Design of Electronic Road Pricing in Jakarta Based on Willingness and Ability to Pay: Addressing Traffic Congestion with Pricing Effectiveness

Traffic volume is one of the leading causes of congestion in big cities related to transportation, land use, and social welfare [1]. People's travel trends can influence congestion, including land use patterns such as population dispersion and expansion of residential areas that adjust the distance between housing and the destination of work and other destinations or community travel patterns, where people tend to travel by motorized vehicles [2]. That is the push demand for increased mobility and accessibility in transportation [3].

The negative impact caused by congestion can affect the economy, society, and the environment. Economically, congestion can cause travel time to increase and increase vehicle fuel consumption, which leads to reduced performance productivity and economic disruption [4]. The environmental impact caused by this congestion is that emissions from the combustion of motor vehicles can increase the greenhouse effect, which increases the earth's temperature. The most severe impact is climate change [5, 6]. Meanwhile, the social effects caused by this congestion are affecting the community's health. Mokbel's [7] revealed that particulate matter (PM), especially PM_2.5, can pose a cancer risk.

To resolve congestion, the management of cities can solve this problem [8]. Efforts that can be made include improving traffic infrastructure, which can be done by widening roads and building over or underpasses, increasing mass transportation as a manifestation of sustainable transportation that is affordable, efficient, safe, and comfortable, and improving traffic system regulation (traffic management) to support traffic infrastructure and public transportation facilities [9]. One of the efforts that can be made in traffic management to control traffic is called Congestion Charge nowadays, better known as ERP [10].

The ERP system was first implemented in Singapore's central business district, and traffic congestion was successfully solved using a dynamic scheme [10]. After Singapore, several cities in Norway also started to implement it from 1986 to 2001, including Bergen, Oslo, Trondheim, Kristiansand, and Stavanger. The implementation of ERP in London and Stockholm has not only succeeded in overcoming congestion. However, it has also reduced particulate matter by 10-15% and carbon dioxide by 13-16% [11]. Some studies above reveal that congestion is the main problem with implementing ERP, as in Jakarta.

As Indonesia's metropolitan city and economic centre, Jakarta still has a severe congestion problem. Although Jakarta already has good public transportation, the lack of integration of some public transit still makes the public perceive that private transportation is better than public transportation [12]. It is better to use private vehicles than public ones because they are more flexible and offer more accessible travel [13]. BPS [14] states that the number of vehicles in Jakarta in 2022 is 21,856,081 units, of which 21,070,506 units, or 90% of the number of vehicles in Jakarta, are dominated by motorcycles and cars. In 2023, the average number of vehicles in Jakarta is dominated by motorcycles with a total of 89,426 units or 71% of the total average vehicles per day, then cars as many as 33,677 units or 27% of the total average vehicles per day, and buses/trucks as many as 3,057 units or 2% of the total average vehicles per day.

Congestion in Jakarta also affects the economy, society, and environment. World Bank [15] notes that traffic congestion in Jakarta results in an annual economic loss of around USD 5.2 billion, approximately 5.4% of the city's GDP. This loss is primarily due to wasted time in traffic, increased fuel consumption, and delayed goods transportation. Socially, congestion increases inequality and affects the quality of life of Jakarta residents. A study on civil servants in the Greater Jakarta area found that over 51% experienced severe mental health disorders due to prolonged exposure to traffic congestion, with symptoms including anxiety and mood disturbances [16]. Congestion also affects commuters with distress, fatigue, and diminished time for family and socialization, further affecting personal harmony [17]. Environmentally, congestion also decreases air quality in Jakarta. Congestion in Jakarta causes higher emissions of pollutants like carbon monoxide and particulate matter (PM_2.5), which harm air conditions and community well-being [18]. Kriswandanu et al. [19] also state that traffic congestion in Jakarta significantly impacts the quality of the environment for residents by contributing to poor air quality.

Implementing ERP to resolve congestion in Jakarta is still a discussion for the regional government, as well as the pros and cons for the public. It is also still not widespread in Indonesian society due to the costs incurred and the low purchasing power of the people. Li and Robuste [20] stated that public acceptance and economic justice issues are the main concerns for implementing ERP. Some critical factors to consider in public acceptance of the implementation of ERP are as follows: 1. Public interest; 2. Use of revenue from ERP implementation; 3. Equality in socio-economic in society; 4. The pricing scheme and its features [21].

Several Asian countries have conducted studies on ERP. With a higher GDP than Indonesia Singapore, the ERP system is a well-established model, effectively managing congestion through dynamic pricing and advanced technology [22]. Seoul's ERP focuses on reducing traffic in central areas, similar to Jakarta's Central Business District approach. Both cities aim to shift commuters to public transport, though Seoul's system is more mature and technologically advanced than Jakarta's [23]. Hong Kong's ERP system emphasizes environmental benefits and urban mobility. Jakarta's ERP could adopt similar strategies to enhance sustainability and public transport integration [24]. While ERP systems in other Asian cities provide valuable insights, Jakarta's unique socio-economic context (e.g., immature public transportation, higher population, and diverse society income) requires a tailored approach. Implementing ERP must balance congestion reduction with social equity and public transport improvements to achieve sustainable urban mobility.

Since Jakarta will implement ERP, the most fundamental thing in balancing congestion reduction with social equity and public transport improvements is the determination of the ERP tariff scheme. Based on this term, WTP and ATP were analyzed to create an effective pricing ERP. These concepts are often used in transportation economics to understand people's behaviour regarding prices and their ability to pay for transportation services [25]. WTP represents the maximum value a person or group is ready to pay for the advantages of a product or service. In contrast, ATP refers to the financial ability of an individual or group to pay for transportation costs. For ERP to gain public acceptance, the pricing structure must align with users' financial capacity (ATP) and perceived benefits (WTP) value, ensuring fairness and minimizing opposition from different socioeconomic segments.

This research was conducted to design an effective ERP that involves the public. This design is based on the public's WTP and ATP to present a solution to traffic congestion by considering the socio-economic circumstances of the public and not burdening the low-income population disproportionately.

1.png

Figure 1. Map of the study area

3.2 Survey method and sample determination

In this research, the researcher used a survey method by distributing questionnaires to road users on the four road sections. The number of samples was taken according to the Cochran method, using stratified random sampling to conduct sample selection.

The Cochran method is suitable for calculating the number of samples whose populations are large or whose exact number is not known for certain, thus enabling the determination of the minimum sample size necessary to achieve results that accurately represent the target population with the specified degree of precision and confidence [54]. Compared to the Slovin method, Cochran is more suitable for sample sizes whose exact numbers are unknown by considering the margin of error, confidence levels, and estimated proportions of the desired population [55]. The formula for the calculation of the Cochran method shown in Eq. (1). 

$n_0=\frac{z^2 \times p \times(1-p)}{E^2}$        (1)

Let z represent the z-value (1.96 for a 95% confidence level); p denotes the expected population proportion (0.5 if unspecified); and E signifies the margin of error.

Applying the Cochran method to determine the total sample size, utilizing a 95% confidence level, a 5% margin of error, and an estimated population percentage of 0.5 yielded a minimum of 385 respondents. In this study, researchers surveyed 400 respondents.

The respondents were selected using the stratified random sampling technique. This technique can ensure the representation of the main subgroup of interest, in contrast to the simple random or cluster sampling method [56]. By stratifying based on relevant characteristics, the researchers can ensure these groups are adequately represented in the sample. Respondents' inclusion criteria include:

1. Cars and motorcycles user; 

2. Age range 18-65 years;

3. Riders who have a consistent frequency, intensity, and pattern of travel in Jakarta;

4. Have a fixed income;

5. Not as an online motorcycle/taxi driver.

While the stratified random sampling approach was carefully designed to ensure representative sampling, there are potential limitations. The sampling method may introduce inherent biases despite its systematic approach. For instance, the inclusion criteria (age range 18-65 years, fixed income, and consistent travel patterns) could potentially exclude important demographic segments, such as gig economy workers or those with less stable incomes. Additionally, the sampling frame might inadvertently over-represent certain socio-economic groups that are more likely to be available and willing to participate in the survey. The main focus is to ensure that the sample comprises regular road users with a consistent frequency, intensity, and pattern and to avoid oversampling from unwanted groups.

3.3 Design of the questionnaire

The research questionnaire is designed to consist of several questions, such as the socio-demographic characteristics, travel characteristics, income per month, expenditure per month, expenditure for transportation per month, ownership of motor vehicles, and frequency and time required to pass through roads to apply ERP. The next step is to determine the WTP and ATP for the ERP service from the data.

3.4 Data analysis

3.4.1 ATP

In this research, ATP value is obtained based on input data in the form of income, total expenses, transportation costs, number of frequencies, and distance travelled. The distance used in the ATP analysis is the distance travelled through the route the ERP plans to implement. ATP is calculated by considering income, expenses, transportation costs (non-fixed and fixed), and the frequency and distance travelled by respondents. The amount of ATP is the number of transportation costs totalled by the margin of income and expenses minus investment costs divided by the frequency of monthly trips as formulated in Eq. (2):

$A T P=\frac{[(Inc-Exp)-Inv]}{F o M}$        (2)

where, Inc is income, Exp is Expense, Inv is investment cost, and FoM is the frequency of monthly trips.

3.4.2 WTP

The value of WTP is acquired by conducting an analysis using a combination of Stated Preference – Reveal Preference methods [44], i.e., DCE, TCM, and RPM. DCE allows for the estimation of individual preferences for different attributes of the ERP system [33]. TCM provides insights into the monetary value individuals place on travel time and costs [36]. RPM reveals actual behaviour based on observed choices [37]. These methods were chosen over others due to their ability to combine stated and revealed preferences, offering a comprehensive understanding of how users value ERP implementation [44]. The estimation of WTP value using method, i.e.:

1. The formula calculates DCE using three attributes: time savings, reliability, and comfort. To get the WTP DCE value, use the formula in Eq. (3), with attribute weight based on the primary survey result [57].

$W T P_{D C E}=\sum(attval \times attwght)$           (3)

where, attval is attribute value, and attwght is attribute weight.

The attribute value is calculated using Eqs. (4)-(6).

$ Time \,\, Saving =(P T T-E T T) \times VoT/min $    (4)

$Reliability =(P T T-E T T) \times V o T / m i n \times Coeff.Rbl $     (5)

$Comfort =(PTT-ETT) \times VoT/min \times Coeff.Cmft $     (6)

where, PTT is the present travel time, ETT is the expected travel time, and Coeff.Rbl is coefficient of reliability, and Coeff.Cmft is coefficient of comfort

With VoT/min calculated with Eq. (7):

VoT/Min $=\left(\frac{ { Monthly\ Revenue }}{ { WorkHr } \times  { WorkDy } \times 60}\right)$        (7)

where, WorkHr is working hours and WorkDy is working days.

2. TCM involves several components (e.g., fuel, time, and maintenance costs). It is obtained using the formula Eqs. (8)-(10).

$Fuel\ Cost = Fuel / \mathrm{Km} \times PoF/L \times LoR $     (8)

$Time\ Cost = VoT / \mathrm{Min} \times TimeRdc $       (9)

$Maint.\ Cost = Cost / \mathrm{km} \times LoR $      (10)

where, PoF/L is the price of the fuel per liter, LoR is the length of the road, and TimeRdc is the time reduction.

3. Revealed Preference Method (RPM), studying the actions of consumers of products and services in order to tease out the values underpinning them. TCM, for instance, assumes that the cost paid to accomplish a goal indicates the value that consumers are rewarded for this goal pursuit [58]. The analysis of WTP is associated with the coefficient of elasticity, which calibrates calculations by altering demand sensitivity to price variations [59]. RPM analysis with the elasticity of the vehicle, with the formula as Eq. (11), where car coefficient elasticity is -0.3 and motorcycle coefficient elasticity -0.5 [59].

$WTP_{RPM} =T C M \times(1-Coeff. Elasticity)$      (11)

3.4.3 AHP for combined SP RP method

The AHP was integrated to prioritize and weigh the factors affecting user decisions. AHP’s systematic approach enables the aggregation of multi-criteria decision-making, ensuring that the trade-offs between different factors (e.g., cost, time, convenience) are considered in the modelling process, enhancing the robustness of the analysis [48]. From the definition above, AHP can give weight to each WTP method (DCE, TCM, RPM) by assessing relevant factors based on preferences and priorities. The analysis results become more accurate and objectively acceptable by weighting each method according to its relevance factors.

The process of the AHP follows [60]:

Step 1: Defining the AHP framework

The SP RP method combination calculations are organized around three approaches (DCE, TCM, and RPM), each with related criteria structured in levels representing the elements' hierarchical connections. A more intricate structural model with multi-tiered analysis is developed to address this complexity. Given the in-depth nature of the study, the model can have multiple levels; however, the top level consists of a single element, which is the desired goal or outcome (value of combined WTP). The intermediate level delineates the steps necessary to attain this objective (the three WTP methods), which can be categorized into multiple sub-levels. The lowest level outlines the actions and decisions required to achieve the goal, including criteria such as data reliability (DR), implementation cost (IC), comprehensiveness (C), and temporal relevance (TR). Once the hierarchical structure is established, the relationships among the factors at each level become clear. This hierarchy is illustrated in Figure 2.

2.png

Figure 2. AHP framework for combine SP RP method

The significance of each factor is assessed using a scale of 1-9 [61], with Table 3 providing the meanings for each value on the 1-9 scale.

Step 2: Building a pairwise comparison matrix

Once the criteria are determined, the next step is to compare each criterion in pairs using a predetermined assessment scale. A paired comparison matrix A=[a_ij] will be formed, where each element a_ij shows a comparison between the criteria i and j.

A pairwise comparison matrix can be described as Eq. (12):

$A=\left[\begin{array}{cccc}1 & m_{12} & \cdots & m_{l n} \\ m_{21} & 1 & & \vdots \\ \vdots & & 1 & \\ m_{n l} & \cdots & & 1\end{array}\right]=\left(A_{i j}\right)_{n \times n}$         (12)

Table 3. Importance scale for prioritization

Once the pairwise comparison matrix is arranged, the next step is to normalize the matrix. Normalization is conducted by dividing each element of aij by the number of columns, thus producing a normalization matrix n_ij, where each element n_ij is calculated with the formula as Eq. (13):

$n_{i j}=\frac{a i j}{C_j}$      (13)

where, C_j is the number of columns j in the pairwise comparison matrix.

Then, the criterion's weight will be calculated by taking the average of the normalized value on each line. Weight criteria w_i for criteria i calculated with the formula as Eq. (14):

$w_i=\frac{1}{m} \sum_{j=1}^m n_{i j}$        (14)

Step 3: Consistency test

After the weight calculation steps are done, testing the consistency in the paired comparison matrix is important. The consistency test aims to ensure that the comparisons made between the elements in the matrix are consistent and not contradictory and that the results are valid. The first step is to calculate the consistency index (CI). CI is used to measure whether the pairing ratio in the matrix is consistent. CI is calculated using the following formula as Eq. (15):

$C I=\frac{\lambda_{\max }-n}{n-1}$        (15)

where, $\lambda_{\max }$ represents the maximal eigenvalue of the paired comparison matrix and n is the sum of the criteria or alternatives in the comparison matrix.

Upon establishing the CI value, the subsequent step is to compute the consistency ratio (CR) to evaluate the consistency of the entire comparison matrix. CR is found by dividing the CI by the random consistency index (RI) based on the matrix's size (number of criteria or alternatives), as shown in the formula in Eq. (16).

If the CR value is less than 0.1, then it is consistent.

$C R=\frac{C I}{R I}$      (16)

Step 4: Evaluate the final decision

To get the final weights in each method, multiply each method’s weight for each criterion by the criterion’s importance.

One of the interesting findings in this study is the significant difference between WTP and ATP. In general, the results of all WTP methods, even their combinations, show that the people of Jakarta have relatively low rates of WTP compared to their ability to pay ATP. Even with high payment ability, Jakarta people are reluctant to pay ERP. The notable disparity between WTP and ATP in Jakarta's context can be discussed by the prospect theory formulated by Kahneman and Tversky [72]. In theory, decisions are not always rational in an economic sense. Several psychological factors may contribute to Jakarta citizens' reluctance to pay ERP despite their high payment ability. One factor could be loss aversion, where the pain of losing money (paying ERP) is perceived as greater than the pleasure of gaining benefits (reduced traffic congestion) [73]. Framing effects also play a role. Suppose the ERP is framed as a loss (a tax or a fee) rather than a gain (an investment in better traffic management). In that case, it may lead to a lower willingness to pay [74]. Based on this theory, policymakers may need to consider these behavioural aspects when designing and communicating about the ERP system to increase public acceptance and willingness to pay [75].

The discrepancy between ATP and WTP has important implications for policy design, particularly concerning equity and public acceptance [76]. In equity, people have the financial means to pay when ATP exceeds WTP. However, they are unwilling to do so because the perceived benefits (such as time savings or reduced congestion) may not justify the cost [77]. Moreover, this discrepancy can create inequities. For instance, high-income individuals (with higher ATP) might not feel the need to pay. Lower-income individuals (who may have a lower ATP but a higher WTP due to stronger reliance on the service) could be disproportionately affected by high tariffs. This could lead to a regressive pricing structure that burdens those with lower incomes despite having a higher inclination to pay for service improvements. In public acceptance, the discrepancy between ATP and WTP can affect this acceptance. For instance, if policies are not adjusted to align the pricing with actual usage behaviour (WTP), people may view the policy as unfair, leading to decreased support and engagement [42]. Ensuring that public transport options are enhanced and affordable alternatives are available is crucial to alleviate concerns from low-income groups, improving overall public acceptance.

Suppose ERP is implemented later with a fixed-cost scheme. In that case, it is necessary to assess its implementation because there is a discrepancy between ATP and WTP. If inefficiencies are found, implementing a dynamic fare system can address these differences by adjusting costs based on times, travel routes, or congestion levels. Jia et al. [78] revealed that implementing ERP with the characteristics of road users dominated by private vehicles with an average travel pattern of five times per week and implementing a dynamic fare system is effective in overcoming congestion. However, in its implementation, individual traveller information (e.g., demographics, income, expenses, travel costs) should be considered to improve the effectiveness of the ERP scheme. Anjomani [79] also argues that the effectiveness of the pricing scheme is determined by evaluating the demographic factors in addition to the available transportation system and environmental factors. ERP dynamic schemes offer promising solutions for urban traffic management, but challenges remain in their implementation. These include the need for robust data collection and analysis systems, public acceptance, and integration of such schemes with broader urban planning and environmental goals.

As a pioneer in ERP, Singapore's system is very sophisticated, using real-time data to adjust pricing dynamically based on traffic conditions [1]. London and Stockholm also effectively implement ERP, which can reduce congestion and emissions. The effectiveness of ERP depends on the availability of supporting infrastructure, such as efficient public transport systems and modern ticketing technologies [80]. Suppose a dynamic pricing scheme would be implemented in Jakarta. In that case, careful planning and adaptation to local conditions in Jakarta is needed. Jakarta's unique challenges, such as high population density, traffic congestion, immature public transport, and various income societies, may require tailored solutions rather than direct comparisons with other countries' systems. Leveraging computing technologies and models that can help ERP effectiveness is also very important. In addition, considering that Jakarta has just started ERP, data implementation in Jakarta is very limited. Most data collected is limited to interviews with potential user respondents, which should also get attention.

Although this study contributes useful insights into ERP implementation in Jakarta, various methodological limitations should be carefully considered. Firstly, the research applies to only a sample of 400 respondents, which can’t reflect all segments of Jakarta's diverse population, even with sound sampling methodologies. Although scientifically valid, the stratified random sampling might induce selection biases that could affect the findings.

Additionally, the study is limited to several road sections in Jakarta's central business district. Though important to the city's transportation infrastructure, these roads may not capture the full breadth of the urban transportation ecosystem. The specificities of these roads, including their economic relevance, integration with mass transit systems, and particular traffic patterns, may not be fully generalizable to the rest of Jakarta or other Indonesian cities.

Finally, more subtle ATP and WTP realizations include methodological insights in the pricing mechanism design. Even though the metrics are actionable, the alarming gap between these two measures indicates that user behaviour can’t be predicted by economic models alone. There are important psychological dynamics, framing effects, and individual perceptions that cannot be perfectly quantified.

These concerns should not detract from the study's importance. However, they reflect the complicated nature of urban transportation research. Their unique contributions emphasize the importance of continued, multi-method research approaches to reflect the nuanced realities of urban mobility in rapidly developing cities such as Jakarta.