analysis of paratransit drivers' stated job choice

16
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011 Analysis of Paratransit Drivers' Stated Job Choice Behavior under Various Policy Interventions Incorporating the Influence of Captivity: A Case Study in Jabodetabek Metropolitan Area, Indonesia Gang LI Doctor Candidate Graduate School for International Development and Cooperation, Hiroshima University, 1-5-1 Higashi Hiroshima, Japan Fax: +81-82-424-6919 E-mail: [email protected] Junyi ZHANG Associate Professor Graduate School for International Development and Cooperation, Hiroshima University, 1-5-1 Higashi Hiroshima, Japan Fax: +81-82-424-6919 E-mail: [email protected] Sudarmanto Budi NUGROHO Assistant Professor Graduate School for International Development and Cooperation, Hiroshima University, 1-5-1 Higashi Hiroshima, Japan Fax: +81-82-424-6921 E-mail: [email protected] Tran Ngoc LINH Doctor Candidate Graduate School for International Development and Cooperation, Hiroshima University, 1-5-1 Higashi Hiroshima, Japan Fax: +81-82-424-6921 E-mail: [email protected] Akimasa FUJIWARA Professor Graduate School for International Development and Cooperation, Hiroshima University, 1-5-1 Higashi Hiroshima, Japan Fax: +81-82-424-6921 E-mail: [email protected] Abstract: This study analyzes paratransit driversstated job choice behavior under various policy interventions. First, the current profiles of four typical types of paratransit drivers (becak, ojek, bajaj, and angkot drivers) are clarified in Jabodetabek Metropolitan Area, Indonesia. Second, factors affecting their future driver job choices under policy interventions from social and environmental aspects were investigated based on a stated preference (SP) survey, conducted in 2010. In the SP survey, future driver jobs are assumed by considering salary level, employment opportunity, employment status, operation cost, subsidy for low- emission vehicles, vehicle fuel types, and so on. Third, direct evaluation of different factors on job choices were conducted by cross-tabulations. Dogit model was further applied to properly represent the captive job decision. The model estimation results confirmed that drivers are relatively captive to ojek and bajaj, and subsidy for low-emission vehicles is influential to driversstated job choices about ojek and bajaj. Key Words: paratransit, policy interventions, captive job, dogit model, SP survey 1. INTRODUCTION The concept of paratransit is quite different between developed and developing countries. In developed countries, paratransit usually refers to the demand-responsive and door-to-door 1144

Upload: others

Post on 29-Mar-2022

12 views

Category:

Documents


0 download

TRANSCRIPT

Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
Analysis of Paratransit Drivers' Stated Job Choice Behavior under Various Policy Interventions Incorporating the Influence of Captivity: A Case Study
in Jabodetabek Metropolitan Area, Indonesia
Gang LI
Doctor Candidate
Fax: +81-82-424-6919
E-mail: [email protected]
Junyi ZHANG
Associate Professor
Fax: +81-82-424-6919
E-mail: [email protected]
Fax: +81-82-424-6921
E-mail: [email protected]
Fax: +81-82-424-6921
E-mail: [email protected]
Akimasa FUJIWARA
Fax: +81-82-424-6921
E-mail: [email protected]
Abstract: This study analyzes paratransit drivers’ stated job choice behavior under various
policy interventions. First, the current profiles of four typical types of paratransit drivers
(becak, ojek, bajaj, and angkot drivers) are clarified in Jabodetabek Metropolitan Area,
Indonesia. Second, factors affecting their future driver job choices under policy interventions
from social and environmental aspects were investigated based on a stated preference (SP)
survey, conducted in 2010. In the SP survey, future driver jobs are assumed by considering
salary level, employment opportunity, employment status, operation cost, subsidy for low-
emission vehicles, vehicle fuel types, and so on. Third, direct evaluation of different factors on
job choices were conducted by cross-tabulations. Dogit model was further applied to properly
represent the captive job decision. The model estimation results confirmed that drivers are
relatively captive to ojek and bajaj, and subsidy for low-emission vehicles is influential to
drivers’ stated job choices about ojek and bajaj.
Key Words: paratransit, policy interventions, captive job, dogit model, SP survey
1. INTRODUCTION
The concept of paratransit is quite different between developed and developing countries. In
developed countries, paratransit usually refers to the demand-responsive and door-to-door
1144
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
transport service exclusively for the elderly and people with disability. In developing countries,
lower living standard, poor public transportation service, cheap labor force, etc, have
contributed paratransit as public transportation to serve the whole population, especially the
poor. Paratransit consists of a variety of transport modes from pedicabs, motorcycles to van-
type minibus, operated by individuals and small companies, adoptable in its routing and
scheduling to individual user's desires in varying degrees, and takes the role of “gap filler”
between conventional bus and private automobiles (Shimazaki and Rahman, 1996). The use
of paratransit provides developing countries with several advantages, such as the mobility,
especially for the poor, urban employment, especially for the unskilled persons, feeder
connections between neighborhoods and trunk routes, the flexibility and sensitiveness to the
changing market and so on. On the other hand, paratransit also creates several problems such
as traffic congestion, traffic accident, environmental pollution and so on (Cervero, 2000).
In the past decade, paratransit in developing countries is attracting more and more attentions
due to the above-mentioned debating features. Consequently a variety of studies focus on the
operation service of paratransit in terms of the demand side such as users’ satisfaction and
loyalty on the usage of paratransit (Joewono and Kubota, 2007a; Joewono and Kubota, 2007b;
Tarigan et al., 2010). With respect to the supply side of paratransit, the research mainly
focuses on the descriptive analysis of social-economic characteristics of paratransit drivers
and the corresponding features of paratransit vehicles by using simple statistical tools
(Cervero and Golub, 2007; Joewono and Kubota, 2005; Shimazaki and Rahman, 1996). The
research, particularly and comprehensively for the paratransit drivers’ job condition and their
relevant operation situation, is quite rare. It is still limited to the drivers’ lives in the socio-
economic aspects and the corresponding industry of that paratransit type (Azuma, 2000;
Etherington and Simon, 1996). Actually from the viewpoint of well balancing the demand side
and the supply side of paratransit in the urban area, the deep and comprehensive
understanding about the current situations of various types of paratransit drivers and their job
choice under various policy interventions in the future is quite important. Such policy
interventions can be viewed from the social prospect, which argues the importance of
providing driving-job opportunities to unskilled (low-income) persons, and the environmental
prospect, which argues the importance of reducing emissions from paratransit systems. The
regulation and its effects of paratransit have been partially discussed in literature (Cervero and
Golub, 2007; Diaz and Cal, 2005; Schalekamp et al., 2009). Authors believe, in a
considerably long period of time with the current economic developments in developing
countries, regulating and incorporating paratransit into formal public transportation system
rather than simply banning it, is a wise method to realize the sustainable urban transportation.
Therefore, in this study, authors suppose, in the near future, the government regulate
paratransit operation by offering the most possible driver job options for each type of current
paratransit drivers, respectively (for example, becak drivers can become ojek and bajaj drivers
but not medium bus drivers). For the new jobs, current paratransit drivers are required to buy
new vehicles with either electricity/CNG or gasoline power, and to operate their business
either as access/egress modes or trunk modes in respective pre-planned and restricted areas by
government. The government offers the subsidy and facilitates the maximum five-year loan
for drivers to buy new vehicles, especially the low-emission types. The benefit for the current
paratransit drivers is legal status, vehicle ownership (especially for the drivers without
ownership currently) and increased monthly income (especially for the drivers with ownership
previously) due to the saving of operation cost of new vehicles and orderly operation. Here
income (salary) doesn’t have any specific definition such as gross income or take home pay,
and its unit is million Rp (Rp, i.e., Indonesian rupiah). Through such policy interventions, the
1145
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
current paratransit drivers can be effectively persuaded to shift to new paratransit drivers, so
that the government can make a comprehensive and sustainable urban transportation plan to
cover all transport services. To the authors’ best knowledge, no research has clarified social
and environmental policy interventions from the paratransit point so far. Therefore, this paper
attempts to employ the dogit model (Gaudry and Dagenais, 1979) to capture paratransit
drivers’ stated job choice behavior under various policy interventions and identify whether
there are “captive jobs” for paratransit drivers in order to achieve the sustainable paratransit
system in developing countries in future.
The rest of the paper is organized as follows. In Section 2, the basic knowledge of paratransit
in Jabodetabek area, Indonesia is introduced. Data used in this study to explore paratransit
drivers’ current profiles and their stated job choices under future hypothetical situations is
described in Section 3, where influential factors affecting future job choices are also examined
based on cross-tabulation. Dogit model is described in Section 4 and the corresponding
estimation results are discussed in Section 5. Finally, the study is concluded in Section 6 along
with a discussion about future research issues.
2. PARATRANSIT IN JABODETABEK METROPOLITAN AREA, INDONESIA
Paratransit represents various transport modes mainly including becak, ojek, bajaj and angkot
and constitutes a hierarchy of services complementary to the insufficient formal public
transportation system in Jabodetabek Metropolitan Area (JMA), Indonesia. Becak is a
manpower-bicycle taxi with three wheels to offer the door-to-door service in neighborhood for
the fare that can be negotiated. It is banned in DKI Jakarta but not in other places in JMA.
(Cervero, 2000). Ojek is a motorcycle taxi for a negotiated fare. It also provides the door-to-
door connectivity but with advantages of speed and travel range compared with becak. It is
actually private mode but for public use, so it is completely illegal public transport mode.
Bajaj is a sort of registered auto-rickshaw taxis with three wheelers and also offers the door-
to-door service for a negotiated fare. Additionally it is allowed to cross major roads but can’t
travel on them in DKI Jakarta. Angkot is a popular public mode for passengers with a fixed
route, but without a fixed schedule, and follows the designated routes in the city’s network.
Additionally various types of cars and vans with a capacity of 12-16 seats are used as angkot
(Joewono and Kubota, 2007c). To clarify the image of four typical types of paratransit in
JMA, Indonesia, basic characteristics are summarized in Table 1 and the corresponding
pictures used here are from websites.
Table 1 Basic characteristics of paratransit in JMA, Indonesia Type Routes Schedules Capacity (Unit: person) Service Niche
Angkot Fixed un-Fixed 12--16 Mixed
Bajaj un-Fixed un-Fixed 2--3 Feeder
Ojek un-Fixed un-Fixed 1--2 Feeder
Becak un-Fixed un-Fixed 1--2 Feeder
Becak Ojek Bajaj Angkot
1146
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
3. DATA In this study, data were collected in JMA, Indonesia, which has a population of about 21
million and consists of DKI Jakarta, the capital city, and 7 local governments (Kabupaten and
Kota Bogor, Kabupaten and Kota Tangerang, Kota Depok and Kabupaten and Kota Bekasi).
Its gross regional domestic product is estimated at 351,000 billion Rp in 2002, or 22% of the
national gross domestic product, showing that JMA is strategically the most important region
of the nation (JICA, 2004). 3.1 Profiles of current paratransit drivers The data was collected based on face-to-face interviews to assure the quality of survey. The
targeted sample size was about 200 for each type of paratransit drivers considering the budget
limitation and the survey was conducted in February, 2010. The questionnaire consists of 7
parts. The first part begins with the requirements of filling in the drivers’ one-day travel diary.
The second part focuses on observations of passengers from drivers’ viewpoint such as main
destinations of passengers and so on. The questions about the basic information of currently
used paratransit vehicles are arranged in the third part. The fourth part contains the questions
about jobs as current paratransit drivers. The SP design about job choices in the future is
arranged in the fifth part. Finally the sixth and seventh parts are used to collect paratransit
drivers’ individual information and their family information respectively. The profiles of four
types of paratransit drivers are listed in Table 2.
Table 2 Profiles of four types of paratransit drivers
Fq % Fq % Fq % Fq %
Gender Male 192 100% 208 100% 194 100% 205 100%
< 30 years old 67 34.9% 73 35.1% 39 20.1% 21 10.2%
30 ~ 39 years old 76 39.6% 74 35.6% 67 34.5% 99 48.3%
40 ~ 49 years old 43 22.4% 45 21.6% 52 26.8% 70 34.1%
> 50 years old 6 3.1% 16 7.7% 36 18.6% 15 7.3%
Single 47 24.5% 54 26% 24 12.4% 14 6.8%
Married 145 75.5% 154 74% 170 87.6% 191 93.2%
Has children 138 71.9% 137 65.9% 160 82.5% 185 90.2%
Elementary school and below 98 51.0% 18 8.7% 60 30.9% 15 7.3%
Secondary school 89 46.4% 76 36.5% 93 47.9% 78 38.0%
High school and above 5 2.6% 114 54.8% 41 21.1% 112 54.7%
< 1 million Rp 165 85.9% 66 31.7% 57 29.4% 29 14.1%
1 ~ 2 million Rp 17 8.9% 112 53.8% 132 68.0% 107 52.2%
> 2 million Rp 29 14.0% 5 2.6% 69 31.7%
Missing 10 5.2% 1 0.5% 4 2.0%
< 1 million Rp 145 75.5% 57 27.4% 43 22.2% 30 14.6%
1 ~ 2 million Rp 15 7.8% 94 45.2% 117 60.3% 95 46.3%
> 2 million Rp 55 26.4% 31 16.0% 79 38.6%
Missing 32 16.7% 2 1.0% 3 1.5% 1 0.5%
DKI Jakarta 22 11.5% 121 58.2% 183 94.3% 143 69.8%
Other places 161 83.8% 79 38.0% 7 3.6% 40 19.5%
Missing 9 4.7% 8 3.8% 4 2.1% 22 10.7%
With ownership 97 50.5% 201 96.6% 2 1.0% 24 11.7%
Without ownership 95 49.5% 7 3.4% 192 99.0% 181 88.3%
Fq and % : the frequency and the corresponding percentage of each category
Eduvation
level
Marital
status
Vehicle
ownership
income
Household
monthly
income
Living
district
208
1147
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
It is clear that all the paratransit drivers are male. Compared with other paratransit drivers,
becak drivers are quite younger from the age structure that the shares of the persons with the
age less than 40 years old and more than 50 years old are the largest and the smallest,
respectively. This corresponds to the obvious features of becak job that is a much-needed
manpower job and is not suitable for the older. The bajaj drivers is just the opposite side,
whose age structure has the smallest share of the group with less than 30 years old and the
largest share of the group with more than 50 years old. Ojek drivers are a little older than
becak drivers in terms of the share of the oldest group (> 50 years old). The age structure of
angkot drivers concentrates on the range from 30 to 49 years old, which indicates this job is
much norm compared with the other three.
Roughly speaking, single persons more engage in ojek driver jobs inferred by the highest share
of single status in marital status reaching about 26% among four types of paratransit drivers.
Single persons have to be becak drivers perhaps due to low education level and less skilled
resulting in limited job choices for them and much easier to access becak job in the sense that
only manpower is needed to a great extent. In contrast, the shares of single persons are lower
in bajaj drivers and angkot drivers. Accordingly, the proportions of having children in
paratransit drivers’ families follow this same order: angkot drivers > bajaj drivers > becak
drivers > ojek drivers.
In terms of education level, ojek drivers and angkot drivers obviously have the similar
education structures and accepted the highest education among four types of drivers. Then
bajaj drivers and becak drivers follows in sequence. Correspondingly from the common sense
that persons with better education obtain better salaries, the order of paratransit job monthly
income (taking home pay) justifies this point that angkot drivers > ojek drivers > bajaj
drivers > becak drivers on average. Accordingly this survey indicates that household monthly
income follows the same order among four types of drivers, because the paratransit job
incomes of most drivers are the main source of household incomes.
From this survey, it can infer that living districts of four types of drivers actually reflect the
current distributions of their business as paratransit drivers. The high percentage (83.8%) of
becak drivers living outside of DKI Jakarta actually is the real reflection that becak business is
banned in DKI Jakarta and only a few of becak drivers are still struggling at the edge of DKI
Jakarta. Ojek and angkot are quite common modes in JMA and their higher proportions in
DKI Jakarta than the ones in surrounding areas are partly due to both greater densities and are
partly caused by survey location. But for bajaj drivers, their operations are only located in
DKI Jakarta, which is also indicated by 99% of bajaj drivers living in DKI Jakarta.
The vehicle ownership clearly shows that for becak drivers, with becak ownership and without
becak ownership are almost half versus half. As expected, mostly all of ojek drivers (96.6%)
have ojek ownership, most of bajaj and angkot drivers don’t have the vehicle ownership,
reaching 99.0% and 88.3%, respectively.
3.2 SP design The SP survey was designed to explore a sustainable paratransit system from drivers’
viewpoints. The policy intervention is viewed from both social and environmental
perspectives. From the social perspective, the availability of current job (yes or not),
employment opportunity (2 or 3 levels: defined in the format of percentage compared to the
respective current job) and employment status (2 levels: self-employed vs. union member (or
1148
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
company-employed)) as a driver in the future are focused. From the environmental
perspective, vehicle fuel type (2 levels: gasoline vs. CNG (or electricity)) and subsidy for low-
emission vehicles (new vehicle type) (3 levels: no subsidy, low and high levels) are dealt with.
Other factors included in the SP design are operation cost of paratransit (2 levels: low and
high) and salary (2 or 3 levels based on current salaries for respective paratransit drivers).
Levels of each factors included in the SP survey are given based on the current situations,
opinions of local experts, and literature review. Orthogonal experiment design results in 16 SP
profiles (cards), which are further randomly grouped into 4 balanced blocks. Each driver in
the survey was only asked to answer one block with four SP questions, each of which includes
2 ~ 4 job options. Before answering each SP question, new types of jobs are briefly described
and expected salaries under different employment opportunities are also calculated and shown
in the card for ease of understanding. Table 3 shows an example of SP card for bajaj drivers,
where the alternatives with “New” mean that drivers should buy this type of new vehicles and
do such jobs. Table 4 lists the choice set for each type of paratransit drivers and the
corresponding results of choices. The resulting sample size is 768 for becak drivers, 832 for
ojek drivers, 776 for bajaj drivers, and 820 for angkot drivers, respectively.
Table 3 An example of SP card in bajaj drivers’ questionnaire
Alternative Vehicle
Fuel Type
New Ojek
5 million
Your Current
Job 4You continue to do the current job in future and all the attributes of the current job in future including the salary don't change
3.3 Drivers’ responses to various policy interventions In this study, the factors, which are assumed to have influences on job choices, are classified
into job attributes (JA) and individual characteristics (IC). To directly evaluate the drivers’
response to various policy interventions, the cross-tabulations between JA and job choices are
conducted, where JA includes the availability of current job, employment opportunity (EO),
and employment status (ES) from the social perspective, vehicle fuel type (FT) and subsidy
for low-emission vehicle (SU) from the environmental perspective. Due to the space
limitation, the cross-tabulations regarding with EO and SU from the respective perspectives
are taken as an example and showed in Figures 1 and 2.
Additionally, “New Bajaj Driver” option and “New Becak Driver” option in ojek and bajaj
drivers’ job choices are combined into one job option called “mix of bajaj and becak”
respectively, due to the limited choices of these two job options. Correspondingly, the
attributes of the combined option are given based on weighted attributes of both options
according to proportions of both options in choice results. As a result, four types of paratransit
drivers all have three options in the following analysis. The first and the third options are ojek
driver and current job, respectively. The second option differs among the four types of drivers,
where “Bajaj job” for becak drivers, “Mix job of bajaj and becak” for ojek and bajaj drivers,
and “Medium bus job” for angkot drivers, indicated by green bars with the name of “Bajaj job
/ Mix job of bajaj and becak / Medium bus job” in the legend in Figure 1 and Figure 2.
Figure 1 clearly shows that there is variation between EO and job choice and correspondingly
the basic trend is the increase from low level to middle level and then the decrease to high
1149
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
level for each type of drivers. It can be explained that the larger EO, the more drivers are
attracted, however, to balance EO and salary, most drivers tend to choose the jobs with middle
level of EO. Meanwhile, it is found that the choice percentage in high level is larger than that
in low level except becak drivers, which means motorized paratransit drivers are relatively
more concerned with EO than their salaries, comparing with becak drivers. It is quite
reasonable that as the poorest group, becak drivers are more greatly eager to change for a
better life indicated by salary (here low level of EO).
Figure 1 Cross-tabulation between EO and job choice
There seemingly be an obvious trend that with the decrease of SU, the increase of job choices
on the respective type of drivers indicated by Figure 2. Dose it means that the SU has a
negative influence on job choices? Actually although there are three levels in SU for each type
of drivers, none level of SU are 50% of 16 cards after conducting orthogonal experiment
designs, respectively, which makes the choice percentages in none level of SU inherently
larger than the ones in low and high level. Such result may indicate that many current
paratransit drivers prefer to buy vehicles with traditional FT (gasoline) due to the unfamiliarity
of vehicles with new power and/or high investment of such vehicles. This speculation should
be further carefully examined in the following model estimation.
Figure 2 Cross-tabulation between SU and job choice
4. DOGIT MODEL
The choice results in Table 4 show that a particular type of paratransit drivers obviously prefer
one or at most two jobs in the respective job choice set, where becak drivers desire “New Ojek
Driver” jobs, ojek drivers want to do the same jobs by buying new motorcycles, bajaj drivers
1150
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
Table 4 Choice sets and the corresponding results of paratransit drivers in SP survey
Set Fq % Set Fq % Set Fq % Set Fq %
Medium Bus Driver + 410 50.0%
New Bajaj Driver + 107 13.9% + 41 4.9% + 537 69.2%
New Ojek Driver + 536 69.8% + 627 75.4% + 194 25.0% + 358 43.7%
New Becak Driver + 47 5.6% + 14 1.8%
Current Job + 78 10.2% + 94 11.3% + 31 4.0% + 33 4.0%
No response 47 6.1% 23 2.8% 19 2.3% Set, Fq and % indicates the unique job choice set, the corresponding frequency and percentage of choice results for each type of paratransit
drivers. "+" shows the options (alternatives) in such unique job choice set.
Becak Driver Ojek Driver Bajaj Driver Angkot DriverGeneral Paratransit
Job Choice Set
prefer to own bajaj and continue this job (99% bajaj drivers don’t have ownership of bajaj as
shown in Table 1) and angkot drivers are willing to change the current job into either medium
bus drivers or ojek drivers. Additionally, to better depict the drivers’ response to various policy
interventions, which cannot be fully analyzed by cross-tabulations and to deal with discrete
choice behavior such as stated job choices here, multinomial logit (MNL) model is usually
used due to the simplicity and ease of estimation. However, the MNL model suffers from the
independence of irrelevant alternatives (IIA) property. Moreover, Swait and Ben-Akiva (1987)
indicated that if the choice set formation did not allow individuals to be captive to specific
alternatives or included alternatives never considered by the individuals, the resultant
misspecification of the choice set would potentially produce biased and inconsistent parameter
estimates of the choice models. In order to overcome such potential problems and together
respond to the concentrated distributions of choice results, i.e., particular types of paratransit
drivers may have great preferences for one type of job, this study attempts to apply dogit
model proposed by Gaudry and Dagenais (1979) to analyze stated job choice behavior of
paratransit drivers in JMA, Indonesia. Dogit model obviates the IIA difficulty without losing
the intuitive and practical appeal of the logit format which allows the relative probabilities of
some pairs of alternatives to be consistent with IIA property, but without simultaneously
destroying itself as a distinct model for the rest of the alternatives considered. The general
form of dogit model is showed below (the subscript notation indicating individual is omitted).
, , 1,..., , (1 )
i k





m n
where,
ip : probability of choosing the i th job from K job options in future
kV : deterministic term of the utility of the k th job option with the k th job’s constant ( k ), m
job attributes ( mkX ) and n individual characteristics ( nkX )
i : job-specific preference parameter, 0i
The probability ratio in Equation (3) shows that the dogit model does not have the IIA
property, meaning that changes of alternatives in choice set will change the relative
probabilities of alternatives. It is noted that the relative probabilities /i jp p is consistent with
the IIA property, only if 0i j or / / ji vv
i j e e . If 0i j holds for all pairs of
1151
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
alternatives, Equation (1) collapses to the logit model (Gaudry and Dagenais, 1979).
i k
j k
V V
Additionally, using Manski's (1977) framework, the dogit model can be conceptualized as
arising from a two-part choice process consisting of a choice set generation process and
conditional on choice set selection, an outcome selection process (Ben-Akiva, 1977; Swait
and Ben-Akiva, 1987; Fry and Harris, 1996), where the general form of dogit model in
Equation (1) is then rewritten as follows.
1
e
(4)
This form is widely considered that dogit model represents an individual either is captive to
one of the K alternatives (first term of Equation (4)) or chooses freely from the full set of
available alternatives (second term of Equation (4)) in a simple formula. Ben-Akiva (1977)
showed that, in the context of choice set generation, i can be interpreted as an index of
captivity, which can be further explained as a “preference” or a “loyalty” parameter.
Specifically in this study, the dogit model can be interpreted that allowance of captivity of
paratransit drivers to some specific jobs. The larger the value of i , the more captive a driver
to a specific job i of interest.
The maximum likelihood method is used to estimate the dogit model whose program is coded
by authors in the software Time Series Processor 5.0. By using dogit model, this study
attempts to explore what kinds of factors (JA and IC) and to what extent have influences on
job choices of paratransit drivers in hypothesized various policy interventions in the future
and further identify whether there is a “captive job” for a particular type of paratransit drivers.
Such analysis could contribute to the better understanding of job choice behavior of
paratransit drivers and further offer the sound foundation of the supply of paratransit as a
formal part of sustainable public transportation systems and the relevant policy making.
5. MODEL ESTIMATION The price of becak about 1 million Rp on average is much cheaper than the one of ojek about
10.6 million Rp, the one of bajaj about 19 million Rp and the one of angkot about 60 million
on average indicated in this survey, and fear of having one’s becak confiscated is another
reason for some becak drivers without ownership of becak (Cervero, 2000). Therefore, here, it
is assumed that there is no significant difference between becak drivers with becak ownership
and becak drivers without becak ownership. However, for ojek drivers, bajaj drivers and
angkot drivers, there should be a great difference between drivers with vehicle ownership and
drivers without vehicle ownership in economic aspect. Together with the three options
(alternatives) for all drivers and the assumed influencing IC mentioned above, four dogit job
choice models toward various policy interventions are established based on four types of
paratransit drivers, respectively. That is, becak drivers’ job choice model is based on all
1152
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
drivers regardless of vehicle ownership; ojek drivers’ job choice model is based on ojek
drivers with ownership; bajaj drivers’ job choice model is based on bajaj drivers without
bajaj ownership and angkot drivers’ job choice model is based on angkot drivers without
vehicle ownership. With the consideration of vehicle ownerships and deletions of missing
values in some influencing factors involved in dogit models, the valid sample for becak, ojek,
bajaj, angkot drivers are 715, 671, 684 and 612, respectively. The estimation results are then
shown in Table 5 and Table 6.
Generally speaking, for four types of paratransit drivers, dogit models can well capture their
job choice behaviors in response to the assumed various policy interventions in terms of
goodness-of-fit indicators (adjusted McFadden’s Rho-squared at zero) ranging from 0.185 to
0.431. Dogit models further confirm that there is “a captive job” for each type of drivers.
Becak drivers, ojek drivers and angkot drivers have significant preferences on “New Ojek
Driver” jobs reflected by 90% or higher confidence intervals and bajaj drivers are
significantly captive to “New Bajaj Driver” jobs indicated by 99% confidence interval. In JA,
EO from the social perspective and SU from the environmental perspective are identified to
play a significant role in job choices when paratransit drivers are facing the various policy
interventions in the future. Meanwhile, their own IC are also proven to affect their job choice
behaviors widely, in which, their geographical feature is regarded as the common significantly
influencing factor for motorized paratransit drivers. The specific explanation for each type of
drivers is then taken as follows.
For becak drivers in Table 5, the statistically significant EO with 99% confidence level
indicates that, as the lowest social-economic group of paratransit drivers, the most important
thing may be to find works for survival, irrespective of job salary (SA), vehicle FT and ES of
that job in future revealed by insignificant parameters of SA, FT and ES, respectively. Becak
drivers also pursue to change the current job for motorized vehicle drivers in order to achieve
a higher economic level, which can be inferred by the 95% significant parameter of operation
cost (OC). Since OC of ojek and bajaj are much higher than the one of becak, the positive
sign of parameter of OC implicitly points out that, how eager they want to change the current
job into the motorized type job with larger OC. The significant ojek-specific parameter
further reveals that this motorized type job is “New Ojek Driver” job that is the captive job in
future for becak drivers. It implicitly confirms their desire to change the current jobs, too.
Moreover, SU with 90% confidence level shows its impact on job choices, which explains
that becak drivers definitely need the financial support to purchase the new vehicle regardless
of FT. Certainly ES, FT and SA also have positive effects on job choices although their effects
are not obvious (significant). As for IC, due to high missing rates such as household income
(HI) with 16.7% and concentrated distributions of some IC such as living districts and marital
status (MS) indicated in Table 1, only age greatly affects job choices indicated by both
alternative-specific parameters. The older becak drivers become, they more tend to choose
“New Ojek Driver” job and are more reluctant to continue the current becak job in future.
Additionally, the marginal reluctance to continue is greater than the marginal preference on
ojek job reflected by magnitudes of both parameters (-0.141 and 0.065). The survey result
reveals that ojek drivers have more freedom in working schedule, less working hours and
enjoy relatively higher salaries among four types of drivers.
1153
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
Table 5 Estimation results of becak drivers’ and ojek drivers’ stated job choice
Explanatory Variable Parameter S.E. T-Statistic Explanatory Variable Parameter S.E. T-Statistic
Alternative Specific Constant Alternative Specific Constant Ojek -0.446 1.227 -0.364 Ojek 0.241 1.228 0.196
Bajaj 0 0 0 Mix of Bajaj and Becak 0 0 0
Current Job 6.120 1.740 3.517 Current Job -4.896 3.058 -1.601
Ojek, Bajaj, Current Job Ojek, Mix of Bajaj and Becak, Current Job Operation Cost 0.243 0.103 2.358 Operation Cost -0.277 0.220 -1.258
Employment Status 0.950 0.780 1.218 Employment Status 0.663 0.542 1.225 (1: union member; 0: self-employed) (1: union member; 0: self-employed)
Employment Opportunity 3.380 0.611 5.530 Employment Opportunity 5.865 1.591 3.686
Salary 0.199 1.882 0.106 Salary 2.889 3.028 0.954
Fuel Type 0.435 0.592 0.735 Fuel Type 0.129 0.489 0.264 (1: electricity for ojek; CNG for bajaj; 0: gasoline) (1: electricity for ojek; CNG for bajaj; 0: gasoline)
Subsidy 10.255 5.709 1.796 Subsidy 5.462 5.407 1.010
Ojek Ojek Age 0.065 0.031 2.103 Marital Status (1: married; 0: single) -0.244 0.516 -0.473
Current Job (Becak Driver) Age 0.009 0.034 0.279
Age -0.141 0.048 -2.921 Household Income (Less than 1 million Rp) -1.346 0.495 -2.719
Theta Living in DKI Jakarta (1: yes; 0: no) 0.567 0.467 1.214
Ojek 0.795 0.256 3.100 Current Job (Ojek Driver) Bajaj (as reference) 0 0 0 Marital Status (1: married; 0: single) -4.383 1.913 -2.290
Current Job 0.000 0.063 0.005 Age 0.358 0.141 2.535
Log-likelihood at Zero Household Income (Less than 1 million Rp) -4.463 1.788 -2.496
Log-likelihood at Convergence Living in DKI Jakarta (1: yes; 0: no) -5.612 2.157 -2.601
McFadden's Rho-squared at Zero Theta Adjusted McFadden's Rho-squared at Zero Ojek 1.183 0.289 4.090
No of Cases Mix of Bajaj and Becak (as reference) 0 0 0
Current Job 0.161 0.065 2.462
Log-likelihood at Zero
Log-likelihood at Convergence
No of Cases
-643.19
-610.66
715
-329.18
0.356
0.461
0.431
671
1154
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
Table 6 Estimation results of bajaj drivers’ and angkot drivers’ stated job choice
Explanatory Variable Parameter S.E. T-Statistic Explanatory Variable Parameter S.E. T-Statistic
Ojek, Bajaj, Current Job Ojek, Medium Bus, Current Job Operation Cost -0.352 0.109 -3.23 Operation Cost -2.733 1.080 -2.53
Employment Status 0.563 0.516 1.09 Employment Status 0.039 0.271 0.15 (1: union member; 0: self-employed) (1: union member/company staff; 0: self-employed)
Employment Opportunity 0.320 0.683 0.47 Employment Opportunity 1.251 0.660 1.90
Salary 0.088 1.848 0.05 Salary -4.273 1.717 -2.49
Subsidy 6.328 3.833 1.65 Fuel Type 0.079 0.210 0.38 (1: electricity for ojek, electricity/CNG for medium bus; 0: gasoline)
Ojek Ojek Fuel Type (1: electricity; 0: gasoline) -0.214 0.466 -0.46 Subsidy -4.224 2.857 -1.48
Mix of Bajaj and Becak, Current Job Education Level (High school and above) -0.474 0.333 -1.42
(Bajaj Driver) Living in DKI Jakarta (1: yes; 0: no) -1.203 0.456 -2.64
Marital Status (1: married; 0: single) 2.572 0.588 4.37 Household Income (Less than 1 million Rp) -1.647 1.110 -1.48
Education Level (secondary school) -1.071 0.339 -3.16 Marital Status (1: married; 0: single) -0.563 0.491 -1.15
Education Level (High school) -2.066 0.439 -4.71 Medium Bus and Current Job Household Income (Less than 1 million Rp) 1.025 0.395 2.59 (Angkot Driver) 0 0 0
Household Income (1 million ~ 2 million Rp) 1.464 0.411 3.56 Theta Living in South DKI Jakarta (1: yes; 0: No) -0.474 0.397 -1.19 Ojek 0.471 0.253 1.859
Living in East DKI Jakarta (1: yes; 0: No) 1.642 0.391 4.20 Medium Bus (as reference) 0 0 0
Fuel Type (1: CNG; 0: gasoline) -0.875 0.651 -1.35 Current Job 0.015 0.017 0.870
Theta Log-likelihood at Zero
Ojek (as reference) 0 0 0 Log-likelihood at Convergence
Mix of Bajaj and Becak 0.517 0.188 2.75 McFadden's Rho-squared at Zero
Current Job 0.000 0.057 0.00 Adjusted McFadden's Rho-squared at Zero
Log-likelihood at Zero No of Cases
Log-likelihood at Convergence
No of Cases
-549.09
-435.75
684
-397.64
0.353
0.327
1155
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
For ojek drivers in Table 5, EO with 99% significant level on job choices can be used to
interpret the seriously fierce competition among ojek drivers and between ojek and other types
of paratransit due to some reasons. First, ojek jobs have no entry limitation in the sense that
anyone who has a motorcycle can do this illegal business. Second, ojek jobs are the captive
jobs for becak drivers and ojek drivers and angkot drivers as indicated by statistically
significant job-specific preference parameters with confidence interval of 99%, 99% and
90% in the respective driver job choice model. OC here is not a significant JA, which can be
attributed to the quite cheap daily OC. Since OC is not important on job choices, the financial
saving from the cheaper OC of ojek with electric power compared with the relatively huge
investment of the purchase of new motorcycles is not attractive at all. So it may be the reason
of insignificant FT. Presumably due to the high share of ojek drivers with ojek ownership, it is
not surprised that SU for buying a new vehicle doesn’t play a important role any more. The
inherent feature of ojek jobs that enjoys more freedom in working schedules and suffers from
serious competitions, may explain that they don’t expect too much on SA and have no interest
in ES. Additionally, IC affect the choice of “Current Job” (continuing the current job in the
future) a lot. The single ojek drivers tend to continue the current job indicated by MS
parameter, maybe because they can’t afford new vehicles or just keep the saving for marriage
in the future. The older ojek drivers become, they more tend to keep using the current vehicle
in the future, indicating the inherent feature “don’t want to change” of the older. Ojek drivers
belonging to the lowest level of HI (here less than 1 million Rp) obviously want to change the
current job in the future, probably due to the dissatisfaction of current paratransit income
level. Geography of residential location is identified as one significant influencing factor on
the choice of current job in the future, showing that ojek drivers living in DKI Jakarta have the
obvious tendency to quit the current job in the future. This could be explained that they may
realize that the living space of the current job is quite limited due to the consistently
improving public transportation system, serious pressure from the government in terms of
punishment of illegal operation of ojek and competition with other transport modes. In
contrary, the new ojek drivers with new power type are promoted in DKI Jakarta, but this
trend is not significant. For “New Ojek Driver job” option, ojek drivers in the lowest income
group show obviously negative attitudes on the current jobs, too.
For bajaj drivers in Table 6, OC, directly determining SA, affects significantly on job choices
with 99% confidence interval. Maybe it is the reason why SA has a little impact on job
choices. Although SU obviously impact on job choices with 90% confidence interval, it seems
that SU still can’t promote the usage of low-emission vehicles with cheaper OC, which can be
reflected by two negative parameters of FT for ojek job option and mix of bajaj and becak job
option (Since the choice result of becak drivers in the future is only 2.5% of mix of bajaj
driver choice and becak driver choice, it could be ignored when explaining.). From this point,
it can be inferred that SU is not enough and/or the saving of OC from vehicle changes can not
compensate for the relative huge price of purchasing the low-emission vehicles (ojek is
electronic power and bajaj is CNG power); Another possible reason could be that bajaj
drivers suspect the practical performance of such vehicles due to the unfamiliarity. If it is
true, more advertisements and detailed information about its advantages (e.g., cheaper OC can
compensate for the vehicle price from the long term) on low-emission vehicles are definitely
needed. With the purpose of examining IC influences, married bajaj drivers have obvious
tendency to choose “New Bajaj Driver” jobs. The education level (EL) also plays a significant
role on this job choice, showing that higher EL is, less desires to choose “New Bajaj Driver”
jobs. This impact is revealed by two parameters with distinct significances (-3.16 for
secondary school level and -4.71 for high school level), comparing with drivers with
1156
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
elementary school level. Comparing with bajaj drivers with highest HI, bajaj drivers with HI
(less than 1 million Rp) and HI (1 ~ 2 million Rp) prefer “New Bajaj Driver” jobs. The
estimation result also manifests bajaj drivers from East Jakarta more likely to choose Mix of
bajaj and becak option (i.e., New Bajaj Driver jobs) shown by the dummy variable (Living in
East Jakarta) with 99% confidence interval due to huge flat roads there and one of main
resource of migration to become bajaj drivers. Statistically significant parameter i proves
that current bajaj drivers obviously would like to have their own bajaj and do this job in the
future, i.e., they are captive to new bajaj job (Mix of Bajaj and Becak in modeling).
For angkot drivers in Table 6, OC is still a great influence factor with 95% confidence
interval. SA, another statistically significant JA, is negative, which indicates angkot drivers’
preference on “New Ojek Driver” jobs (i.e., preferring the job with a little lower salary
comparing with current angkot driver job). The possible explanation is that, currently angkot
drivers without ownership of angkot would like to have their own ojek first so that they can
enjoy the more flexible work style of ojek job at the expense of losing a little salary.
Meanwhile, job-specific parameter for ojek with statistical significance in 90% further
proves that some extent of attractiveness of ojek job to angkot drivers, i.e., ojek driver job is
the captive job for current angkot drivers in the future. EO with 90% significance shows its
positive influence on job choices. The insignificant parameter of FT can be explained partially
by the negative influence of SU on job choices. Since the purchase of the traditional vehicle
with gasoline power in SP design are offered with less subsidies, the negative sign of SU
exactly reveals that they don’t acknowledge the low-emission vehicle, i.e., the never used
electric motorcycle. With respect to IC, the only statistically significant dummy variable
(living in DKI Jakarta) shows that angkot drivers from DKI Jakarta are obviously reluctant to
take ojek drivers as a future job option. It is similar choice behavior with that of ojek drivers in
DKI Jakarta when facing the option of continuing to use the current ojek in the future. It may
be confirmed again that the serious situation of ojek drivers with old type power in the future.
6. CONCLUSION
Recognizing the important roles played by paratransit systems in providing valuable job
opportunities to low-income people and relatively seamless transport services to residents in
developing cities, this study has attempted to comprehensively investigate the employment
issues of paratransit drivers in developing cities by taking Jabodetabek Metropolitan Area,
Indonesia as an example. Four typical paratransit drivers (becak, ojek, bajaj, and angkot
drivers) are targeted. Aiming to create more competitive, attractive and sustainable paratransit
systems in future, this study especially looked at how the current paratransit drivers prefer
different types of new paratransit driver jobs equipped with low-emission vehicles under
different competitive employment circumstances. The objective of this study is to analyze
paratransit drivers’ job choice behavior under various policy interventions, clarify the
influencing factors and further identify whether there is a “captive job” for particular
paratransit drivers. With such considerations, this study conducted a questionnaire survey in
February 2010 by interviewing nearly 800 paratransit drivers of the above four types. In the
survey, drivers were asked to report their actual job situations and stated job choices under the
hypothetically various policy interventions based on a stated preference (SP) survey method.
The SP survey was designed for the current paratransit drivers to report their stated job
choices by incorporating the influences of salary level, employment opportunity, employment
status, operation cost, subsidy for low-emission vehicles by government, vehicle fuel type and
1157
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
so on. As a result, 715, 671, 684 and 612 valid SP profiles for becak, ojek, bajaj angkot
drivers were used for analysis, respectively.
To directly evaluate drivers’ response to job choices in the assumed various policy
interventions, cross-tabulations between factors reflecting both social and environmental
policy interventions and job choices results are conducted, which roughly and qualitatively
reflects the variation of job choices on different levels of JA. Additionally, the dogit model is
applied to overcome the limitations of cross-tabulations and to capture drivers’ job choice
behaviors, where the selected choice model is used to respond to the choice concentration on
some particular job options and identify whether there are some “captive jobs”. The dogit
model can also partly relax the IIA property of the widely applied MNL model for discrete
choice such as job choice here. The estimation results clearly show that dogit models could
well represent such job choice behaviors for different types of paratransit drivers in terms of
goodness-of-fit indicators. It is also revealed that employment opportunity (EO) reflecting the
social perspective of policy intervention and subsidy (SU) for low-emission vehicles by
government reflecting the environmental perspective of policy intervention have widely
different influences on the job choices across four typical types of paratransit drivers.
Meanwhile, significantly different influences of individual characteristics on some particular
job choices infer the inherent different features of each type of paratransit drivers. In addition,
geographical feature of drivers - living districts is identified to play an important role on some
specific job choices for motorized paratransit drivers. Moreover, the existences of “captive
job” are identified especially for ojek driver job and bajaj driver job. For policies related
variables, only subsidy to new vehicles shows significant influences on becak driver and bajaj
driver job choices, and given the current energy subsidies in Indonesia, it seems that fuel type
(FT) almost has no impacts on paratransit drivers. This suggests that the financial incentive is
the most important tool to encourage paratransit drivers to shift to sustainable jobs. In
contrast, the savings from operation cost (OC) of vehicles with much cleaner power seem not
influential to drivers’ future job choices. To the authors’ best knowledge, this is the first time
in the transportation literature to comprehensively examine influential factors affecting the
choices of paratransit driver jobs under social and environmental policy interventions in the
future situations made by the various types of current paratransit drivers. Such analysis results
obtained by deeply understanding job choice behaviors of paratransit drivers could be useful
to policy decisions on transforming the current ill-functioned paratransit systems in
developing cities into more sustainable, more competitive and attractive transportation
systems.
As for the future research, it is still necessary to explore better model structures to more
properly represent paratransit drivers’ job choice behavior, especially considering drivers’
heterogeneous responses under different choice situations and drivers’ actual employment
capabilities. Imputation methods should be applied to resolve the missing data issues related
to some key variables in the supply side of paratransit. Furthermore, promoting a wiser use of
paratransit system in developing cities requires the understanding of trip makers’ travel mode
choice behavior considering the availability of better paratransit modes as both line-haul and
access/egress modes. Integrating the new insights about paratransit systems from both
employment and transport perspectives, it is important to explore how to re-design the
currently ill-functioned transportation systems in developing cities from the comprehensive
viewpoint such as congestion, safety, energy, equity and poverty and so on to achieve the
sustainable public transportation system.
1158
Journal of the Eastern Asia Society for Transportation Studies, Vol.9, 2011
REFERENCES
Azuma, Y. (2000) Socioeconomic changes among Beca Driver in Jakarta, 1988-98, Labour and Management in Development, Vol. 1, No. 6.1-31.
Ben-Akiva M. (1977) Choice models with simple choice set generating process. Working
Paper, Dept. of Civil Engineering, MIT, Cambridge, MA.
Cervero, R. (2000) Informal Transport in the Developing World, United Nations
Commission on Human Settlements, Nairobi, Kenya.
Cervero, R. and Golub, A. (2007) Informal transport: A global perspective, Transport Policy, Vol. 14, No. 6, 445-457.
Diaz, C. E. D. and Cal, P. C. (2005) Impacts of government regulation on the sustainability of
paratransit services in the Philippines: Case of FX services between Manila city and
Quezon city, Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, 214-224.
Etherington, K. and Simon, D. (1996) Paratransit and employment in Phnom Penh: The
dynamics and development potential of cyclo riding, Journal of Transport Geography, Vol. 4, No. 1, 37-53.
Gaudry, M. J. I. and Dagenais, M. G. (1979) The dogit model, Transportation Research Part B, Vol. 13, 105-111.
Japan International Cooperation Agency (JICA). (2004) The study on integrated transportation
master plan for Jabodetabek (Phase 2).
Joewono, T. and Kubota, H. (2007a) Exploring negative experiences and user loyalty in
paratransit, Transportation Research Record, No. 2034, 134-142.
Joewono, T. B. and Kubota, H. (2005) The characteristics of paratransit and non-motorized
transport in Bandung, Indonesia, Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, 262-277.
Joewono, T. B. and Kubota, H. (2007b) The multigroup analysis regarding user perception of
paratransit service, Journal of the Eastern Asia Society for Transportation Studies, Vol. 7, 1651-1663.
Joewono, T. B. and Kubota, H. (2007c) User Perceptions of private paratransit operation in
Indonesia, Journal of Public Transportation, Vol. 10, No. 4. 99-118
Schalekamp, H., Mfinanga, D., Wilkinson, P. and Behrens, R. (2009) An international review
of paratransit regulation and integration experiences: Lessons for public transport system
rationalisation and improvement in African cities. Proceedings of the 28th Annual Southern African Conference, Pretoria, South Africa.
Shimazaki, T. and Rahman, M. M. (1996) Physical characteristics of paratransit in developing
countries of Asia, Journal of Advanced Transportation, Vol. 30, No. 2, 5-24.
Swait, J. D. and M, Ben-Akiva. (1987) Empirical test of a constrained choice discrete model,
mode choice in Sao Paulo, Brazil. Transportation Research Part B, Vol. 21, No. 2, 103-
115.
Swait, J. D. and M, Ben-Akiva. (1987) Incorporating random constraints in discrete models of
choice set generation. Transportation Research Part B, Vol. 21, No. 2, 91-102.
Tarigan, A. K. M., Susilo, Y. O. and Joewono, T. B. (2010) Negative experiences and
wilingness to use paratransit in Banung, Indonesia: An exploration with ordered probit
model, Compendium of Papers CD-ROM, the 89th Annual Meeting of the Transportation Research Board, Washington, D.C., January 10-14.
1159