SERVICE TIME IMPROVEMENT PROPOSITION FOR INSURANCE CUSTOMER OPERATION AT BPJS KESEHATAN WEST BANDUNG OFFICE: A DISCRETE EVENT SIMULATION APPROACH

BPJS Kesehatan is one of the public legal entities in Indonesia and provides public services to the public, one of which is the administration process service and the provision of face-to-face information directly to the nearest BPJS Kesehatan office. The Customer Satisfaction Index is an assessment of whether the services provided by BPJS Kesehatan are in accordance with participants' expectations. To create excellent service for participants, BPJS Kesehatan refers to 5 dimensions of service quality, namely Tangibility, Empathy, Responsiveness, Reliability, and Assurance. One of the face-to-face service problems faced by BPJS Kesehatan West Bandung Office every day is the waiting time and long queues. This problem was even listed on the Customer Satisfaction Index as one of the participants' dissatisfaction points. To find out what causes long queues and the long waiting time for participants, interviews, and observations of conditions in the field are needed. This can be searched in a structured manner using Fishbone Analysis. To help find solutions to these problems, we use Discrete Event Simulation in the Anylogic application to analyze the existing service conditions and find alternative solutions to solve the issue. As a result, we propose several alternative operational scenarios and define the best scenario that can be applied to increase CSI and reduce participant complaints about waiting times and long queues.


Introduction
Public Services is the important thing for the government or public service providers as an effort to fulfill public needs and implement the provisions of laws and regulations. BPJS Kesehatan is a public legal entity in Indonesia that is directly responsible to the president whose job is to provide health insurance for all Indonesian people. Previously, BPJS Kesehatan had the name ASKES and only had the task of administering the health guarantee program for all Indonesian people [1]. With the increasing number of registered participants, the number of participants who visit the BPJS Kesehatan office is increasing. Even BPJS Kesehatan provide many channels, mostly the participants visit the office to process the administrative needs and information request. The number of participants who CAKRAWALA -Repositori IMWI | Volume 6, Nomor 3, Juni 2023 p-ISSN: 2620-8490; e-ISSN: 2620-8814 come to the office has a big impact on the queue number for participants and makes an impact on the length of waiting time for participants from getting the queue number until getting the services. From this situation, the waiting time and queue length become an issue in face-to-face services at BPJS Kesehatan West Bandung Office. With this situation, the researcher have to identify the root cause, searching for alternative solutions, define the best solution and how to implement the best solution to solve the issue.
Find solutions and answers to these problems, It will take a long time when observing real systems. So, it can affect the face-to-face service process at the BPJS Kesehatan West Bandung Office. To help see whether the new design which is considered a solution to these problems can answer these problems or not, the researchers used a simulation on the Anylogic application. One simulation that is suitable for queuing problems is the Discrete Event Simulation. Discrete Event Simulation is widely used by researchers to solve queuing problems because they can carry out simulation experiments without having to do it in real life by emulating real conditions into a simulation made in the Anylogic application [2,6,11]. Thus, this simulation can be said to represent the conditions that will occur if changes are made in the face-to-face service process.

Service Excellence
In public services, service quality is the main spearhead basis for providing the best service to customers. There are 5 dimensions of service quality (Riyanto, 2012), Tangible, Empathy, Responsiveness, Reliability, and Assurance [3]. The main factor that affected the service quality is the service expected by the customer/participant and the public's perception of the service. The value of service quality is based on the ability of the company and all the employees to fulfill the expectation of customers consistently. Customer satisfaction is the introduction to the customer to use again the services, be a loyal customer, stay as a customer, and in the end give a profit to the company.

Fishbone Diagram
The Fishbone diagram is one of the methods for the researcher to know the problem that is related to its causes [4]. There are parts of the fish that represent different meanings. The 'head' of the fish is the problem that we know as an issue. And for the 'skeleton' leading into the fish's backbone are the specific issue. The researcher mostly used the 'six Ps' in the fishbone diagram.

Discrete Event Simulation
A simulation is the imitation of the operation of a real-world process or system over time (Banks, Nelson, Nicol 2007) [3]. Studying the behavior of a system as it evolves with collecting the data as if a real system and then being observed. The simulation-generated data is used to estimate and the measure performance of the system. To design the system, we have to understand the concept of a system and the system boundaries between the system and its environment. There are components of system simulation, and we have to decide to make a simulation of the system, an entity, an attribute, an activity, an event and state variable. There are 2 categories of systems, discrete and continuous. The difference between both categories is the state variable. For a discrete system, the state variable changes only at a discrete set of points in time, but it changes continuously over time for a continuous system. An example of discrete is the number of customers in the bank changes only when the customer arrives. In a continuous system, water flows in the lake will increase during rain [16][17][18][19][20][21][22][23][24].

Simulation of Queueing System
A queuing system is described by its calling population, the nature of the arrivals, the service mechanism, the system capacity, and the queueing discipline (Banks, Nelson, Nicol 2007). [3] a. The Components of the queue are the calling population/customers, the waiting line/system capacity, and the server. This component is the basic point to make a simulation of queueing system. The calling population may be assumed to be infinite or finite based on the case [8][9][10][11][12][13][14][15]. b. Queueing Notation. Because the queueing system is diverse, Kendal [1953] proposed the format A/Bc/N/K as a notational system [3]. c. Queueing Discipline is the rule by which customers or populations are served or the service discipline which contains the order in which customers receive service. There are 4 types of queueing discipline, FIFO (First In, First Out), LIFO (Last In. First Out), SIRO (Service in random order), and PS (Priority Service) [8][9][10][11][12][13][14][15]. d. The Arrival Distribution is usually calculated by the time between arrivals, i.e. the time between the arrivals of two successive customers [3]. e. The Service Time Distribution is the length of time needed to serve the customers. Usually characterized as a sequence of independent and identically distributed random variables. For service time, the probability distribution that is often used in queues is Exponential Distribution. f. Service Design in the queue can be classified into channel and phase [8][9][10][11][12][13][14][15].

Fig. II Research Design (Source: Discrete Event System
Simulation, Jerry Banks, 2007) For solving the issue, the researcher uses the step of simulation study by Jerry Banks in his fourth edition book of Discrete-Event System Simulation and there are 12 steps to simulate the real-life system to simulation in application. That is problem formulation, the setting of objectives and overall project plan, model conceptualization, data collection, model translation, verified, validated, experiment design, production runs and analysis, more runs, documentation and reporting, and implementation [3].

Fig. III Fishbone Diagram
From the results of interviews with internal and external stakeholders, the researcher finds the root cause of the queues and the waiting time problems faced in the BPJS Kesehatan West Bandung office and draw it using fishbone diagram in Fig III. From the data it can be used by the researcher to CAKRAWALA -Repositori IMWI | Volume 6, Nomor 3, Juni 2023 p-ISSN: 2620-8490; e-ISSN: 2620-8814 make an alternative simulation design and analyze whether the result of the alternative design is better or not when it will be implemented in real life. However, to find out whether solving problems after knowing the root cause can answer all research questions, researchers use simulations to find out whether adding or changing variables in face-to-face services can reduce service time and participant queues. In the next stage, researchers will enter changes or additional variables into the simulation which will be made as a substitute for conditions in the real service system. The components of queue simulation:

Model Conceptualization
The flow of face-to-face service has to know before design it to simulation design. The participants who come to the office has to meet the front guard. Then, Front security guards will carry out filtering, for participants who have WhatsApp and needs that can be processed through other channels will be directed to use these channels to carry out administrative processes and request information, the security guard will filter which ones can be directed directly to the service officer and which participants must be directed to fill out a checklist form. The participants who can enter the office to fill out a form or directly to the service officer to get the check again and get the queue number. For participant who fills out a form, they can join the queue of service officer after finishing it. Next, the service officer will check again the requirements and the needs of the participants and give them the number of queues. After the participants get the number, they will wait for a call from the frontliner in a different room.

Data Collection
In this step, the researcher did quantitative research with an observation in the face-to-face service at BPJS Kesehatan West Bandung office to collect the data needed to input data in the simulation.
The research made observations on participant services for 5 working days, where to get the data needed for the simulation, namely, the number of participants who attend each day, the number of participants who attend every hour, the number of participants who have finished being served, the number of participants who enter the counter, the service time of participants by the security guard, service time for participants by service officers, service time for participants by front-liner officers, waiting time for participants from taking the queue number until they have been served.
a. The number of arrival participants in each hour

Distribution of Service Time Security
c. The number of participants who enter the office to get the services from the service officer, the number of participants who directly go to the service officer or fill out the form.

Model Translation
The model translation is the step for a researcher to enter the model design of the simulation into a computerrecognizable format. For reducing the time of simulation, the researcher used software that is a multimethod simulation modeling tool developed and can support discrete event simulation, that is Anylogic.

Verification and Validation
The next step, the simulation output and real-life data collected before must have a little gap. If the result has a little gap, so the simulation design can be used to find the solution. CAKRAWALA -Repositori IMWI | Volume 6, Nomor 3, Juni 2023 p-ISSN: 2620-8490; e-ISSN: 2620-8814

Fig. VIII The Analysis of services in 1 day
Experiment Design, Production Runs And Analysis, More Runs, Documentation And Reporting a) Experiment 1. In this experiment, the researcher changed some conditions, doing checks at the service officer before entering the counter, reducing participant filtering, the probability of bringing from so to other channel/incomplete file has been added, because a check has already been done in SO, information needs probabilities, administrative needs probabilities and reducing the security guard service time. b) Experiment 2. In this experiment, the researcher changed some conditions, adding the number of Front liner, front liners have 1 hour for break, service time for all resources do not change, the probabilities for each conditions do not change. c) Experiment 3. In this experiment, the researcher changed some conditions, adding the number of front liner, front liners have 1 hour for break, reducing participant filtering, the probability of bringing from so to other channel/incomplete file has been added, because a check has already been done in so, information needs probabilities, administrative needs probabilities, reducing the security guard service time, and do checks at the service officer before entering the counter Experiment 4. For last experiment design, there are the conditions that researcher changed, adding the number of front liner, front liners have 1 hour for break, reducing participant filtering, information needs probabilities, administrative needs probabilities, and reducing the security guard service time.

Result
The following are the results of alternative designs that are processed using simulation and we can see in Table  1: a. Service officer as a bottleneck In experiments 1 and 3, the researcher tried to change the bottleneck to service officer. But it increases the waiting time of the service officer and the services from the frontliner finish longer than 4.30 PM. Because they are too long to get the services from the service officer and go to the frontliner later. And the participant has to wait longer in front guard because the queue capacity service officer was already full. We can see the number of waiting times in frontliner decrease but the total waiting time increase.
But for both experiments increasing the number of waiting times CAKRAWALA -Repositori IMWI | Volume 6, Nomor 3, Juni 2023 p-ISSN: 2620-8490; e-ISSN: 2620-8814 to the frontliner can reach a number smaller than the target waiting time from getting the number of the queue until being called by the frontliner. And decreasing the lost potential income with checks mostly the participants who come to the office as the service officer. Most participants can be directed to the right channel and make sure they're using it properly. b. Adding the number of frontliner For this condition, the researcher implemented experiments 2,3 and 4. For all experiments, it can decrease the waiting time for the frontliner. It made a big impact on all conditions. But, to add the number of frontliner need time for company to recruit and map new employee. And in experiments 2 and 3, after adding 1 of the frontliner, the utilization for each frontliner is under 70%, and it can waste the cost of the company. From the results of simulation, the researcher prefers to use the output using simulation in experiment 4 for decreasing the total waiting time and making sure all the participants who come to the office got the right solution at the counter. And the results using the simulation used are more certain because the simulation is used as a possible output that will be encountered. Not having to check the needs of the participants in the service officer can increase the waiting time service officer like in experiment 3, and this design also answers all the issues. In contrast to the advice given by stakeholders that changing service officers will reduce waiting times and queues, this is the opposite of the simulation results when changing conditions.
The mean total waiting time is decreasing, the mean waiting time to the frontliner is decreasing to 34 minutes and approaching the waiting time set by the company is 30 minutes. The mean of queue length at the security is decreasing, although the length of the queue at the service officer increases from 1 to 3. But, a queue length of 3 is still not that much. And can reduce the loss of potential income due to referrals to other channels by 50%.
From the results of interviews and simulations, there are points that produce contradictory results, namely changing the service officer into a bottleneck where one can check data. The results of the interview said that by changing these points it would reduce waiting time and queue length, but this was not in accordance with the results of the simulation output using discrete event simulation. Using simulation, it is found that changing the bottleneck adds.