. A. Public Management in the 1980s and
Beyond. The ANNALS of the American Academy of Politi-
combinations of efficiency measures.
cal and Social Science, Vol. 466, No. 1, 1983, pp. 91–102.
 The present work provides an insight into the eva- https://doi.org/10.1177/0002716283466001006.
luation of multiple public transport modes which 11. Chu, X., G. J. Fielding, and B. W. Lamar. Measuring
together could pave the way to integrated policy Transit Performance using Data Envelopment Analysis.
formulation. Transportation Research Part A: Policy and Practice, Vol.
30 Transportation Research Record 2675(3)

26, No. 3, 1992, pp. 223–230. https://doi.org/10.1016/0965- 25. Tongzon, J. Efficiency Measurement of Selected Australian
8564(92)90033-4. and Other International Ports using Data Envelopment
12. Fielding, G. J. Transit Performance Evaluation in the Analysis. Transportation Research Part A: Policy and Prac-
U.S.A. Transportation Research Part A: Policy and Prac- tice, Vol. 35, No. 2, 2001, pp. 107–122.
tice, Vol. 26, No. 6, 1992, pp. 483–491. https://doi.org/ 26. Tone, K., and T. Sawada. An Efficiency Analysis of Public
10.1016/0965-8564(92)90029-7. versus Private Bus Transportation Enterprises. Operational
13. Hawas, Y. E., B. Khan, and N. Basu. Evaluating and Research, Vol. 9, 1991, p. 365.
Enhancing the Operational Performance of Public Bus Sys- 27. Obeng, K. The Economic Cost of Subsidy-Induced Techni-
tems using GIS-Based Data Envelopment Analysis. Journal cal Inefficiency. International Journal of Transport Econom-
of Public Transportation, Vol. 15, No. 2, 2012, pp. 19–44. ics, Vol. 21, 1994, pp. 3–20.
14. Gajera, H., N. Dawda, S. Dave, G. Joshi, and S. Arkatkar. 28. Nolan, J. F. Determinants of Productive Efficiency in
Evaluating Operational Performance of City Bus Transit Urban Transit. Logistics Transport Review, Vol. 32, No. 3,
Service using TCQSM Guidelines: A Case Study of Surat 1996, pp. 319–342.
City. Journal of the Eastern Asia Society for Transportation 29. De Borger, B., K. Kerstens, and A. Costa. Public Transit
Studies, Vol. 13, 2019, pp. 1361–1381. https://doi.org/ Performance. What Does One Learn from Frontier Stud-
10.11175/easts.13.1361. ies? Transportation Review, Vol. 22, No. 1, 2002, pp. 1–38.
15. Hendrickson, C. A Note on Trends in Transit Commuting 30. Oh, M. Y., S. S. Kim, and M. J. Kim. Analyzing Efficiency
in the United States Relating to Employment in the Cen- in the Seoul’s Urban Bus Industry using Data Envelopment
tral Business District. Transportation Research Part A: Analysis. Journal of Korean Society of Transportation, Vol.
General, Vol. 20, No. 1, 1986, pp. 33–37. https://doi.org/ 20, No. 2, 2002, pp. 59–68.
10.1016/0191-2607(86)90013-0. 31. Swami, M., and M. Parida. Comparative Appraisal of
16. Karlaftis, M. G., and P. S. McCarthy. Subsidy and Public Metro Stations in Delhi using Data Envelopment Analysis
Transit Performance: A Factor Analytic Approach. Trans- in a Multimodal Context. Journal of Public Transportation,
portation, Vol. 24, No. 3, 1997, pp. 253–270. https:// Vol. 18, No. 3, 2015, p. 3. https://doi.org/10.5038/2375-
doi.org/10.1023/A: 1004956532174. 0901.18.3.3.
17. Boschken, H. L. Behavior of Urban Public Authorities 32. Seth, C., K. Triantis, and D. Teodorović. Performance
Operating in Competitive Markets: Policy Outcomes in Mass Evaluation of Bus Routes: A Provider of Passenger Per-
Transit. Administration & Society, Vol. 31, No. 6, 2000, spective. Transport Research Part E: Logistics and Trans-
pp. 726–758. https://doi.org/10.1177/00953990 022019308. portation Review, Vol. 43, No. 4, 2007, pp. 453–478.
18. Obeng, K., and G. A. Azam. Allocative Distortions from 33. Lao, Y., and L. Liu. Performance Evaluation of Bus Lines
Transit Subsidies. International Journal of Transport Eco- with Data Envelopment Analysis and Geographic Informa-
nomics, Vol. 22, No. 1, 1995, pp. 15–34. tion Systems. Computers, Environment and Urban Systems,
19. Pina, V., and L. Torres. Analysis of the Efficiency of Local Vol. 33, No. 4, 2009, pp. 247–255.
Government Services Delivery. An Application to Urban 34. Sun, D. J., S. Chen, C. Zhang, and S. Shen. A Bus Route
Public Transport. Transportation Research Part A: Policy Evaluation Model Based on GIS and Super-Efficient Data
and Practice, Vol. 35, No. 10, 2001, pp. 929–944. Envelopment Analysis. Transportation Planning and Tech-
20. Garcı́a-Meca, E., and J. P. Sánchez-Ballesta. Corporate nology, Vol. 39, No. 4, 2016, pp. 407–423. https://doi
Governance and Earnings Management: A Meta-Analysis. .org/10.1080/03081060.2016.1160582.
Corporate Governance: An International Review, Vol. 17, 35. Sun, L., J. Rong, and L. Yao. Measuring Transfer Effi-
No. 5, 2009, pp. 594–610. https://doi.org/10.1111/j.1467- ciency of Urban Public Transportation Terminals by Data
8683.2009.00753.x. Envelopment Analysis. Journal of Urban Planning and
21. Barnum, D., and J. M. Gleason. Technical Efficiency Bias Development, Vol. 136, No. 4, 2010, pp. 314–319.
in Data Envelopment Analysis Caused by Intra-Output 36. Banker R. D., A. Charnes, W. W. Cooper. Some Models
Aggregation. Applied Economics Letters, Vol. 14, 2007, for Estimating Technical and Scale Inefficiencies in Data
pp. 623–626. https://doi.org/10.1080/13504850500461647. Envelopment Analysis. Manage Science, Vol. 30, No. 9,
22. Barnum, D. T., S. Tandon, and S. McNeil. Comparing the 1984, pp. 1078–1092.
Performance of Bus Routes after Adjusting for the Environ- 37. Karim, Z., and J. Fouad. Measuring Urban Public Trans-
ment using Data Envelopment Analysis. Journal of Trans- port Performance on Route Level: A Literature Review.
portation Engineering, Vol. 134, No. 2, 2008, pp. 77–85. Proc., MATEC Web of Conferences, Rabat, Morocco, Vol.
23. Karlaftis, M. G., and D. Tsamboulas. Efficiency Measure- 200, 2018, pp. 1–9.
ment in Public Transport: Are Findings Specification 38. Nguyen-Hoang, P., and R. Yeung. What Is Paratransit
Sensitive? Transportation Research Part A: Policy and Worth? Transportation Research Part A: Policy and Prac-
Practice, Vol. 46, No. 2, 2012, pp. 392–402. https://doi tice, Vol. 44, No. 10, 2010, pp. 841–853. https://doi.org/
.org/10.1016/j.tra.2011.10.005. 10.1016/j.tra.2010.08.006.
24. Zhu, J. Quantitative Models for Performance Evaluation 39. Behrensa, R., D. McCormick, R. Orero, and M. Ommeh.
and Benchmarking: Data Envelopment Analysis with Spread- Improving Paratransit Service: Lessons from Inter-City
sheets. Operational Research, Vol. 1, Springer, New York, Matatu Cooperatives in Kenya. Transport Policy, Vol. 53,
NY, 2009. 2017, pp. 79–88. https://doi.org/10.1016/j.tranpol.2016.09.003.
Dawda et al 31

40. Kheang, P. V. Characteristics and Perceptions of Paratran- 53. Dandapat, S., M. F. Cheranchery, and B. Maitra. Is Fare
sit Users in Phnom Penh. Journal of the Eastern Asia Society Increment Desirable for Ensuring Operational Viability of
for Transportation Studies, Vol. 11, 2013, pp. 1451–1466. Private Buses? Transport Policy, Vol. 59, 2017,
41. Joewono, T. B., and H. Kubota. User Perceptions of Pri- pp. 134–141. https://doi.org/10.1016/j.tranpol.2017.07.010.
vate Paratransit Operation in Indonesia. Journal of Public 54. Arora, A. Integrating Intermediate Public Transport Within
Transportation, Vol. 10, No. 4, 2007, p. 5. Transport Regulation in a Megacity: A Kolkata Case Study.
42. Rahman, F., T. Das, M. Hadiuzzaman, and S. Hossain. Shakti Sustainable Engery and Foundation, Kolkata, 2016.
Perceived Service Quality of Paratransit in Developing 55. Prabhu, A. D., S. Madhu, L. Ramamurthy, and D. Dha-
Countries: A Structural Equation Approach. Transporta- nuraj. Study on Paratransit Sector in Chennai. Civitas
tion Research Part A: Policy and Practice, Vol. 93, 2016, Urban Solutions for City Connect Foundation Chennai
pp. 23–38. (CCCF), Chennai, India, 2011.
43. Census of India. Ministry of Home Affairs, Government 56. Lave, R. E., and J. W. Billheimer. Paratransit Integration,
of India. https://censusindia.gov.in/. Accessed March 12, State-of-Art Report. Transportation Research Board, 1978.
2018. http://onlinepubs.trb.org/onlinepubs/millennium/00107.pdf.
44. Regional Transport Office. Government of Gujarat. Accessed December 15, 2020.
https://cot.gujarat.gov.in/statistics-surat-en.htm. Accessed 57. Gkiotsalitis, K., Z. Wu, and O. S. Cats. Bus Allocation to
July 29, 2019. Short-Turning and Interlining Lines. Presented at 98th
45. Ministry of Urban Development. Comprehensive Mobility Annual Meeting of the Transportation Research Board,
Plan (2008) – Comprehensive Mobility Plan and Bus Rapid Washington, D.C., 2019.
Transit System Plan. Ministry of Urban Development, 58. Vuchic, V. R. Skip-Stop Operation as a Method for Tran-
Government of India, New Delhi, 2008. https://smartnet. sit Speed Increase. Traffic Quarterly, Vol. April, 1973,
niua.org/sites/default/files/resources/file_1016201405372097. pp. 307–327.
pdf. Accessed December 15, 2020. 59. Zhang, H., S. Zhao, Y. Cao, H. Liu, and S. Liang. Real-
46. National Academies of Sciences, Engineering, and Medi- Time Integrated Limited-Stop and Short-Turning Bus
cine; Transportation Research Board; Transit Cooperative Control with Stochastic Travel Time. Journal of Advanced
Research Program; Kittelson & Associates, Inc.; Parsons Transportation, Vol. 2017, 2017, p. 9. https://doi.org/
Brinckerhoff; KFH Group, Inc.; Texas A&M Transporta- 10.1155/2017/2960728.
tion Institute. TCRP Report 165: Transit Capacity and 60. Daganzo, C. F., and Y. Ouyang. Public Transportation Sys-
Quality of Service Manual, 3rd ed. Transportation tems: Principles of System Design, Operations Planning and
Research Board of the National Academies, Washington, Real-Time Control. World Scientific Publishing Co., Singa-
D.C., 2013. pore, 2019. https://doi.org/10.1142/10553.
47. Service Level Benchmark of Urban Transport. Ministry of 61. Wu, W., R. Liu, and W. Jin. Integrating Bus Holding Con-
Urban Devlopment. http://utbenchmark.in/UsersidePages/ trol Strategies and Schedule Recovery: Simulation-Based
CityProfile.aspx?City=6. Accessed July 27, 2019. Comparison and Recommendation. Journal of Advanced
48. Lao, Y., and L. Liu. Computers, Environment and Urban Transportation, Vol. 2018, 2018, p. 13. https://doi.org/
Systems Performance Evaluation of Bus Lines with Data 10.1155/2018/9407801.
Envelopment Analysis and Geographic Information Sys- 62. Lin, Y., X. Yang, N. Zou, and M. Franz. Transit Signal
tems. Computers, Environment and Urban Systems, Vol. 33, Priority Control at Signalized Intersections: A Comprehen-
No. 4, 2009, pp. 247–255. https://doi.org/10.1016/j.com sive Review. Transportation Letters, Vol. 7, No. 3,
penvurbsys.2009.01.005. 2015, pp. 168–180. https://doi.org/10.1179/1942787514Y
49. Ardila, A. How Public Transportation’s Past Is Haunting .0000000044.
Its Future in Bogotá, Colombia. Transportation Research 63. Chakraborty, P. NMC to Rationalize Routes to Cut Losses
Record: Journal of the Transportation Research Board, in City Bus Ops. The Times of India, June, 2018.
2007. 2038: 9–15. 64. Hafezi, M. H., and A. Ismail. Balancing between Headway
50. Ardila, A. Limitation of Competition in and for the Public and Frequency Scheduling for Bus Service. Applied
Transportation Market in Developing Countries: Lessons Mechanics and Materials, Vol. 97–98, 2011, pp. 669–673.
from Latin American Cities. Transportation Research https://doi.org/10.4028/www.scientific.net/AMM.97-98.669.
Record: Journal of the Transportation Research Board, 65. Rodrigues, M. M., C. C. de Souza, and A. V. Moura. Vehi-
2008. 2048: 8–15. cle and Crew Scheduling for Urban Bus Lines. European
51. Tang, S., and H. K. Lo. The Impact of Public Transport Journal of Operational Research, Vol. 170, No. 3, 2006,
Policy on the Viability and Sustainability of Mass Railway pp. 844–862. https://doi.org/10.1016/j.ejor.2004.06.035.
Transit – The Hong Kong Experience. Transportation 66. Vuchic, V. R. Urban Transit: Operations, Planning and Eco-
Research Part A: Policy and Practice, Vol. 42, No. 4, 2008, nomics. J. Wiley & Sons, Hoboken, NJ, 2005.
pp. 563–576. https://doi.org/10.1016/j.tra.2008.01.022. 67. Directorate of Urban Land Transport. Davanagere City
52. Feng, S., X. Shen, and B. Hu. Over-Supply in Public Trans- Bus Service Evaluation Report. Directorate of Urban Land
portation: Case Study of Bus and Metro Lines in Harbin Transport, 2013. http://www.urbantransport.kar.gov.in/
City, China. Promet - Traffic & Transportation, Vol. 28, projects_query.html. Accessed December 15, 2020.
No. 5, 2016, pp. 471–477. https://doi.org/10.7307/ptt.v28i 68. Tangphaisankun, A., T. Okamura, F. Nakamura, and R.
5.1936. Wang. A Study in Integrating Paratransit as a Feeder
32 Transportation Research Record 2675(3)

System into Urban Transportation and Its Effects on 72. Dawda, N., G. Joshi, S. Arkatkar, and N. Vasudevan.
Mode Choice Behavior: A Study in Bangkok, Thailand. Multimodal of Lateral Transport System: A Case Study of
Proc., 12th WCTR Conference, Lisbon, Portugal, 2010, Successful Cities Worldwide. In Innovative Research in
pp. 1–22. Transportation Infrastructure – Lecture Notes in Intelligent
69. Ferro, P. S., and R. Behrens. From Direct to Trunk-and- Transportation and Infrastructure (D. Deb, V. Balas, R.
Feeder Public Transport Services in the Urban South: Ter- Dey, and J. Shah, eds.), Springer, Singapore, 2019,
ritorial Implications. Journal of Transport and Land Use, pp. 101–110.
Vol. 8, No. 1, 2015, pp. 123–136. https://doi.org/ 73. Schalekamp, H., and R. Behrens. Engaging Paratransit on
10.5198/jtlu.2015.389. Public Transport Reform Initiatives in South Africa: A
70. Kalpakcı, A., and N. K. Ünverdi. Integration of Paratran- Critique of Policy and an Investigation of Appropriate
sit Systems with Inner-City Bus Transport: The Case of Engagement Approaches. Research in Transportation Eco-
Izmir. Public Transport, Vol. 8, No. 3, 2016, pp. 405–426. nomics, Vol. 29, No. 1, 2010, pp. 371–378. https://doi.org/
https://doi.org/10.1007/s12469-016-0136-9. 10.1016/j.retrec.2010.07.047.
71. Shen, C.-W., and L. Quadrifoglio. Evaluation of Zoning 74. Sharaby, N., and Y. Shiftan. The Impact of Fare Integra-
Design with Transfers for Paratransit Services. Transporta- tion on Travel Behavior and Transit Ridership. Transport
tion Research Record: Journal of the Transportation Policy, Vol. 21, 2012, pp. 63–70. https://doi.org/10.1016/
Research Board, 2012. 2277: 82–89. j.tranpol.2012.01.015.

You might also like