As previously explained, the role of Yogyakarta as a business and recreational city attracts people living in its surroundings, such as Solo and Kutoarjo. Solo is located on the eastern side of Yogyakarta with the distance of approximately 63 kilometers, while Kutoarjo is located on the western side of Yogyakarta with the distance of nearly 72 kilometers. To support the connectivity among those cities, there is a line of railway system operated since 1994 called as Prambanan Ekspres. The frequency of its service is ten times in a day. While the ticket fare is IDR 8000 or approximately 0.62 USD (flat tariff) and it is a single trip ticket only. There are 18 railway stations along Kutoarjo, Yogyakarta, and Solo. Three stations are located in Yogyakarta: Maguwo, Lempuyangan, and Tugu Railway Station.

In Yogyakarta, there are five public transport modes that can be chosen by commuters for their egress trip: bus, para-transit, taxi, motorcycle taxi, and non-motorized taxi (e.g., trishaw and horse carriage). Commuters who choose bus for their egress mode have to walk 15 meters, 352 meters, and 197 meters to reach bus stop in Maguwo, Tugu, and Lempuyangan railway station, respectively. Ticket fare is IDR 3600 (0.28 USD) for bus mode and IDR 4000 (0.31 USD) for para-transit mode (flat tariff).

Nevertheless, some commuters choose to be picked up by their family members or colleague. Also, some of them opt to park their private vehicle (e.g., motorcycle) at the destination railway station. There are three parking area alternatives for car and motorcycle in Lempuyangan railway station. The firstparking area is located in the front of the railway station (on street parking). This parking area has been managed by local people with the permission from the Yogyakarta Transport Agency. Nonetheless, it does not serve overnight parking. This parking area can accommodate approximately 50 units of motorcycles and several cars. The parking fee to use these facilities is IDR 2,000 or 0.15 USD (flat parking fee). The second option is located at people’s houses. This parking area is used for motorcycle parking only. There are approximately 12 parking areas along Lempuyangan Street which are located in settlement areas at the opposite of the railway station. The parking service hours are from 05.00 to 22.00. The fee varies and usually depends on the distance from the parking areas to the railway station. However, the fee is IDR 3,000 (0.23 USD) for daily or overnight parking, and approximately IDR 50,000 (3.85 USD) a month for regular customers. These areas can accommodate approximately 50-300 units of motorcycle. The third option is the official parking area within the station. This parking area has been managed by the third party appointed by the railway company. It can accommodate approximately 250 units of motorcycle and 20 cars. Unlike other parking area, passengers that park their vehicles in these areas are imposed with a flat fee of IDR 2,000 (0.15 USD). However, if their parking duration is more than 8 hours, the parking fee will rise to IDR 8,000 (0.62 USD).

There are two parking facilities located at Tugu Railway Station. The first one is located on the east side of the station, and the second one is on the west side. The east parking area is only intended for motorcycle mode. In order to reach the station, precisely at the east entrance of the station, the passengers must walk along 324 meters from the parking area. The fee for using the facility is a progressive tariff, IDR 2,000 (0.15 USD) and IDR 4,000 (0.31 USD) for first two hours for motorcycles and cars, respectively. For the next one hour, the passengers are imposed with fees of IDR 1,000 (0.08 USD) and IDR 2,000 (0.15 USD) for motorcycles and cars, as well as maximally IDR 8,000 (0.62 USD) and IDR 16,000 (1.23 USD) for non-overnight and overnight parking, respectively. Meanwhile, the west parking area can accommodate 450 units of motorcycle and 170 cars. To reach a southern entrance of the station, the passengers must walk for 365 meters. The fees imposed are the same with those imposed in the east area of the station.

The last station is the Maguwo Railway Station. Maguwo railway station is a railway station integrated with the airport. This station does not have the parking areas so the passengers of the train must park their vehicles in the airport parking areas. The parking areas for motorcycles and cars are located approximately 210 meters from the station. To use the facility, the passengers are imposed with a fee of IDR 3,000 (0.23 USD) for daily park and IDR 5,000 (0.38 USD) for overnight parking.

The questionnaire items were divided into five components. The first component is personal characteristics such as gender, age, education level, monthly income, driving license ownership, and occupation. The second component is household characteristics, such as the presence of colleague or family member who can pick up at the station, the availability of private vehicle in a household or workplace. The third is trip characteristics, such as trip purpose and frequency. The fourth component is a detailed information on bus attributes, such as trip chain, distance, travel mode, travel time, waiting time and location of departure and alighting if the respondents using bus for egress trip. The final component is a stated preference related to the desire to use bus mode in their egress trip.

Regarding the stated preference survey, the aim is to understand the passenger preferences for particular attributes of public transport that affect their mode choice decision. Since this paper focuses on increasing the demand for public transport on egress trip, the determined dependent variables only consist of two means: public transport mode (bus and para-transit) and motorcycle mode. The reason why only motorcycle mode involved in stated preference survey is related to our preliminary survey that is showing the very low demand of train passengers who choose the car, walking, and non-motorized transport mode for egress trip. This is also in line with the survey result shown in Fig. (2) indicating that there are 85.38% of train passengers who use public transport and motorcycle, and the rest (14.62%) are using the car, walking, car taxi, and non-motorized mode. However, there is also a ‘no choice’ option if the respondents do not prefer to choose both public transport and motorcycle mode.

For independent variables, six attributes considered influencing mode choice for egress trip consist of walking distance, egress travel time, waiting time, a number of transfer, trip cost, and parking fee at the railway station. The selected attributes refer to the existing literature of mode choice decision on egress trip. For example, Polydoropoulou and Ben-Akiva [20] show that parking fee, walk access time, the number of transfer and transit fare are the dominant factors in selecting travel mode on access and main mode. Meanwhile, by using latent class nested logit model, Wen et al. [21] had been analyzing the high-speed rail access mode choice in Taiwan. The result showed that access cost, access time, parking fee, and waiting time have the significant impact on the access mode choice. Kim et al. [22] explored several factors that influence the access/egress mode choice to/from light rail stations. Three alternative modes were considered in the model: bus, walk and drop off/picked up. They found that bus availability, park and ride lot, and distance have significant correlation influencing access/egress mode choice decision. Chaucan et al. [23] also found that ingress distance to train station has a significant impact on mode shifting to Delhi Metro in India.

We offered eight scenarios to our respondents. Those scenarios were determined by adopting the orthogonal planning method (for detail see [24, 25]). Initially, the number of scenarios were determined by using a full factorial design method based on the number of independent variables and their levels. By using M (i.e., mode choices) of 2, A (i.e., attributes or independent variables) used was 6, and the maximum L (i.e., level) of attributes was 4 (i.e. the overnight parking fee), scenario combinations. However, the number of scenario combinations was impossible to accommodate. Therefore, the reduction is necessary to do. The formulation of the alternative hypothesis was carried out by using the concept of orthogonal design so that the number of scenarios became (L–1) x M.A + 1. The minimum number of scenario combinations based on the resultant orthogonal design was (4-1) x (2x6) + 1 = 37 alternative scenario combinations. Conversely, 37 alternative combinations are still too much to offer to respondents. Consequently, eight scenarios were determined to be used by eliminating other scenarios with less good combinations of attributes. The scenarios used are shown in Table 1.

The analysis of egress travel mode choice is performed using a discrete choice modeling approach. Since this study focuses on several attributes that can be managed to increase the demand of bus mode on egress trip, a binomial logit model is used to compare between bus mode and motorcycle mode. Due to this state, we neglect other characteristics, such as socio-demographical factors. The data used in our model is obtained from the stated preference survey.

The binomial logit model assumes that a commuter selects a choice having the highest utility value, in which the utility value by using mode m for traveler n can be expressed as follows.

(1)

Where β.X_nm is observed as variable and ε_nm is the unobservable part. The probability of commuter n selects (m) motorcycle mode out of bus mode (b) can be formulated as [26]:

(2)

The coefficients of β parameters in this model are estimated using the method of maximum likelihood function as shown by:

(3)

Given the fact that L is the product of many terms, it is clearly easier to maximize its logarithm of L, rather than L itself. This is because the function of ln is strictly monotone; therefore, maximizing the logarithm of a function f(x) is equivalent to maximizing the function.

(4)

in which θ_mn is equal to 1 if commuter n has chosen mode m and 0 if chosen otherwise.

A face to face interview was conducted from early August to mid of September 2015. Passengers were requested to fill out a questionnaire, which took approximately 20 minutes to complete. More specifically, the questionnaire survey was distributed from Monday to Friday during peak hours (06.00–10.00 and 16.00–20.00). There are initial restrictions for the respondent. First, passengers must get off at Maguwo, Lempuyangan, and Tugu Railway Station. Second, the frequency of using Prambanan Ekspres Train is not less than four times in a month and preferably for the passenger who uses the train for daily travel. Third, the duration of this activity is one-day on destination place (respondents depart and return home on the same day). During the period of the survey, only 154 samples were obtained because surveyors faced the difficulty in identifying the qualified respondents, such as because of the jostle of passengers, while surveyors had to ask the passenger whether or not they satisfy the initial requirements as respondent. Many travelers also disagreed to participate in the study. In the preliminary survey, the questionnaire form was not only distributed in the train but also in the railway station. However, since passengers had a limited time at the transfer point (from destination railway station to their final destination place), they tend to be more reluctant to participate in this survey. Of the 154 collected questionnaires, only 130 data can be analyzed due to incomplete value and have no choices in the stated choice experiment. It should be noted that the average number of Prambanan Ekspres passenger in 2014 is 4,630 travelers/day [27]. However, there are only 27% who commute within the same day [11]. For this reason, a number of 130 respondents could express 10.4% of train users with one-day activity duration on destination place.

Table 2 illustrates the descriptive statistics on the demographic profile of the commuters using Prambanan Ekspres Train. As previously explained, of the 154 respondents surveyed, only 130 respondents (82.28%) were analyzed. Since there are 8 scenarios asked to respondents, a number of 1040 observations (130 respondents x 8 scenarios) are obtained to be analyzed. However, there are only 889 observations modeled in this study due to some missing data.

From Table 2, it can be seen that male was more dominant with 61%. Respondents were also dominated by age group of 22-33 years old (43.08%) and followed by 34-45 years old (31.54%). Looking into income variable, 58.93% of the commuters had a monthly income ranging from IDR 2,600,000 to IDR 6,000,000 (200 USD to 462 USD), while there are 24.11% of respondents with income more than IDR 6,000,000 a month.

Fig. (1) shows the activity end destination of respondents related to the travel mode chosen. While Fig. (2) reports the frequencies for the departure station and egress mode chosen by the commuters. The choice of destination stations in Yogyakarta was closely related to egress trip mode used by the commuters to continue their travel. Approximately, 70% of commuters chose Lempuyangan Railway Station as a destination station in Yogyakarta. Meanwhile, the other two stations (Tugu and Maguwo Railway Station) were chosen by 17% and 13%, respectively, of commuters as a destination station.

Fig. (1). Respondents’ activity end destination and their egress mode in each railway station.

Almost all commuters chose motorcycle as their egress mode alight in Lempuyangan Railway Station. Perhaps, this is due to the overnight parking fee particularly in the parking areas managed by local people, was relatively inexpensive. Travel modes used by commuters alighting in Lempuyangan Railway Station also more varied compared to other stations. There were also walking mode, and riding trishaw selected by commuters indicate that the sites of activity are near to the station. On the contrary, commuters who opt bus mode for their egress trip have only accessed Maguwo and Tugu Railway Station.

Vehicle availability is one of the most important factors that determine the choice of egress mode as it encourages people to use private vehicle compared to public transportation. Fig. (3) depicts a relationship between vehicle availability in Yogyakarta as the destination location and the mode used for egress trip. Vehicle availability refers to not only privately owned vehicle but also the official vehicle that can be used by respondents for activity.

The majority of respondents (64%) have private vehicles in Yogyakarta with the ownership of 53%, 7%, and 4% for motorcycles, cars, and both motorcycles and cars, respectively. The motorcycle which owned by respondents tended to use it as the egress mode. However, this does not occur among the owners of cars; the later tends to choose other modes. Probably, the parking areas for cars are more limited, and the overnight parking fee is more expensive than that for a motorcycle. Meanwhile, the commuters with no vehicle in Yogyakarta have a tendency to choose motorcycle also as their egress mode; being met by their relatives or using a private commercial motor service (motorcycle taxi).

Fig. (2). Egress travel mode choice in each railway station.

Fig. (3). The effect of private vehicle availability on egress trip mode choice.

Moreover, trip frequency also affects the choice of egress trip mode. As previously discussed, one of the criteria for selecting commuters as respondents is that the commuters who have a commuting trip at a minimum of 4 times a month. However, the selection of samples with a high frequency of making a commuting trip was put first. We expect that this selection could provide more accurate results in identifying the factors affecting the egress mode choice.

As shown in Fig. (4), 56% of respondents performed 15-30 trips per month by using the train as their main commuting mode. Most of them tend to opt using private vehicle mode for their egress trip mode either driving alone (by parking their owned vehicle in the destination station) or picked up by their relatives. The similar situation also occurs in 13% of commuters with a high frequency of commuting trip by train (23-30 times a month). Meanwhile, travelers with the lower frequency of commuting trips (less than 15 trips per month), the distribution of egress mode choice tend to be spread evenly in several choices of transport means.

Fig. (4). The effect of commute frequency by train on egress trip mode choice.

The results of logit model are presented in Table 3. Except for a number of transfer variables, all the variables considered had significant effects on the egress mode choice with the confidence level is less than 0.95. Moreover, the attributes of public transport waiting time had a significant value 0.99.

After the coefficient of each affecting attribute was found, the probability of using bus and motorcycle at eight scenarios could be calculated with the results as shown in Fig. (5). The results show that Scenario 7 gives the highest probability of using bus mode for egress trip followed by Scenario 4. Conversely, the scenarios with the highest opportunity of using the private vehicle as the egress mode are Scenarios 2 and 5 with the probability values 81% and 79.2%, respectively.

Fig. (5). Probability of egress mode choice in each scenario.

The simulation was carried out by formulating several scenarios of change in attributes, such as parking fee, trip cost, and travel time in a stratified manner, then examining how large changes in values of utility and probability occurred in each trip mode. The basic assumptions used in the simulation are shown in Table 4 and the simulation results are described as follows.

5.2.1. Role of Walking Distance

As shown in Fig. (6), the scenario was carried out by gradually changing the walking distance with a difference of 100 meters. The result shows that there was no significant shift to use the bus as egress mode by motorcycle users. If the walking distance to the bus stop is similar to the walking distance to the parking area, the probability of using the bus mode was 0.282. If the bus stop is located in the proximity of the railway station exit gate, with the assumption of walking distance is 0 meter, the probability of commuters to use bus mode is 0.334.

Fig. (6). The role of attributes on egress travel mode choice.

Also, in order to support commuters to use the bus for their egress trip, it is necessary to make a change in the walking distance to the parking area within the station. As shown in Fig. (6), if motorcyclists have to walk about 500 meters to reach the parking area, the demand of bus mode will increase from 12.9% to 28.2%. However, when the walking distance is extended and becomes 900 meters, the probability of using bus mode could reach more than 0.5.

Table 5 shows the sensitivity values of using the bus. From Table 5, considering the total travel time, the bus waiting time is more effective to shift to bus mode rather than in-vehicle travel time. Meanwhile, walking distance to the parking area of the private vehicle shows a higher value of sensitivity as compared to the walking distance to bus stop.