Operational Characteristics of Paratransit Services with Ride-Hailing Apps in Asian Developing Cities: The Phnom Penh Case

Rapid adoption of ride-hailing apps (RHAs) has greatly influenced the way people travel—there is no exception for paratransit users. However, it remains unclear whether RHAs would be regarded as threats or opportunities among paratransit operators in Asian developing cities. While RHAs have been viewed as disruptive transportation, several studies explored the threats of RHAs on taxi industry—but only a few examined such threats on other paratransit services (e.g., auto-rickshaws). This study assessed the changes in the operational services among paratransit operators who have adopted RHAs. The changes were examined by statistical comparisons using data collected from questionnaire survey with 182 Bajaj drivers in Phnom Penh, January 23-27, 2018, as a case study. Results showed that majority of the interviewed drivers started new services with RHAs less than a year ago—they were younger (<40 years old), married, and higher education. After adopting RHAs, the drivers could increase their revenue by increasing the number of trips and customers. They could further achieve a higher revenue by operating with multiple RHAs. Most drivers (>88%) satisfied with RHAs and acknowledged improvements on their operational services. The results suggested that RHAs would be opportunities for those paratransit drivers who have adopted them, while they would be threats for those who have not. The collected data serve as useful inputs for future public transport planning in Asian developing cities.

. The cumulative number of ride-hailing apps (in parentheses) by years in some Asian countries. The first app name is also shown in the parentheses (Source: https://play.google.com).
RHAs, in July 2016, China established new regulations to govern all for-hire services. The new Chinese regulations provide basic principles for local governments to create innovative and flexible (de-)regulation on supply quantity, fare, labor protection, and market entry [10]. In the Philippines, RHS has not yet been considered as a disruptive innovation-that is, the traditional taxi industry can still compete with RHAs [3]. While regular taxi operators continue to protest against RHAs, they fail to realize and prioritize the needs faced by both customers and drivers. The Philippines has viewed RHAs as safer and more convenient commuting options for citizens, and established legal framework to regulate RHS and to tackle with the increasing urban mobility. In Thailand, there have been conflicts between ride-hailing firms (RHFs) and paratransit operators [11].
In May 2016, Department of Land Transport (DLT) banned UberMoto and GrabBike for violations of existing public transport regulations including license and fare. In July 2016, GoBike was launched as the Thailand's first official and fully legal RHA for motorcycle taxis, that complied with the strict regulations set by the DLT. Later, the RHFs conformed the national regulations and pursued strategies to attract more drivers. For example, Grab Thailand introduced a financial incentive (i.e., 1000 THB) campaign to draw more motorcycle taxi drivers into GrabBike. As Grab did not have enough drivers with motorcycles registered as public vehicle under Thai laws (i.e., yellow-plated license), Grab also recruited drivers with unregistered motorcycles (i.e., white-plated license) for providing RHS. The new business model of RHAs, however, is very different from the traditional motorcycle taxi services-which are mostly organized as groups (know as Win in Thai) and operated within pre-defined servicing areas [12].
Since RHA drivers could pick up a customer anywhere, this has created conflicts when Win members realized that RHA drivers picked up customers from their Many studies assessed the impacts of RHAs on existing transport services [9] [13] [14]. The studies can be categorized into "rule change" from the viewpoint of market regulation, "competitive effects" from an industry standpoint, and "labor changes" from the stakeholder perspective [10]. The impact assessment is quite important for successfully designing appropriate regulations for RHS particularly in Asian developing cities in the long run. While several RHA studies concern taxi industry [3] [4] [13] [15] [16], only few concern other paratransit modes such as motorcycle taxis and auto-rickshaws [17]. There was no mass transit system, and the available public transport modes at that time were mainly paratransit. Following the rapid urbanization, the traffic The current urban public transport modes include public bus and paratransit.
The bus is the only available mass transit system, which has been introduced since 2014. From September 2017, the bus services have been expanded to ten routes (total length of 148 km), covering several parts of the city [19]. Since the bus is operated in mixed traffic, its operating speed is relatively slow (approximately 10 km/h) and the expected waiting time at a bus stop is uncertain. In addition, 40.0% of the bus users appear to be free riders [20], which leads to concern over overall revenue of bus operations. However, the bus service is still rather new and seems to have little appeal to general citizens because many still prefer door-to-door trips. This also includes the poor accessibility (e.g., sidewalk,

Ride-Hailing Service
In Cambodia, we conducted two online searches in Google Play website to see  Among local RHFs, some RHAs become more popular (e.g., PassApp, iTsumo, and WeGo) while some RHAs become less popular or disappeared (e.g., Exnet, CamGo, Natra), as evidenced by the changes in the number of android installs. Each RHA has been developed separately for drivers and general users. In particular, PassApp has become the most popular paratransit RHAs-the number of its installs by users and drivers increased by ten times within the observed period, indicating high acceptance among general citizens and paratransit drivers in Phnom Penh. PassApp has also planned to expand its services for Motodop and to other provinces (e.g., Siem Reap). Grab has already expanded to Siem Reap. Moreover, we observed that most RHFs require the registered paratransit drivers to attach an RHA logo sticker on their vehicles or wear a uniform, indicating the RHF with which they operate.

Data
We use questionnaire survey to collect data about the operational characteristics of paratransit drivers, who have operated with RHAs in Phnom Penh. The questionnaire was first written in English and was later translated into Khmer. We then verified its consistency between the English and Khmer versions in a pilot test. The questionnaire includes five parts. Part 1 inquired about RHAs in general. Part 2 inquired about the changes in the operational services after adopting RHAs. Part 3 inquired about the behaviors of paratransit drivers when waiting for customers booking via RHAs. Part 4 inquired about subjective responses towards RHAs. The subjective questions were based on a 5-point scale (1: very unlikely, 2: unlikely, 3: neither, 4: likely, 5: very likely). And part 5 inquired about personal information.
Four surveyors were trained to fully understand and administered the questionnaire. There were no specific interview locations had been pre-defined. Paratransit drivers operated with RHAs were acquired by three different methods. First, surveyors sought for RHA drivers on city streets by a visual inspection. They approached the drivers whose vehicles were attached with RHA logos. Second, surveyors visited RHA companies. There, the drivers who already registered to RHAs came to pay for the commission fee or have their vehicles maintained. And third, surveyors made a booking via RHAs, calling drivers to a designated location. The shortest trip-distance was selected for minimum fare. Upon arrival, the drivers were queried whether they would participate in the survey. Yet, only four drivers were recruited using this method.  The surveyors did not target all paratransit drivers with RHAs they spotted. They first observed if a driver was free and then verbally confirmed if he was willing to participate in the survey. Between January 23 and 27, 2018, surveyors requested approximately 300 drivers, of which 197 voluntarily participated in the survey. The drivers refused to join the survey because they were waiting for customers, tired, busy on phone or having lunch, and not in a good mood.
It took the average from 10 min to 15 min for each respondent to answer the questionnaire. Most respondents were Bajaj drivers and only three were Motodop drivers. Respondents were recruited with a financial incentive (i.e., 1 USD) for their time spent in the survey. About ten respondents refused to accept the incentive because they humbly wanted to contribute to this study. After the in-Journal of Transportation Technologies formation screening process, only data from 182 Bajaj drivers were with effective information. Table 2 reports the descriptive characteristics of interviewed Bajaj drivers with RHAs. Majority were younger than 40 years old (80.2%) and married (86.7%).

Driver Characteristics
Regarding the highest education attainment, majority finished at least secondary school (85.3%). This proportion is relatively higher than that (69.7%) reported in [23], suggesting that Bajaj drivers with RHAs had higher education than the traditional paratransit drivers (i.e., Motodops and Remorks). Some 8.8% had dual jobs-they were not only the drivers of Bajaj, but also government officer, farmer, worker, or teacher. RHAs seem to create jobs that attracted people with various professional backgrounds-including employees (15.4%), drivers (14.8%), skilled workers (11.0%), and sellers (6.0%). The period of working as Bajaj drivers with RHAs ranged from one week up to 30 years, with the average of 22.4 months. Majority were new drivers; whose working period was less than a year (78.0%). Minority had a driving license (29.1%) and a vehicle insurance (6.0%).
6.6% claimed that they were also members of two labor associations-that is,

Cambodia for Confederation Development Association (2.2%) and Independent
Democracy of Informal Economy Association (4.4%). These are the two largest labor associations in Cambodia [24].
The period that drivers had been using a smartphone ranged from one week to 13 years, with the average of 2.7 years. There were two types of smartphones' operating systems: android (86.5%) and iOS (13.5%). Android phones were more popular because their prices-mostly second-hand devices-were relatively cheaper than an iOS device (i.e., iPhone or iPad). With mobile internet service, the drivers could operate with RHAs. Majority registered to one RHA (77.5%)-and without a strict internal rule set by RHFs, some drivers registered to 2 -4 RHAs (22.5%). The average period since the drivers registered to RHAs ranges from 19 days to 5.6 months. The most recent registration was two days, while the longest period was up to three years ago. Each driver who registered to a RHF is required to pay a commission fee. The commission fee refers to the percentage of the total fare collected from a customer upon consuming paratransit service via the firm's RHA. The reported commission fees ranged from 0% to 20%, with the average of 9.2%. Although each RHF has a set of internal rules like the commission fee, the drivers seem to understand the fee in different way. As such, they reported   ). This is more practical when considering the current working conditions of informal drivers in the city. As most drivers were poor and had no stable income, they had no obligation to have an official bank saving account. In line with this situation, many drivers needed to visit RHA offices to refill their temporary accounts. Further, respondents were requested to describe three major reasons why they registered to an RHF-175 of them reported 227 reasons. Mixed reasons were received and were classified into three categories, namely to get more passengers (75.8%), to get higher revenues (11.5%), and other reasons (12.8%) including modernization and no need a pick-up station. Similarly, respondents were requested to describe three good and three bad points they experienced with the RHAs registered. 152 drivers reported 219 good points and 62 drivers reported 79 bad points-their mixed responses were categorized as in Figure 2 (left) and Figure 2 (right), respectively. For the good points, majority of the drivers claimed that RHAs facilitated them in getting more customers and higher revenue (79.9%). Some also reported that RHAs helped them to save time (5.9%) and to avoid fare negotiation (4.6%). Other good points include safety and maps/GPS navigation (9.6%). For the bad points, the drivers claimed high commission fee charged by RHFs (38.0%), the app errors (34.2%), and the issues with customers (15.2%). The other bad points include the strict internal rules set by the app companies (12.7%). The app errors include internet and system errors. The customer issues include the trip cancelation and complaints. RHFs might want to improve these bad points in order to minimize negative image on their RHS. For example, RHFs should improve their RHA features to minimize errors, cooperate with high-speed internet company to support their RHA drivers, and encourage the drivers with special incentives or discounts. Minimal errors within the RHAs would also help reduce customer issues and maintain the reputation of RHFs.

Operational Services
In part 2 of the questionnaire, respondents were requested to report their operational services before and after adopting RHAs. Their responses were summarized in Table 4 Approximately 40.0% of the interviewed drivers reported their operational services both before and after adopting RHAs, suggesting that they continued to serve paratransit services by embracing advance technology (i.e., RHAs). Their operational services were summarized separately under "Before" and "After", respectively. Two types of group mean comparison tests were performed to examine changes in the operational services before and after adopting RHAs. One is the comparison between "Before" and "After (overall)" using unpaired t-test with a degree of freedom (D.F.). The other is the comparison between "Before" and "After" using paired t-test. All t-tests are significant (p < 0.01), except that for daily working hours between "Before" and "After (overall)" which is not significant (p > 0.05).
We now examine the actual changes in the operational characteristics for spe-  To see whether there are differences in the operational services for drivers operated with multiple RHAs, we performed additional comparison tests in Table   5. We first divided the respondents into two groups: one operated with one RHA (77.5%) and the other operated with 2 -4 RHAs (22.5%). We then examined the V. K. Phun et al. RHFs (e.g., PassApp) have specific rule not to allow their driver members registering to the other RHAs. In the current situation, however, this rule is not so strict and the drivers could register to multiple RHAs, which is a good strategy for them to enjoy a higher revenue.
We also conducted more statistical tests to see whether there were differences in the operational services with respects to drivers' socio-demographic characteristics, as shown in Table 2-that is, the groups of age, marital status, education, and period of working as a paratransit driver. One-way analysis of variance was applied for comparison of more than two groups, but it does not tell which group is more superior than the other groups.

Driver Behaviors
This section explores the common behaviors of paratransit drivers with RHAs. Journal of Transportation Technologies First, respondents were requested to report the high demand hours, during which they experienced more RHA customers. Figure 3 shows the hourly share of demand for RHS. Majority of drivers reported that they could get more RHA customers during the peak hours from 7:00 to 8:00 (68.1% -74.7%) and from 16:00 to 18:00 (71.4% -87.9%). With these high demand hours in mind, the drivers would work smarter and be ready for their RHA customers.
In addition, the drivers were asked to describe three of their favorite standby locations during low demand hours. 249 locations were reported and categorized as in Figure 4 (left). Majority (89.9%) liked to standby on-streets and near public places (e.g., residential areas, markets, hotels, restaurants, schools, and bus terminals). Basically, the drivers operated with RHAs require no specific standby stations. The drivers asserted that, after dropping off a customer, they looked for a vacant space with shadow along the streets to wait for the next customers booking via RHAs. When they were not in the working mood, they would simply turn off RHAs on their smartphones for not accepting customers. Besides, 180 drivers reported 283 favorite activities at the standby locations. These activities were summarized in Figure 4 (right). Majority (64.3%) involved with leisure activities via smartphones (e.g., Facebook, music, movies, game), 12.4% would wait for the next customers. Other activities include reading (7.8%), chitchatting with friends (6.0%), resting (5.3%), having coffee (2.5%), and others (1.8%). The additional behaviors of paratransit drivers with RHAs are reported in Table 6. In average, the drivers spent 6.7 min to go picking up a customer upon they accepted his/her booking via RHAs. About half (51.7%) took less than 5 min, while 40.1% took 5 -10 min to pick up RHA customers. This could be as fast as 1 min depending on the level of traffic congestion as well as the customers' origins. The drivers further reported that they would carefully accept RHA customers whose pick-up locations were far from them (e.g. 5 -10 min), especially during the congested hours. This is because they wanted to avoid customers from waiting longer time, which could possibly lead to complaints and cancellations of the booking trips. The drivers were also aware of the punishments set by RHFs for whom frequently cancel the trips booked by customers. Experienced drivers would know this situation well and chose to react accordingly. When asking how many times they failed to locate the booking customers, 14.8% reported that they never missed their customers, 35.2% lost from one to  7h 8h 9h 10h 11h 12h 13h 14h 15h 16h 17h 18h 19h 20h 21h 22h 23h Time of the day Journal of Transportation Technologies  five, and 50.0% lost more than six customers per month. The failures to locate customers were caused by several factors including the issues within RHAs, slow internet, errors in digital map and GPS, and the matters with customers (e.g., waiting at an inconvenient place where drivers were unable to pinpoint). When there were no customers, the drivers checked their RHAs about once in average, and up to 15 times per hour. About half (53.3%) did not check RHAs, 42.8% checked from once to 3 times, and 3.9% checked more than 3 times per hour. Usually, RHAs would alert the drivers when there were incoming booking customers. However, they may come to check the apps because they wanted to review their working history and updates, including incentive benefits and balance of deposit account. Although RHAs assisted the drivers to get customers, 35.2% still tried to cruise for additional passengers along the streets. They cruised 6.5 times and spent 42.5 min per day in average looking for additional customers. On daily basis, majority of them spent from 1 min to 45 min (74.7%) and cruised from once to 6 times (73.3%). In addition, the drivers explained up to three main reasons why they decided to cruise for additional customers. Eighty-three drivers reported 108 reasons, which were categorized as: because they wanted to get more customers (53.7%), they wanted to increase daily revenue (35.2%), and there were errors within RHAs, including internet and system errors (11.1%). When asking whether they wanted a proper standby station waiting RHA customers, 72.0% gave positive responses. Such standby station could help reduce the number of paratransit drivers idling on-streets, and thus traffic congestion on the city streets to some extents, especially during peak hours.

Operating Speed and Routes
To collect data regarding the operating speed and routes of paratransit drivers with RHAs, we first requested about 30 drivers to attach each of their vehicles with a portable GPS device (GPS Route Logger GT-730FL-S, Canmore Electronics Co., Ltd., Hsinchu, Taiwan). This device records time and geographical coordinates (i.e., latitude, longitude, and altitude), and speed for every five seconds. Eleven drivers agreed to have their vehicle attached with the GPS device, each for a single day 9:00 -22:00, January 26-27, 2018. Other drivers refused to GPS attachments because they concerned with violations of the rules in ride-haling firms, they were busy, and they simply rejected without offer any reasons. Table  7 gives the summary statistics of the collected GPS data. The total GPS-measured time ranged from 6.4 hours to 12.0 hours, with the average of 8.6 hours. The servicing distance ranged from 36.1 km to 120.5 km, with the average of 79.4 km. The total moving time ranged from 2.1 hours to 6.4 hours, with the average of 4.2 hours. The average operational speed was measured as slow as 4.0 km/h and as fast as 12.1 km/h. The average maximum speed of Bajajs is 46.3 km/h, which is relatively faster than that (35.7 km/h) of traditional Remorks [23]. The maximum speed could be as high as 60.0 km/h, almost equivalent to the highest speed specification indicated from Bajaj manufacturer. Figure  5 illustrates the operating routes of Bajaj drivers with RHAs in Phnom Penh. While fewer trips seem to be scattered in the suburban areas, more trips concentrated in the downtown area, penetrating through both major and small roads. This may also suggest that majority of the Bajaj trips via RHAs were short-distance trips within the downtown area.

Subjective Responses
Respondents evaluated seven subjective questions related to their satisfaction and future anticipation towards RHAs, based on the 5-point scale (1: very unlikely, 2: unlikely, 3: neither, 4: likely, 5: very likely). Their evaluation scores were reported in Figure 6. Most drivers (96.7%) rated high scores (scored of 4 or 5), Journal of Transportation Technologies  number of paratransit drivers operated with RHAs. They (69.8%) also thought that it would be better to have proper standby stations near bus stops or railway stations in the city.

Discussion
Among the interviewed Bajaj drivers in Phnom Penh, about 40.0% were identified as former paratransit drivers who have upgraded their services by adopting RHAs. Adaptation is the key to survival. These drivers might have anticipated changes in the demand for paratransit services following market penetration of RHSs-and thus, they have gradually upgraded their vehicles (e.g., from Motodops/Remorks to Bajajs) and adopted RHAs in order to improve their operational services and livelihood. The improvement was well supported by the data in this study. As about 40% were former paratransit drivers, this also means that about 60.0% were new drivers who came into paratransit market directly with RHAs. This proportion justifies that, although RHAs may affect traditional paratransit operators, RHAs have indeed generated extra jobs into labor market, especially for many low-income or unskilled people. However, new drivers may also include less experienced or less professional individuals to serving paratransit services, which in turns raises questions on their operational safety. If the traditional paratransit operators do not embrace new technology, following recent changes in the paratransit market, they may eventually lose their market to RHA drivers. One observation is that, the arrival of Bajajs with RHAs have lessened the number of Remorks on the city streets. While RHA drivers required no specific pick-up stations, traditional Remorks seemed to spend more time waiting their customers at the secured pick-up locations such as near the airport, hotels, markets, and bus stops/terminals. Other potential customers for traditional Remorks include those who did not use a smartphone and loyal customers (e.g., relatives, friends, and neighbors). Another observation is that only a small number of Remork drivers registered to RHAs. Based on opinions of the interviewed drivers, operating Remorks with RHAs would not be much attractive among RHA users because Remorks consume more gasoline, which would cost higher comparing to Bajajs that consume cheaper liquefied petroleum gas.
In line with this situation, the fare structure for Remorks has been defined to be relatively higher than that of Bajajs. For example, the fare of PassApp began at 3000 and 4000 KHR (0.75 and 1.0 USD) with the rate of 1200 and 1300 KHR (0.3 and 0.325 USD) per subsequent kilometer for Bajajs and Remorks, respectively. The fare was distance-based, but not time-based. When Remork drivers wish to operate with RHAs, they may also want to upgrade their vehicles-e.g., from Remork to Bajaj-in order to be more profitable via lower operating cost. Although there was news reported that the demand for traditional paratransit services have been squeezed by RHA drivers, no protests against RHAs have been seen so far. Even there would be protests, as these paratransit drivers have been considered as informal transport operators; however, they may receive little attention from the government. One of the best strategies for the impacted traditional paratransit drivers might be that they should adapt themselves in the era of advance technology-e.g., by offering their customers better transport services via RHAs. As the popularity of RHAs continues to arise, the government may want to consider minimum regulations (e.g., standardized fare structure, safety regulations, and labor protection), which shall benefit all relevant stakeholders, including general citizens and paratransit operators.

Conclusions
This study assessed changes in the operational services among paratransit operators with RHAs in Asian developing cities. The changes were assessed via data from interview survey with 182 Bajaj drivers who operated with RHAs in Phnom Penh, as the case study. Results showed that approximately 40.0% of the drivers-who have upgraded their services by adopting RHAs-could increase the number of their daily trips and customers by about 63.0%. They could also increase their monthly revenue by 45.5%. Those who operated with multiple RHAs could further increase their monthly revenue by 14.8%, compared with those who operated with single RHA. In sum, most drivers (>88.0%) satisfied with RHAs and acknowledged that these RHAs helped improve their overall paratransit services as well as their living conditions. The combination of Bajajs and RHAs (i.e., LAMAT) is an example of advanced utilization of old-fashioned paratransit modes in Asian developing countries. Such combination has attracted not only traditional paratransit operators (i.e., the case study showed 40.0%), but also new drivers (60.0%) to provide paratransit services with RHAs. Results from the case study implied that RHAs would be opportunities for those paratransit drivers who have adopted them, while they would be threats for those who have not. Despite concerns over the impacts of RHAs on traditional paratransit operators, RHAs might help improve urban mobility, add alternative transport options, and reduce traffic congestion via reduction in private vehicles. Since RHAs appear to have more benefits than drawbacks, traditional paratransit operators should adapt themselves to new technology-so that, they could meet the transport needs of citizens in modern Journal of Transportation Technologies society. Otherwise, traditional paratransit operators would continue to lose their customers, following higher levels of RHA market penetration. This is surely the choice of customers, who usually prefer more affordable and better transport services. Future study should focus on users' perspective, regarding their preferences and negative experiences with RHAs. Analyzing the effects of RHAs on mass transit ridership could also be useful input for policymakers to consider appropriate and flexible regulations for RHS in Asian developing cities, where there is lack of mass transit system and paratransit remains an indispensable mode.