Low Emission Zone (LEZ) Expansion in Jakarta: Acceptability and Restriction Preference

Abstract

After the Jakarta government enacted a policy to improve air quality in high-priority areas and established Low Emission Zone (LEZ) in the Kota Tua Tourism Area (KTTA), there is now a plan to replicate this LEZ in other locations. However, the current scale of the LEZ in the KTTA and the ongoing issues with public acceptance, raise the question of how the Jakarta Government effectively replicates this policy in other locations. This study aims to explore the socio-psychological factors that affect acceptability and restriction preference (i.e., type of vehicle and strategies) on LEZ replication in Jakarta. For those purposes, questionnaires were distributed to the commuters in Jakarta, and hierarchical Structural Equation Modeling (SEM) and Multinomial Logistics Regression (MNL) were used to analyze the data. The support for LEZ expansion to other locations was found to be shaped by the trust of the government, the level of environmental concern, and personal and social norms concerning LEZ implementation. The perception of the form of full-scale restriction (i.e., combinations of the vehicle types to be restricted and/or the TDM strategies) that should be implemented within the LEZ area is associated with higher acceptability and environmental concern. Moreover, younger and wealthier people are among the groups who are more concerned about the effectiveness of LEZ implementation.

1. Introduction

Urban areas in developing countries have been facing serious issues related to air pollution due to substantial motorization and low public transport use [1]. To address various public health and wellbeing risks from deteriorating air quality [2,3], several countries (e.g., the UK, Spain, Sweden, China, Singapore, Indonesia, etc.) have implemented policies restricting private vehicles. One of the policies is the enactment of a Low Emission Zone (LEZ) [4]. In 2019, Jakarta was among the world’s top five most polluted cities [5], resulting in 10 million people exposed to serious health risks that have caused five million illnesses annually [6]. Studies show that the transportation sector is a major contributor to air pollution in Jakarta [7].

Consequently, Jakarta Governor Instruction No. 66 of 2019 on Air Quality Control was passed to reduce air pollution via LEZ as a strategy in the transportation sector, along with other strategies such as public transportation development, congestion charges and vehicle emission monitoring. The LEZ has been shown to be effective in improving air quality. The LEZ in London has reduced PM₁₀ emission by 1.9% and NOx emission by 2.4%, while the LEZ in Beijing is estimated to eliminate 90 metric tons of air pollutants daily [4]. Jakarta has begun implementing an LEZ pilot project in the Kota Tua Tourism Area (KTTA) in early 2021. Unlike the LEZ in London or Beijing [4,5], the KTTA LEZ covers a relatively small area (approximately 12 ha). Rather than charging for private vehicles as implemented in London, the KTTA implements private vehicle restrictions, excluding vehicles that have passed the emission test, only allowing active mobility and public transport, similar to the policy implemented in Barcelona [8,9]. Moreover, the government has a further plan to replicate the LEZ in other locations that aim to improve air quality. However, with the unique implementation in KTTA, there is a question regarding acceptability of LEZ expansion and the restrictions that can be implemented within this expansion.

Public acceptance is one of the crucial success factors in LEZ implementation [10,11]. In order to improve public acceptance, the public must be aware of the problems being addressed and the effectiveness of the policy in delivering benefits to address these problems [12]. More importantly, the LEZ is often seen as limiting personal freedom or as having an economic impact due to its distributional implication [13,14]. Therefore, there is a need to better understand public perception of the factors driving LEZ implementation. Various countries—mostly in developed countries [7,13,14]—have explored the public perception of and public acceptance of LEZ in order to evaluate its impact and effectiveness. However, considering the complexities involved in measuring public acceptance and its factors, several scholars have included socio-psychological factors such as attitude, perception, evaluation, beliefs and effects in studying public perception of LEZ implementation. For instance, Oltra et al. [10] included prior attitude, institutional trust and process legitimacy as the factors that affect public acceptance of LEZ in Barcelona. Public perception has also been studied to measure the acceptability of a sustainability policy, covering such attributes as environmental awareness, perception or perceived LEZ benefits and personal norms, as in Morton et al. [15] in Scotland. The study by Wang et al. [11] added more personal, social and city indicators to measure the acceptance of mobility restriction policies in Latin American cities.

Investigating the acceptability and restriction preference within LEZ expansion in Indonesia is important for several reasons. Firstly, the amount of motorization and urbanization has substantially increased in the past decades and is predicted to continue increasing in the following years [1]. This increase makes the implementation of LEZ, as a part of push policy [16], important when influencing people to use public transport, as is currently being developed by the government. Secondly, the characteristics of Indonesian LEZ are different from LEZs in most countries. Therefore, investigating the type of restriction that should be applied is also important in the light of public perception. Other cities such as Stockholm or Singapore have implemented congestion charging to complement their LEZ policy [11], while in KTTA the implementation relies on emission-based restriction. Lastly, unlike other cities, Jakarta is starting LEZ in small areas of KTTA and the expansion plan aims to maximize its impact on air quality and, consequently. the well-being of citizens.

This study aims to investigate the factors that affect the acceptability of LEZ expansion as well as the type of restriction implemented. Specifically, the novel contributions of this study are threefold. First, earlier studies have explored the acceptance and perceived benefits of LEZ [7,15,16,17]. To the best of authors’ knowledge, this is the first study that focuses explicitly on LEZ expansion and explores the type of restriction (i.e., vehicle and policy) that the public prefer. For this study, we used the extension of the socio-psychological indicators and framework from Morton et al. [15]. Second, most research on LEZ acceptance comes from cities in developed countries [4,15] and there is little information on how commuters from emerging economies perceive LEZ. Developing countries, such as Indonesia, have still been pursuing economic investments including accelerating infrastructure development in recent decades. Thus, to some extent, policy instruments that might affect economic growth and investment attractiveness tend to be limited. LEZ is one of them and it would be interesting to comprehend how it affects people’s acceptability. Lastly, the unique characteristics of LEZ implementation in Jakarta will add insights into the perception and attitude of the people toward LEZ implementation. The unique factor in Jakarta’s LEZ implementation, which affects the surrounding traffic, might have shaped the public perception of the policy’s benefits, which differs from other countries. By doing so, this study is seeking to contribute to expanding the knowledge of LEZ acceptance in developing countries. We performed hierarchical Structural Equation Modeling (SEM) and Multinomial Logistics Regression (MNL) [18] on the data collected in online and offline surveys conducted in Jakarta from April to May 2022 (Supplementary Materials). The classification and dimension reduction [18] analyses were also performed for the data.

The remainder of the paper is structured as follows: the following section presents the review on LEZ and its implementation in Indonesia; the third section describes the research design, questionnaire development and data collection processes; the fourth section explains the result of the model estimation 1divided into two analyses. The first analysis focused on the acceptance of LEZ expansion and the second on the type of restriction (i.e., type of vehicle and policy), followed by an explanation and discussion of the findings. The fifth section concludes the paper.

2. Literature Review

2.1. Low Emission Zone (LEZ)

Generally, an LEZ is defined as a dedicated emissions control area established to improve air quality by reducing vehicle pollutants [4]. LEZ is one of the Transport Demand Management (TDM) [13,16,17,19] strategies that usually implement restrictions on the movement of certain vehicles within set areas. The set areas that are decided within the LEZ scope are usually based on pollution level, with areas with higher pollution levels having the highest priority for LEZ implementation. The type of restriction within the LEZ is mostly based on emissions, such as in London [20] and Brussels [21]. However, some cities have combined the LEZ with other restrictions. For instance, London and Stockholm have combined the LEZ with congestion charging (CC), while Berlin and Singapore only select one of the LEZ and CC policies to implement [4]. Even though LEZ and CC are different in terms of objectives, in general both of these policies deal with the urban transport challenges that emerged from the increasing amount of private vehicle use. With the effect on road use (often also implying an increase in public transport use), a positive effect on air quality is also generated.

Similar to other forms of TDM strategy implementation [22], implementing LEZ often raises various conflicts with users and is often controversial. Firstly, users might consider that it reduces their rights to move freely. Secondly, as it generates revenue (i.e., from congestion pricing), users might have low confidence in the uses of this revenue by the government. Lastly, the benefit of LEZ for the environment is not immediately visible (such as improved air quality). Therefore, users might believe that this policy is a waste of public money [13,22]. Moreover, some frustrations may arise because users are offered no alternative routes or no means to pay a fee if they use their vehicles within the LEZ. Nonetheless, various researchers have investigated the effect of LEZ implementation. The results of LEZ implementation have been promising in various countries. In London in 2013, after 5 years of LEZ implementation throughout the city, particulate matter (PM) concentration decreased by around 2.46–3.07% in the inner zone and by 1% outside the zone [4]. PM emissions from heavy-duty vehicles across the city decreased by 40% four years after introducing LEZ in Stockholm [4]. In Paris Region, 13–43% of residents vulnerable to dense NO₂ concentrations outside the LEZ perimeter fell below the critical threshold [23]. In Lisbon, Ferreira et al. [24] found a 23% decrease in annual average PM₁₀ after 2 years of implementation.

2.2. LEZ Implementation in Jakarta

Jakarta—the capital city of Indonesia and an emerging megapolis comprising also the city’s outskirts—has experienced rapid urban population growth that has resulted in congestion, poor air quality and ultimately economic downturn [1]. The PM2.5 concentration increased by 50% from 20 µg/m³ (with a safe limit of 10 µg/m³ according World Health Organization (WHO) standards [25]) in Jakarta between 2017 and 2018. By 2019, the pollution level had exceeded those numbers, given the extensive growth of public activities [25]. Jakarta has been mentioned as the region with the highest Particulate Matter (PM) 2.5 content in the world on several occasions [8,26]. In 2019, the Jakarta Governor decreed Instruction No. 66 of 2019 concerning Air Quality Control, establishing various policies to improve urban air quality. Under the consideration that half of the pollutants in Jakarta are generated by land transportation [7], several policies restricting mobility—odd-even restrictive driving policy and Low Emission Zone (LEZ)—were implemented in Jakarta.

The Kota Tua Tourism Area (KTTA) in Jakarta—a popular tourist area for Indonesians and one of the strategic tourism areas in the National Development Masterplan [27]—has been turned into an LEZ. After several weeks of trial at the end of 2020, on 8 February 2021 the KTTA LEZ implemented the open-closed policy [28] throughout the day for 24 h. The KTTA LEZ is implemented based on Government Regulation No. 32 of 2011 concerning Management and Engineering, Impact Analysis, and Management of Traffic Needs, and Governor Regulation No. 36 of 2014 concerning the Old Town Master Plan. The KTTA LEZ is an area where access is restricted for certain types of polluting vehicles to improve air quality in the area, which is quite similar to LEZ implementation in various cities such as Barcelona and some towns in Scotland [7,16]. However, there are differences in its implementation in Jakarta. For instance, the KTTA LEZ does not impose congestion charges as does the LEZ in London or Beijing [11]. The odd-even restrictive driving policy [29] is implemented on several roads around the KTTA to support the LEZ, which differs from LEZ implementation in other cities. The most significant difference is the area covered by the KTTA LEZ. The London LEZ covers more than 1500 km², the Stockholm LEZ covers more than 100 km² [4], while the Jakarta LEZ only covers around 0.12 km² of the KTTA area (Figure 1).

A series of demand management and infrastructure development programs are being implemented in KTTA. The infrastructure development consists of improving pedestrian facilities and the public transport infrastructure (i.e., BRT corridor, multimodal to the Regional Rail Stations, and MRT Jakarta North–South corridor). Demand management is also being implemented by revitalizing the micro-shops in the KTTA and its surroundings and monitoring the vehicle emission restriction implementation. To ensure the traffic flow within the areas, the government implemented traffic management on several links road within the KTTA. Moreover, air quality monitoring tools were also placed in the KTTA areas to evaluate the improvement of air quality due to the LEZ.

2.3. LEZ Acceptance

Due to its implications for public mobility, similar to other TDM strategies, various research projects have started to investigate the acceptance of LEZ. However, tjhere is only a small amount of literatures investigating the acceptance of LEZ, and most of the research has used the framework of TDM strategies such as CC. A study has found that the effect of socio-demographic characteristics (i.e., income, age and gender) have shaped LEZ acceptance, with higher income and younger people tending to vote against the LEZ [30]. Moreover, acceptance has been investigated via a more complex socio-psychological construct by using Theory Planned Behaviour (TPB) and the Norm Activation Model (NAM) under policy-specific beliefs, along with distal factor theory [10,13,22,30]. The use of TPB and NAM aim to reveal specific individuals’ driving factors, i.e., attitudes, self-efficacies, personal and social morals, on their actions. On the other hand, policy-specific beliefs and distal factors are used to represent perceptions or preferences on the effect or impact of the operational-specific policy used regarding various TDM strategies’ acceptance.

By applying a framework based on Theory Planned Behaviour (TPB) and the Norm Activation Model (NAM), Morton et al. [15] found that distal construct (problem awareness), attitude and policy-specific beliefs (trust in government) play a direct and indirect role in LEZ acceptability in Scotland. Studied by Mehdizadeh and Shariat-Mohaymany [30] in Teheran, Iran and Oltra et al. [10] in Barcelona, Spain also underlines the importance of policy-specific beliefs along with trust in government policy (i.e., the transparency on LEZ revenue allocation), the belief that LEZ is ineffective and suffers from complicated details, and lack of awareness regarding transport and air pollution to LEZ acceptance. This is in line with the previous research on TDM acceptance such as the finding related to trust in government policy by Balbontin et al. [31] in Sydney, Australia and, hence, increasing people’s awareness of how the revenue from LEZ will be spent is crucial for public acceptance [32].

While previous studies have revealed the role of socio-psychological constructs in LEZ acceptance in general, the effects of specific vehicle restriction or management strategies within LEZ implementation remains missing. Therefore, that is the gap that aims to be filled by this study. By extending the previous TDM strategies’ [22,33,34] study framework, this study uses socio-psychological explanatory variables on the choice of LEZ vehicle restriction/management strategies combinations. Moreover, most of the research above has been carried out in developed countries, given that the implementation of LEZ in developing countries is limited. Implementation in developing countries might result differently due to two factors. Firstly, the unique characteristics of the economy, culture, infrastructure and built environment of developing countries are different from those of developed countries [35]. Secondly is the technical implementation of the LEZ itself. For instance, the uniqueness of the KTTA LEZ (i.e., implementation coverage and type of mobility-restriction policy) compared to LEZs in other developed cities might influence public perception of the LEZ policy. More importantly, Jakarta has a plan to expand the LEZ to other locations, which is not the case in other locations in other countries. Therefore, the investigation into LEZ implementation in Jakarta will contribute to extending the literature on LEZ implementation. Moreover, Jakarta has planned a set of rigorous policies to reduce air pollution. These policies comprise LEZ expansion, electronic road pricing implementation and intermodal urban public transportation development (i.e., mass rail transport, light rail train and bus rapid transit) for the coming decade [36]. Therefore, it is important to understand the people’s acceptability and preference for future LEZ in Jakarta in ensuring that the appropriate policy is put in place.

3. The Data Utilized

3.1. Research Design

To explore the factors that affect the acceptability of LEZ expansion and the type of restriction implemented, this study has developed two structural models illustrated in Figure 2. The first structural model, on the left side of Figure 2, is based on the framework of citizen acceptance of sustainable policy, with various socio-psychological indicators from Morton et al. [15]. The indicators used are combinations of theories of behaviour and citizen acceptance. The main indicators used for individual behaviour are based on TPB [30] which underlines the role of attitude, perceived behaviour control and self-efficacy. The modification of TPB is done by applying the NAM [37]. The NAM stipulates that individuals’ behaviour is influenced by one’s relationship with personal morals and social conventions. With the LEZ policy implemented by the government, the socio-psychological indicators related to policy-specific beliefs are also measured. One of the indicators related to the implementation, which consists of perceptions of the efficacy, cost, fairness, and the assignment of emotions to an issue, represents the value of policy implementation. This indicator on the government side was also measured from the perspective of trust held in government [22].

Moreover, the urgency of the policy implemented is also measured in relation to the problem of perception/awareness and from the perspective of environmental concern. These indicators were previously examined in past TDM acceptability research [22]. In summary, there are ten LEZ acceptability indicators for sustainable policy implementation: trust, problem perception, positive affect, negative affect, perceived benefit, perceived cost, perceived risk, outcome efficacy, distributive fairness and procedural fairness. Four indicators represent individuals’ behaviour in TPB and NAM: personal norm, social norm, attitude and perceived behavioral control.

Moreover, to extend the acceptability analysis, the framework of restriction preference analysis is illustrated on the right side of Figure 2. To extend the analysis, the framework divides the restriction preference into two models of type of restriction: restriction based on the vehicle (e.g., car, motorcycle and goods)—Y1—and based on the strategy/restriction (e.g., parking tariff management, emission, congestion charging and odd-even)–Y2. The strategy/restriction (Y2) options are based on the availability of TDM strategies in Jakarta, mostly strategies that are implemented in other countries [17] with the addition of odd-even strategies that are uniquely implemented in Jakarta. Similar to the acceptability analysis, the various socio-psychological indicators from Morton et al. [15] are also used for the exploratory variables, consisting of policy-specific beliefs, individuals’ behaviour and sustainable policy implementation acceptance [37,38].

3.2. Data Collection

Based on the formulated indicators, a five-section questionnaire was developed. In the first section, respondents were asked about their awareness of LEZ in general and in KTTA. The second section introduces the expansion of LEZ in Jakarta and the respondents are asked about acceptability of and preference for locations. The questions related to the type of restriction are divided into two types (i.e., type of vehicle and type of policy) in the second section. While the type of vehicle restricted consists of three types (i.e., motorcycle, ca, and goods vehicle), restriction strategy consists of four types (i.e., odd-even, congestion charging, parking fare management and emission-based restriction). In both restrictions, respondents respond on a five-point Likert scale ranging from 1 for “strongly disagree” to 5 for “strongly agree”. Questions related to trust, problem perception, positive affect, negative affect, perceived benefit, perceived cost, perceived risk, outcome efficacy, distributive fairness and procedural fairness are asked in the third section. The five-point Likert scale ranges from 1 for “strongly disagree” to 5 for “strongly agree” to capture the respondents’ responses. Moreover, personal norms, social norms, attitude and perceived behavioral control are asked about in the fourth section using a similar five-point Likert scale to capture the response. The questionnaire closes with socio-demography (i.e., age, income, occupation, etc.) and travel characteristics (i.e., origin, destination, mode choice).

Since this study seeks to assess the acceptability of LEZ expansion in Jakarta and its influencing factors, the target respondents are people who regularly perform activities in Jakarta areas. Data collection was conducted offline in five regions within Jakarta (i.e., south, north, central, west and east). While most private transport users will be directly affected by the LEZ, public transport users are also indirectly affected by road and public transport level of services changes. Therefore, both of these users are target respondents of the survey. The surveyors approached respondents in their spare time outside their offices or in public transport stations. Since the survey was conducted during the COVID-19 pandemic in Indonesia, the mode of the questionnaire was still web-based. A strict health protocol (i.e., mask use, vaccination, regular hand washing) in line with the local government regulation was applied during the survey. Only respondents who frequently travel to Jakarta were accepted as valid. Before the survey, the surveyors explained that it would take 10–15 min to complete the survey and that the survey would only be continued for respondents who agreed to that timeframe. The survey was conducted in the second quarter of 2022 from 1 April to 15 May, with a total of 730 valid responses out of 833 collected (Supplementary Materials).

3.3. Socio-Demographic, Socio-Psychological and Restriction Preference Characteristics

The respondents’ demography is presented in

Table 1. Respondents’ Demography (n = 730).

Variables		Proportion
Age	<23 years old	2.99%
	23–40 years old	75.07%
	41–60 years old	21.94%
	>60 years old	0.00%
Gender	Male	61.68%
	Female	38.32%
Occupation	Civil Servant	2.42%
	Medical staff/doctor	0.57%
	Private employee	75.21%
	Teacher/lecturer	4.42%
	Student	0.85%
	Housewife	0.28%
	Freelancer	0.57%
	Unemployed/retired	2.85%
	Others	12.82%
Monthly income a	<IDR.2,000,000	0.28%
	IDR.2,000,000–IDR.4,000,000	3.28%
	IDR.4,000,001–IDR.6,000,000	67.52%
	IDR.6,000,001–IDR.8,000,000	22.79%
	IDR.8,000,001–IDR.10,000,000	4.70%
	IDR.10,000,001–IDR.15,000,000	1.42%
	>IDR.15,000,000	0.00%
Main transport modes	Motorcycle	21.97%
	Car	16.67%
	Bus	33.59%
	Rail	23.22%
	MBRS *	3.42%
	CBRS *	1.14%

^a US$1 equal to IDR14,501 in July 2022. * MBRS: Motorcycle-based Ride-sourcing; CBRS: Car-based Ride-sourcing [39].

. Most respondents were male (61.6%) and in their productive age (23–40 years old; 75.1%). The respondents mostly worked in private companies (75.1%). Their monthly income ranges from IDR4 million to IDR8 million (equal to US$267–551). For their commuting trips, the respondents mostly use public transport. At the same time, private transport users comprise more than 38% of the respondents.

The socio-psychological variables of the respondents are described in

Table 2. Respondents’ Socio-Psychological Variables.

Construct and Statement a	M	SD	FL	VE
Acceptability (α = 0.796)
I support the introduction of LEZ in my nearest area	4.135	0.540	0.885	71.4
I believe the implementation of LEZ is an acceptable policy	4.182	0.574	0.841
I will vote to support LEZ if there is a public discussion by the government or a representative of the stakeholders	4.154	0.611	0.808
Social Norm (α = 0.632)
I think the people of Jakarta are worried about air pollution	4.161	0.488	0.858	73.5
My friends and family are axious with the air pollution level	4.162	0.599	0.858
Perceived Behavioural Control (PBC) (α = 0.508)
It would be easy for me to adapt my travel behaviour in the LEZ	3.994	0.678	0.823	67.7
I don’t need a car to get to the city center so LEZ is unlikely to influence me	3.610	0.977	0.823
Attitude (α = 0.656)
LEZ should not be taken into account by policy makers R	4.101	0.515	0.870	75.6
Public money would be wasted on LEZ investment R	4.020	0.710	0.870
Personal Norm (α = 0.655)
The LEZ concept align perfectly with my values	4.003	0.600	0.863	74.4
I personally care about policy of improving air quality	4.251	0.563	0.863
Positive Affect (α = 0.810)
Establishing LEZ would make me proud	3.828	0.630	0.858	72.4
I would be thrilled about the prospect of Low Emission Zones	3.979	0.640	0.868
I would be happy to see LEZ introduced	4.056	0.608	0.826
Negative Affect (α = 0.830)
The LEZ implementation would irritate me	2.101	0.734	0.835	74.9
I believe that the regulations surrounding LEZ would aggravate me	1.879	0.674	0.914
If LEZ were implemented, I would be disappointed	1.972	0.698	0.847
Perceived Benefit (α = 0.614)
The LEZ would improve public health if they were implemented.	4.100	0.449	0.851	72.4
Walking within a LEZ would be a enjoyable experience	4.306	0.522	0.851
Perceived Cost (α = 0.822)
The public expenses of establishing and maintaining LEZ would be enormous	3.123	1.010	0.923	85.2
Economic prosperity would decline if LEZ were implemented	2.897	1.163	0.923
Perceived Risk (α = 0.874)
LEZ only transferred heavily polluting automobiles to different places.	3.205	0.997	0.937	80.4
People will discover ways to bypass LEZ restrictions.	3.399	1.166	0.910
Introducing LEZ would probably have unanticipated negative effects.	2.764	1.053	0.840
Outcome Efficacy (α = 0.567)
LEZ wouldn’t reduce local air pollution levels in a meaningful way R	4.098	0.323	0.859	73.7
There are better ways to improve air quality than introducing LEZ R	3.694	0.604	0.859
Trust (α = 0.678)
I am confident that the Government would introduce LEZ appropriately	4.147	0.462	0.874	76.3
I trust that LEZ regulations would be developed and implemented effectively	4.221	0.586	0.874
Procedural Fairness (α = 0.607)
The government would be righteous to consider restricting the use of polluting vehicle in towns/cities	4.154	0.500	0.746	56.1
LEZ would be a suitable action to enhance local air quality	4.286	0.547	0.714
The policy makers will carefully select the types of vehicle restricted by a LEZ	4.261	0.514	0.786
Distributive Fairness (α = 0.897)
LEZ may benefit some people, but they would greatly impede others R	2.711	0.973	0.901	83.3
I believe the LEZ would unfairly affect some people R	2.853	1.011	0.928
LEZ would penalize those who are already in precarious circumstances R	3.070	1.157	0.909
Problem Perception (α = 0.601)
The usage of vehicles negatively affects people’s health and happiness	3.922	0.706	0.839	70.4
I am very concerned about the level of air pollution	4.196	0.514	0.839

^a 1 = Strongly disagree to 5 = strongly agree; α = Cronbach Alpha; ^R = Statement has been reverse coded; M = mean; SD = standard deviation; FL = factor loading; VE = variance explained; LEZ = Low Emission Zone.

. The descriptions cover the descriptive statistics (i.e., mean and standard deviation) and factor loadings for each statement within each construct. The confirmatory factor analysis (CFA) is used in each construct to reduce the dimension for each variable used in the analysis [18]. Most of the construct has reliability above the acceptable threshold of 0.6 [40] and only variables in PBC and outcome efficacy (OE) were less than 0.6. Therefore, PBC and OE are excluded from further analysis. The variance explained in all the constructs is above 55%. The description of the socio-psychological factors shows that most respondents have a positive perception of LEZ expansion and have a lower perception that LEZ will negatively impact Jakarta. Moreover, there is trust in government action in implementing LEZ.

LEZ restriction preference is described in

Table 3. Respondents’ Restriction Preference Characteristics.

Variables a	Mean	Std. Deviation
Alternatives Strategy for Restriction within LEZ Areas
Parking fare management	4.026	0.423
Congestion pricing	3.774	1.009
Emission based restriction	4.177	0.672
Odd-even	3.662	0.971
Alternatives of Vehicle that Restricted within LEZ Areas
Car	4.097	0.462
Motorcycle	4.107	0.660
Goods vehicle	4.019	0.818

^a 1 = Strongly disagree to 5 = strongly agree.

. The upper half of Table 3 explains the preference for the TDM strategy that can be implemented within LEZ areas. Emission-based restriction policy, currently implemented in KTTA, is the policy that has the highest value, meaning that this policy is the most preferred by travelers. While the less preferred policy is odd-even, this does not mean that it is not preferred to be implemented at all, since its value (3.662) is higher than the mid-value (3). Moreover, among the three modes (i.e., car, motorcycle and goods vehicle), motorcycle is the most preferred vehicle to be restricted, followed by car and goods vehicle. Since the number of motorcycles in Jakarta is tremendous and consequently contributes to congestion, this might influence travelers’ choice.

4. LEZ Expansion Public Perception Analysis

4.1. LEZ Implementation Acceptability

4.1.1. Analytical Framework

As explained, the structural model is developed based on the framework of behaviour and acceptance of sustainable policies suggested by Morton et al. [15]. The framework combines the TPB, NAM and the theory of sustainability policy acceptance [41] as illustrated on the left side of Figure 2 above. The ovals represent socio-psychological constructs and the arrows represent hypothesized relationships between the constructs. We have adjusted the framework to evaluate LEZ public acceptance. We tested each hypothesis, denoted by links between the constructs, to determine their applicability [15]. The proposed construct is divided into three parts. The right part of the construct represents the two theories of human behaviour: TPB and NAM [37,38]. The construct’s center represents the socio-psychological variables related to sustainability policy beliefs. The last part of the construct, on the left, underlines the distal construct that consists of trust in the government and problem perception. The framework has previously been tested within the context of LEZ or other TDM strategies’ acceptability [15,22,42]. The previous study underlines that the novelty of this construct is within the construct of emotional reactions, specifically in positive and negative affect and the relation of perceptions to costs, benefits, and risks of the policy with attitude formation.

Hierarchical SEM [29,35] was used to investigate the structural construct of LEZ acceptability (Figure 2). The center (PA, NA, PB, PC, PR, OE, PF, DF) and the right side of the construct (AT, PN and AC) are defined as endogenous variables. In contrast, all other variables were treated as exogenous (i.e., age, income, mode of transport, gender, PBC, and SN). However, since the reliability of OE and PBC is low, we exclude the OE and PBC from the analysis. From previous study findings [15,42], the absence of PBC in AC is understandable. All the value of endogenous variables and several exogenous (i.e., T, PP, SN, and PBC) is generated using confirmatory factor analysis (CFA), the factor loading described in Table 2.

The hypotheses are tested by applying the structural form, where the hierarchical SEM is used similarly to 2 Stage Least Square (2SLS). The hierarchical SEM is a multilevel regression analysis that can solve endogeneity issues using instrumental variables (IV). It is similar to the two and three-stage least square method (2SLS and 3SLS) [35]. Compared with traditional SEM, hierarchical SEM has several advantages. It can deal with categorical variables, needs less computational time compared to multilevel path analysis models [35], is able to include multiple endogenous variables or multiple equations, and is able the include more than one nesting effect [35]. Since it accommodates the concept of multilevel regression, three steps of Ordinary Least Square (OLS) similar to 3SLS analysis are performed. The first step of OLS is the investigation of the T and PP at the center of the construct. The second step is caried out after creating all the indicators within the first step of the regression analysis, called the creation of instrumental variable (IV). The incorporation of the estimated (^) of the center constructs (PA, NA, PB, PC, PR, OE, PF, DF) on the AT and PN model is the second step of 2SLS. After the second step of OLS, the creation of estimated (^) AT and PN is carried out and used for the last part of OLS, which is the estimation of LEZ acceptability (AC).

As in 3SLS, the inclusion of the estimated indications in all endogenous variables tries to tackle endogeneity problems using observed variables (age, gender, mode use and income). The detailed equations are Equations (1)–(11). As with 3SLS, the limitations of this study were run separately for the first, the second and the third stage. Thus, the estimated error terms are assumed not to be correlated. Therefore, the model sacrifices the simultaneous and the reciprocal effects. Since simultaneous effects are not expected to be a novelty in this study, this is not regarded as a serious problem. We used OLS for estimating all of the seven models for the first stage, as can be seen in the equations below:

$$P{A}_i=\left({\alpha }_1\right)+\left({\beta }_1{T}_i+{\beta }_{2}P{P}_i+{\beta }_{3}A{g}_i+{\beta }_{4}G{e}_i+{\beta }_{5}I{c}_i+{\beta }_{6}M{t}_i\right)+{\epsilon }_1$$

(1)

$$N{A}_i=\left({\alpha }_2\right)+\left({\beta }_7{T}_i+{\beta }_{8}P{P}_i+{\beta }_{9}A{g}_i+{\beta }_{10}G{e}_i+{\beta }_{11}I{c}_i+{\beta }_{12}M{t}_i\right)+{\epsilon }_2$$

(2)

$$P{B}_i=\left({\alpha }_3\right)+\left({\beta }_{13}{T}_i+{\beta }_{14}P{P}_i+{\beta }_{15}A{g}_i+{\beta }_{16}G{e}_i+{\beta }_{17}I{c}_i+{\beta }_{18}M{t}_i\right)+{\epsilon }_3$$

(3)

$$P{C}_i=\left({\alpha }_4\right)+\left({\beta }_{19}{T}_i+{\beta }_{20}P{P}_i+{\beta }_{21}A{g}_i+{\beta }_{22}G{e}_i+{\beta }_{23}I{c}_i+{\beta }_{24}M{t}_i\right)+{\epsilon }_4$$

(4)

$$P{R}_i=\left({\alpha }_5\right)+\left({\beta }_{25}{T}_i+{\beta }_{26}P{P}_i+{\beta }_{27}A{g}_i+{\beta }_{28}G{e}_i+{\beta }_{29}I{c}_i+{\beta }_{30}M{t}_i\right)+{\epsilon }_5$$

(5)

$$D{F}_i=\left({\alpha }_6\right)+\left({\beta }_{31}{T}_i+{\beta }_{32}P{P}_i+{\beta }_{33}A{g}_i+{\beta }_{34}G{e}_i+{\beta }_{35}I{c}_i+{\beta }_{36}M{t}_i\right)+{\epsilon }_6$$

(6)

$$P{F}_i=\left({\alpha }_7\right)+\left({\beta }_{37}{T}_i+{\beta }_{38}P{P}_i+{\beta }_{39}A{g}_i+{\beta }_{40}G{e}_i+{\beta }_{41}I{c}_i+{\beta }_{42}M{t}_i\right)+{\epsilon }_7$$

(7)

where PA_i = positive affect for a traveler’s i; NA_i = negative affect for a traveler’s i; NA_i = positive affect for a traveler’s i; PB_i = perceived benefit for a traveler’s i; PC_i = perceived cost for a traveler’s i; PR_i = perceived risk for a traveler’s i; DF_i = distributive fairness for a traveler’s i; and PF_i = procedural fairness for a traveler’s i. The right-hand side of Equations (1)–(7) consists of: T_i = trust in government of traveler’s i, PP_i = problem perception of traveler’s i, Ag_i = age of traveler’s i, Ge_i = gender of traveler’s i, Ic_i = income of traveler’s i and Mt_i = mode of transport of traveler’s i.

In the second stage model (for AT and PN), this study incorporated estimated indicators of the first stage, i.e., $\widehat{P{A}_i}$, $\widehat{N{A}_i}$, $\widehat{P{B}_i}$, $\widehat{P{C}_i}$, $\widehat{P{R}_i}$, $\widehat{D{F}_i}$, and $\widehat{P{F}_i}$. Endogeneity problems were expected to be tackled by applying all the estimated indicators instead of the non-estimated indicators and also representing the recursive structure of the framework. Moreover, all of the models’ equations in the second stage are provided below:

$$\begin{gathered} A{T}_i=\left({\alpha }_8\right)+\left({\beta }_{41}\widehat{P{A}_i}+{\beta }_{42}\widehat{N{A}_i}+{\beta }_{43}\widehat{P{B}_i}+{\beta }_{44}\widehat{P{C}_i}+{\beta }_{45}\widehat{P{R}_i}+{\beta }_{46}\widehat{P{F}_i}+{\beta }_{47}A{g}_i \\ +{\beta }_{48}G{e}_i+{\beta }_{49}I{c}_i+{\beta }_{50}M{t}_i\right)+{\epsilon }_8 \end{gathered}$$

(8)

$$\begin{gathered} P{N}_i=\left({\alpha }_9\right)+\left({\beta }_{51}\widehat{P{A}_i}+{\beta }_{52}\widehat{N{A}_i}+{\beta }_{53}\widehat{P{B}_i}+{\beta }_{54}\widehat{P{C}_i}+{\beta }_{55}\widehat{P{R}_i}+{\beta }_{56}\widehat{P{F}_i} \\ +{\beta }_{57}A{g}_i+{\beta }_{58}G{e}_i+{\beta }_{59}I{c}_i+{\beta }_{60}M{t}_i\right)+{\epsilon }_9 \end{gathered}$$

(9)

The estimated indicators of attitude ($\widehat{A{T}_i}$) and personal norm ($\widehat{P{N}_i}$) are incorporated in the last stage of 2SLS for LEZ acceptability (AC). In addition, the role of PP and T to the AC is also added in the last stage, as illustrated in Figure 2. The last models are provided below:

$$\begin{gathered} AC{1}_i=\left({\alpha }_{10}\right)+\left({\beta }_{61}\widehat{A{T}_i}+{\beta }_{62}\widehat{P{N}_i}+{\beta }_{63}SN+{\beta }_{64}A{g}_i+{\beta }_{65}G{e}_i+{\beta }_{66}I{c}_i \\ +{\beta }_{67}M{t}_i\right)+{\epsilon }_{10} \end{gathered}$$

(10)

$$AC{2}_i=\left({\alpha }_{11}\right)+\left({\beta }_{68}{T}_i+{\beta }_{69}P{P}_i+{\beta }_{70}A{g}_i+{\beta }_{71}G{e}_i+{\beta }_{72}I{c}_i+{\beta }_{73}M{t}_i\right)+{\epsilon }_{11}$$

(11)

Moreover, the stepwise method is used for both models, which is the step-by-step iterative construction of a regression model. This involves the selection of independent variables to be used in a final model [18]. We still retained several insignificant variables in the models. This is due to the influence of the variables to the goodness-of-fit of the models after a review during the stepwise process [18].

4.1.2. Model Estimation for LEZ Acceptability

The model estimation for the first stage is described in

Table 4. Estimation model for first step hierarchical SEM.

Variables	Positive Affect		Negative Affect		Perceived Benefit		Perceived Cost		Perceived Risk		Procedural Fairness		Distributive Fairness
	Unstd. β	Std. β	Unstd. β	Std. β	Unstd. β	Std. β	Unstd. β	Std. β	Unstd. β	Std. β	Unstd. β	Std. β	Unstd. β	Std. β
Constant	−0.052	0.000	−0.918 a	0.000	0.150	0.000	−0.494 a	0.000	−0.231	0.000	0.090	0.000	0.640 a	0.000
Trust	0.364 a	0.364	−0.164 a	−0.164	0.468 a	0.468	−0.055	−0.055	0.079 a	0.079	0.463 a	0.463	0.120 a	0.120
Problem Perception	0.224 a	0.224	0.028	0.028	0.101 a	0.101	0.127 a	0.127	0.070 a	0.070	0.208 a	0.208	−0.156 a	−0.156
Private transport users (D)	−0.336 a	−0.152	0.276 a	0.125	−0.230 a	−0.104	0.163 a	0.074	−0.011	−0.005	−0.312 a	−0.141	0.022	0.010
Male travellers (D)	0.144 a	0.070	0.096	0.047	0.046	0.023	−0.039	−0.019	−0.104	−0.051	0.036	0.018	−0.028	−0.013
Age	0.013 a	0.100	0.013 a	0.095	−0.005	−0.040	0.020 a	0.155	0.012 a	0.089	0.006	0.042	−0.018 a	−0.139
Income	−0.118 a	−0.082	0.105 a	0.073	0.020	0.014	−0.069	−0.048	−0.031	−0.022	−0.064	−0.044	−0.001	−0.001
R2	0.31		0.08		0.30		0.05		0.03		0.39		0.04
Adjusted R2	0.30		0.07		0.29		0.04		0.02		0.39		0.04
ANOVA (F)	51.90		9.90		48.58		5.68		3.24		74.25		5.33

^a significant at 5%; (D): 1 if yes and 0 otherwise.

. All of the models have a significant F value, meaning that a significant relationship exists between dependent and independent variables. Moreover, the adjusted R² for the models ranging from 0.03–to 0.39 implies the data variance explained by the models. The models found the influence of trust and problem perception on the social-psychological construct. While trust is found to significantly influence all psychological constructs except perceived cost (PC), problem perception does not influence the negative affect (NA). Personal and travel characteristics were also found to influence the construct. Interestingly, age has a positive impact on PA, NA, PC and PR, while having a negative impact on DF. Meanwhile, travelers with higher income or using private transport (i.e., car or motorcycle) tend to have a lower perception of the LEZ impact, referring to the influence of PA and NA.

The second stage model estimations are described in

Table 5. Estimation model for second step hierarchical SEM.

Variables	Attitude		Personal Norm
	Unstd. β	Std. β	Unstd. β	Std. β
Constant	−0.334		−0.256
Positive Affect ^	0.308	0.171	0.312	0.174
Negative Affect ^	−0.098	−0.026	−0.519	−0.139
Perceived Benefit ^	−14.873 a	−8.137	−8.316 a	−4.550
Perceived Cost ^	−37.097 a	−8.314	−21.088 a	−4.726
Perceived Risk ^	48.694 a	7.641	27.571 a	4.326
R2	0.29		0.36
Adjusted R2	0.28		0.36
ANOVA [F]	46.44		66.44

^a significant at 5%; ^: Estimated.

. The R² parameters show a value in the range of 0.29–0.36, while the F value exceeds the significance threshold. The models show that PB, PC and PR significantly influence AT and PN. Interestingly, PR has a positive influence on AT and PN, while for PB and PC it has a negative influence. This is understandable, since the PC has underlined that the cost of LEZ is not effective. On the other hand, concern about the risk in implementing LEZ is found not to decrease the attitude and the influence of personal norms concerning implementation.

Moreover,

Table 6. Estimation model for final step hierarchical SEM.

Variables	Acceptance 1		Acceptance 2
	Unstd. β	Std. β	Unstd. β	Std. β
Constant	0.199		0.000
Trust			0.550 a	0.550
Problem Perception			0.230 a	0.230
Attitude ^	−2.580 a	−0.936
Personal Norm ^	3.696 a	1.276
Social Norm	0.112 a	0.112
Private transport users (D)	−0.164 a	−0.074
Male travellers (D)	−0.044	−0.021
Age	0.008 a	0.063
Income	−0.093 a	−0.064
R2	0.59		0.46
Adjusted R2	0.59		0.46
ANOVA (F)	125.29		298.15

^a significant at 5%; (D): 1 if yes and 0 otherwise; ^: Estimated.

presents the last stage of the analysis. The first AC model shows the effect of estimated PN and AT as well as SN on the AC, while the second AC model shows the role of T and PP on AC. Both models have significant F values and 0.59 and 0.46 are the values of R². Attitude has no positive implication on acceptability, while it is a significant positive influence on personal norms. Social norms were also found to positively influence the acceptability of LEZ. Meanwhile, higher income earners, younger people and private transport users show lower acceptability of the LEZ implementation. For the second AC model, trust in government has a significant positive influence on the acceptability of LEZ. The strong perception of the problem in Jakarta was also found to influence the higher acceptability of LEZ expansion.

4.2. LEZ Restriction Preferences

4.2.1. Analytical Framework

In this section, we discuss the analysis of the restriction preferences within the LEZ location. The framework of the model is described on the right side of Figure 2. Since the investigation is interested in various social-psychological indicators, the analysis is not developed simultaneously to or recursively with the previous LEZ acceptability analysis. As both dependent variables consist of several indicators captured on a five-point Likert scale, the k-means cluster analysis [18,36] is performed for vehicle-based and strategy-based restrictions. The k-means clustering method is performed to partition the observations into a number of clusters that account for the distinctiveness of observations [43]. Furthermore, this classification group is then used in the subsequent analysis of restriction preference using MNL, as used by others [44].

An MNL specification is appropriate for modelling the LEZ restriction preference when the outcomes of the dependent variables are categorical and non-ordered. The standard form of the multinomial logistic regression model appears in Equation (12) [45].

$$P(Y=j|x)=\frac{e^{g_j\left(x\right)}}{{\sum }_{k=0}^n{e}^{g_k\left(x\right)}}$$

(12)

In Equation (12), the logit, g_j, is linear in its parameters, consisting of β_0j + β_ijX_i, and the logit transformation, P(Y = j|x) is the probability of the outcome j. For estimation of the parameters in gj, we set one of the categories to zero as a reference category and re-interpret the remaining parameters to represent preference differences relative to the reference category [45]. For example, the logit formula for a multinomial logistic regression model with three categories, can be represented by Equation (13).

$$\{\begin{array}{c}{g}_1={\beta }_{01}+{\beta }_{11}{X}_1+\dots +{\beta }_{i1}{X}_i+{\epsilon }_1\\ g_2={\beta }_{02}+{\beta }_{12}{X}_1+\dots +{\beta }_{i2}{X}_i+{\epsilon }_2\\ g_0=0\end{array}$$

(13)

where β_0j is the constant term that is associated with the outcome j, β_ji the unknown parameters that are associated with choice j and covariates i, and X_i is the vector (which in the analysis are the socio-psychological factors, LEZ acceptance and travelers’ personal and travel characteristics) associated with the covariates i. ε_i is the error term of outcome j, which reflects unobserved explanatory variables, which in turn influence the response variable, that is, the logit value in this case. The Maximum Likelihood Estimation (MLE) method has been used to determine the coefficient values in the logit formula.

4.2.2. Model Estimation for Strategy-Based Restriction

We established several cluster groups of travel changes by using the k-means clustering method. To do this, we utilized the stated intentions of whether respondents will or will not change the frequencies of various trip purposes in the post-COVID situation. The response is on a five-point Likert scale ranging from significantly decrease to significantly increase.

The cluster analysis is used to change the trip frequencies post- pandemic for six purposes: work/school trips, grocery shopping, electronics/fashion shopping, out-of-home dining, recreation and socialising. Based on the Within Sum Square (WSS) (Appendix A) [35,46], which indicates the cluster validation of consistency within the data, our analysis shows that the optimal number of clusters based on these attributes will be equal to four. The title of each cluster group was determined after evaluating the estimated value in the upper half of

Table 7. Cluster Analysis of Type of Restriction.

Type of Restriction	Cluster Group
	P, CC and E	All Strategy	P and E	P, E, and OE
Parking fare management	3.92	4.13	3.96	3.83
Congestion pricing	3.78	4.44	2.32	2.48
Emission based restriction	4.10	4.36	3.45	4.41
Odd-even	2.78	4.33	2.27	4.20
Proportion of sample	23.1%	51.9%	14.4%	10.7%
Distances between Final Cluster Centres
Cluster Group	P, CC and E	All Strategy	P and E	P, E, and OE
P, CC and E		1.71	1.69	1.95
All Strategy	1.71		3.10	1.99
P and E	1.69	3.10		2.17
P, E and OE	1.95	1.99	2.17

P: Parking; CC: Congestion Pricing; OE: Odd-even; E: Emission.

. The scale of preference restriction is based on the number of strategies/types of strategies users prefer. The cluster composition shows that the cluster ‘all strategy’ contained the most respondents among the three (51.9%). In contrast, the respondents who prefer parking, emission-based restriction and odd-even combinations are in the smallest cluster (10.7%). Table 7 also shows the Euclidean distances between the final cluster centers. The second and third clusters are the most distinct from each other.

The cluster analysis results are used to define the dependent variables in a multivariate analysis. The estimated model of strategy-based restriction within the LEZ period is presented in

Table 8. Model Estimation for Type of Restriction.

Variables	Type of Strategy
	P, CC and E		P and E		P, E and OE
	Estimate	T-Stat	Estimate	T-Stat	Estimate	T-Stat
Constant	−0.329	−0.386	−1.043	−1.099	−2.323 a	−2.423
Acceptance of LEZ	−0.931 a	−4.488	−0.358	−1.405	−0.051	−0.203
Trust	−0.028	−0.164	−0.486 a	−2.232	0.064	0.289
Problem Perception	−0.964 a	−6.373	−0.377 a	−2.126	−0.202	−1.122
Positive Affect	−0.315 a	−1.922	−0.311	−1.459	−0.371	−1.822
Negative Affect	0.037	0.227	0.242	1.495	0.421 a	2.677
Perceived Benefit	3.090	0.155	1.974	0.192	0.733	0.188
Perceived Cost	−3.610	0.192	−2.862	0.240	−0.699	0.246
Perceived Risk	2.407	0.167	3.320	0.233	2.413	0.224
Outcome Efficacy	−1.019	0.156	−1.333	0.218	−1.624	0.200
Procedural Fairness	2.858	0.163	−2.971	0.222	0.122	0.200
Distributive Fairness	0.019	0.183	−2.722	0.250	−0.903	0.239
Attitude	−0.298 a	−2.028	−0.023	−0.102	0.005	0.027
Personal Norm	−0.219	−1.234	−0.184	−0.894	−0.344	−1.651
Social Norm	−0.032	−0.223	−0.211	−1.040	−0.419 a	−2.173
Private transport users [D]	−0.017	−0.053	−0.518	−1.485	−0.366	−0.998
Male travellers [D]	−0.408	−1.619	−0.800 a	−2.426	−0.335	−1.028
Age	−0.039 a	−2.223	−0.015	−0.712	−0.001	−0.041
Income	0.202	1.110	0.133	0.711	0.326	1.706
McFadden R2	0.223
LL-β [Chi-Square; p-value)	1300.63 [375; 0.000]
% Correct	61.8

Reference Category: All strategies used within LEZ areas; P: Parking; CC: Congestion Pricing; OE: Odd-even; E: Emission; _a: significant at 5%.

. The interpretation of the multinomial logistic model is based on the reference category: ‘all the strategy implemented in the LEZ areas’. With four groups and one as a reference category, three models were estimated each for every combination of strategy. The model fitness is shown by the overall model fitness test (2LL). With a p-value of less than 5%, it indicates that the null hypothesis, which states that there is no difference between the model with independent variables and the model without, can be rejected. In other words, the independent variables could improve the model fitness and explain the variance. The cross-tabulation test shows that more than 61% of data were correctly classified in the models. With a number of significant variables, the model is considered suitable for explaining the variability in the data. The interpretation below is largely focused on variables that have been found significant and depend on the sign of the coefficient as well as whether the covariate value increases or decreases. If the sign is positive and the covariate increases, then it is likely that the subject stays with the column group, or, if the sign is positive but the covariate decreases, it is likely that the subject is in favour of the alternative (i.e., the reference group).

The acceptance of LEZ is found to positively influence the higher number of strategies that are implemented, as it tends to prefer all strategies implemented. Various socio-psychological factors influence the preferred strategies to be implemented. The respondents with higher PA and who have a positive attitude toward LEZ implementation tend to prefer all strategies implemented. However, the respondents concerned about the NA of LEZ implementation tend to prefer to exclude CC from all strategies. External parties’ concerns (i.e., family or friends) on the environmental issue also positively influence the number of strategies implemented. In line with the external factors, problem perception of the environment also influenced the higher number of strategies implemented in LEZ areas. From the personal and travel characteristics of travellers, male travellers were found to prefer to implement more strategies than female travellers. Interestingly, older people also tend to support more of the strategies implemented within LEZ areas.

4.2.3. Model Estimation for Vehicle-Based Restriction

We analyzed the vehicle-based restriction preference within LEZ areas, and k-means clustering was used to develop the cluster groups of three vehicles (i.e., car, motorcycle, and goods). Similar to the previous model, the WSS classifies the vehicle-based restriction into three groups (Appendix A). The interpretation of the cluster group is carried out by evaluating the value in the upper side of

Table 9. Cluster Analysis of Type of Vehicle.

Type of Vehicle	Cluster Group
	PrivateVehicle	All Vehicle	Car and Goods
Car	3.96	4.19	3.75
Motorcycle	3.86	4.35	2.84
Goods	2.77	4.41	4.08
Proportion of sample	21.9%	69.1%	9.0%
Distances between Final Cluster Centres
Cluster Group	PrivateVehicle	All Vehicle	Car and Goods
Private Vehicle		1.727	1.678
All Vehicle	1.727		1.604
Car and Goods	1.678	1.604

, with the scale value of 1 representing a strong disagreement with implementing a restriction for that vehicle and 5 representing an agreement to implement the restriction for that vehicle. Regarding the cluster proportions, all vehicles restricted contributes to the highest proportion of individuals with 69.1% proportions, whereas cars and goods vehicles show the lowest proportion.

The estimated models of multinomial logistic regression of online-activity changes during the new normal are shown in

Table 10. Model Estimation for Type of Vehicle.

Variables	Type of Vehicle Restricted
	All Vehicle		Car and Goods
	Estimate	T-Stat	Estimate	T-Stat
Constant	0.797	1.001	−2.106	−1.660
Acceptance of LEZ	0.940 a	4.643	0.501	1.708
Trust	−0.042	−0.267	−0.099	−0.421
Problem Perception	0.587 a	4.106	−0.029	−0.128
Positive Affect	−0.037	−0.222	0.138	0.495
Negative Affect	−0.228	−1.615	−0.308	−1.233
Perceived Benefit	−0.295 a	−1.998	−0.621 a	−2.646
Perceived Cost	0.916 a	4.822	1.321 a	4.527
Perceived Risk	0.632 a	3.860	−0.415	−1.586
Outcome Efficacy	0.157	1.029	−0.049	−0.190
Procedural Fairness	−0.107	−0.699	0.343	1.466
Distributive Fairness	0.471 a	2.520	0.785 a	2.672
Attitude	0.072	0.488	0.080	0.328
Personal Norm	0.088	0.506	−0.468	−1.733
Social Norm	−0.189	−1.367	−0.056	−0.257
Private transport users (D)	0.098	0.334	−0.033	−0.069
Male travellers (D)	0.076	0.306	−0.145	−0.359
Age	0.031 a	1.908	−0.003	−0.095
Income	−0.113	−0.689	0.293	1.220
McFadden R2	0.256
LL-β (Chi-Square; p-value)	837.892 (289.017; 0.000)
% Correct	79.10%

Reference Category: Only private vehicles ae restricted; _a: significant at 5%.

. The model is fit for purpose since the model with predictors is better than the model without predictors. It is also notable that the cross-tabulation test shows that more than 79% of data were correctly classified in the model. The overall model results are highly acceptable, with significant effects found in various variables and based on the statistical and practical significance. Moreover, the category ‘the private vehicle is restricted’ is the reference category for interpreting the model.

Similar to the type of restriction model, various socio-psychological factors found to have shaped the choice of vehicle restriction (Table 10). Travellers who predict a huge cost for implementation of the LEZ and the risk of LEZ not being effective tend to choose to restrict all vehicles. Interestingly, optimistic travellers about the benefits of LEZ are associated with a preference towards the private vehicle restriction. Travellers who believe in fairness of distributional impact in LEZ implementation tend to believe that restriction should be implemented for all vehicles. Higher acceptance of LEZ expansion also influences traveller to prefer all vehicle restrictions. Moreover, older people tend to believe all vehicles should be subject to restriction in the LEZ areas.

4.3. Discussion

This study examines the effects of socio-psychological factors on the support for LEZ replication and the alternatives to restriction policy. Generally, the study found that understanding and awareness of the environmental problem shape the acceptability and increase the preference for the LEZ socio-psychological construct. Concern about the urgency of the environmental issue has been widely investigated and found to shape positive appreciation, as well as the adoption of various sustainable policy strategies, including TDM strategies [12,22,47]. Interestingly, this includes predicting the benefit, cost and risk of LEZ implementation. While some positive outcomes are expected by travelers with LEZ implementation, they are also concerned about the cost of implementing LEZ, which will burden the city’s fiscal capacity. The perceived cost is also found to negatively shape personal norms regarding approval of the LEZ implementation, meaning that the higher the concern about cost, the lower the acceptance of LEZ extension. Attitude was also shaped by this perceived cost, which is similar to what Morton et al. [15] found in Scotland. However, contrary to Morton et al. [15], in Scotland the perceived benefit is not found to influence the personal norm positively. It appears that the infrastructure for walking and cycling in Jakarta has been a concern for people, and personalization of the small areas of the LEZ will influence better health. Unlike Scotland, which already has good infrastructure quality in supporting activities in LEZ areas [13] and larger LEZ areas, LEZ in KTTA was implemented in small areas with low-quality of pedestrian infrastructure in the first place. While the Jakarta government has continued to improve this infrastructure, there appears to be a concern about the scale of LEZ, which might not be positively influencing greater health, but only improving local air pollution.

The acceptance of LEZ implementation is found to be shaped by personal norms, meaning that the indirect effect of the risk of LEZ being ineffective and the fairness of implementation impact are important. This is in line with previous LEZ [13] and TDM [42] strategy (i.e., rising fuel price and CC) studies that investigate the public response. Social norms also play a significant role in acceptability, indicating that the thoughts of important others may exert pressure when travelers are translating their views of LEZ implementation. Within environmental behaviour (i.e., policy to support climate change actions), we expect conscientious persons to follow social norms [35] and to transform their concerns into actions. Moreover, the importance of trust in government to LEZ acceptability is also found to be in line with several implementations of LEZ worldwide [4,11]. In London, communication about LEZ and CC was introduced by politicians and followed up by clear communications about the plan of implementation, mitigation strategy and equality impact [4]. This socialization plan is a way to increase trust in the government concerning LEZ implementation.

The study also underlined that age strongly affects support for LEZ expansion. Consequently, specific public communication strategies are required to better address the need to improve public support for specific age groups. Findings of this study show that older people are more positively associated with support for LEZ expansion than younger people. Meanwhile, younger people tend to be more politically engaged and can have a substantial ability to drive the policy design and implementation. As the younger people are more inclined to view LEZ implementation negatively, targeting this demographic group in awareness-raising campaigns that emphasize the positive implications of LEZ may increase the chances for policy approval by the government. These younger people, or millennials, make up a substantial proportion of society. Empowering millennials to adopt sustainable travel behaviour, especially on travelling in LEZ areas with low or no emission transport modes, is also important for influencing future transport demand. In addition, the negative perception of the various socio-psychological constructs of LEZ implementation (i.e., PA and NA) as well as acceptability are important factors that should be taken. The higher income group, mainly using private transport, might be more impacted by the policy. Care should be taken to ensure that the implementation of LEZ policy does not adversely affect the well-being and to communicate the importance and the planning of LEZ to this demographic group.

Generally, the choice of the restriction (i.e., types of vehicles or strategies) from a public perspective is shaped by their acceptability. Higher acceptability leads to full-restriction preference (i.e., all vehicles and all strategies implemented). This means that actions to increase LEZ acceptability by persuading change to positive personal and social norms concerning LEZ implementation are crucial. Interestingly, while travelers who understand the benefit of LEZ tend to support restriction only to the private vehicle, those concerned about the risk and cost of LEZ tend to support all vehicle restrictions. Even though people might still be concerned about whether or not LEZ should be implemented, their belief in its benefits to their urban life quality seems to be able to overcome this. Care should be taken on this issue, underlining that LEZ implementation should be assessed in greater detail and communicated to the people. With the link between environmental concern and climate actions [48,49], concern with the environmental issue in Jakarta has been associated with a higher number of restrictions within LEZ areas. Travelers might believe that, if all TDM strategies are in place and all vehicles are restricted, it will contribute significantly to air quality improvement. Thus, the government should take these measures, given that the issue is crucial [26,50].

While this research produced several important findings, we have also laid out the limitations of this study. Considering that data collection was done amid a pandemic where the number of people who travel to the KTTA is significantly lower, this study captures the travel behaviour of the KTTA users during this pandemic. However, while previous studies have underlined the effect of the pandemic on future travel behaviour [51,52,53] and perception of the environment [54,55], further studies of the detailed implementation of LEZ in the post-COVID-19 context should be conducted. This study also does not consider several travel behaviour factors, such as trip frequency, etc., in predicting the users’ support for LEZ expansion. Moreover, this study focus on the general commuters of Jakarta and the effect of LEZ implementation on a specific group of users (i.e., based on the type of modes) is an area that can be explored in future research. Further studies that accommodate these variables will provide more insight into the implementation of LEZ, which could help formulate a more comprehensive sustainable transportation policy in Jakarta.

5. Conclusions and Recommendations

Driven by the critical air quality issue, the Jakarta Government has implemented LEZ in the KTTA and is planning to expand its implementation to other locations. This study aims to explore the social-psychological factors that influence LEZ acceptance and the types of restrictions that are preferred to be enforced within the LEZ areas. The findings of this study are twofold. First, the support for LEZ expansion to other locations was found to be shaped by the government’s trust, the level of environmental concern, as well as personal and social norms regarding LEZ implementation. The indirect effect of fairness (i.e., distributive and procedural) on acceptance is also found to shape the acceptability of LEZ. Second, the perception of the type of restriction (i.e., combinations of the vehicle types and/or the TDM strategies) that should be taken within LEZ areas is associated with acceptability and environmental concern. The risk and cost of LEZ implementation have become a concern for travelers, particularly younger and wealthier people.

Based on these findings, several recommendations are proposed for the future implementation plans of LEZ in Jakarta or other cities that have a plan to implement or expanded LEZ policy. Firstly, as LEZ acceptability and restriction preference is shaped by environmental concern and personal and social norms, a public campaign to increase the awareness of urban environmental issues should be carried out continuously. The approach should be designed specifically for each target, younger and the wealthier people, considering that they have different perceptions of benefits and risks. Secondly, the mediated effect of perceived risk on personal attitudes has prompted the need for further analysis followed by a series of public consultations on the LEZ implementation’s risks and costs. Lastly, to build people’s trust in the government, we propose a participatory approach in planning, monitoring and evaluating the LEZ implementation, including how generated revenue from the restriction would be utilized. With transparent information about each step of implementing LEZ, the public can be informed comprehensively about the progress and, consequently, this can increase the public ownership of the policy itself. In an international context, as suggested by Morton et al. [15], openness in the planning process has been used as a means for the government to identify the extent of equity implication for such a scheme. Once it is implemented, the people must periodically be informed about how much their contributions have made a difference. This would gradually shape their personal and social norms and transform these into “grass-root” collective action instead of “top-down” policy.