Consumer Preference and Willingness-to-Pay for Formal Recycling of Electric Vehicle Batteries: A Discrete Choice Experiment in China

OriginalConsumer preference and willingness-to-pay for formal recycling of electric vehicle batteries: A discrete choice experiment in China Xingjun Huang a,b , Song Lei c,* , Feng Liu d , Yan Li e , Fuli Zhou f,g , Ming K.

OriginalLim h,i a School of Modern Posts, Chongqing University of Posts and Telecommunication, Chongqing, PR China b Department of Land Surveying and Geo-Informatics, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China c School of Management Science and Engineering, Chongqing Technology and Business University, Chongqing, PR China d School of Economics and Business Administration, Chongqing University, Chongqing, PR China e Chongqing Changan Minsheng APLL Logistics Co., Ltd, Chongqing, PR China f School of Automation Science and Engineering, South China University of Technology, Guangzhou, PR China g College of Economics and Management, Zhengzhou University of Light Industry, Zhengzhou, PR China h Adam Smith Business School, University of Glasgow, Glasgow, United Kingdom i Department of Industrial Engineering Department, Khon Kaen University, Thailand

OriginalKeywords:

OriginalElectric vehicle battery recycling Discrete choice model Recycling market preferences Recycling policies Consumer behavior A B S T R A C T The burgeoning electric vehicle (EV) market poses a substantial challenge to battery recycling systems, yet understanding EV battery recycling behavior from the demand side remains limited.

OriginalPrevious studies have analyzed perceptual or attitudinal factors, neglecting the observable attributes of EV battery recycling. To this end, we proposed a discrete choice model to investigate the differences between formal and informal recycling behaviors, identifying consumer preferences and willingness to pay.

OriginalBy analyzing 1190 sample data collected from Chongqing, China, we find that: (1) The formal recycling market exhibits greater sensitivity to prices compared to the informal recycling market.

Original(2) The formal recycling market favors recycling by EV battery producers, whereas the informal recycling market shows the least preference for recycling by automobile pro ducers.

Original(3) Door-to-door recycling services are the most effective in facilitating the transition from informal to formal recycling markets for EV batteries. (4) Capacity subsidy policies outperform one-time fixed subsidy policies in incentivizing formal recycling.

Original(5) The formal recycling market for EV batteries necessitates "trace ability to the recycling outlet", as opposed to being untraceable. (6) The high-awareness group exhibits greater sensitivity to government policies compared to those with lower environmental concerns and less knowledge of EV battery recycling.

Original1. Introduction

OriginalThe increasing numbers of electric vehicles (EVs) are posing a severe waste management challenge for recycling systems at end-of-life (Harper et al., 2019, p. 75; Ding et al., 2020; Dong and Ge, 2022; Li et al., 2023).

OriginalAlthough EV lithium-ion batteries can theoretically last 8–10 years, their capacity drop to 80% or less after 5–6 years, making them unsuitable for powering EVs in fact (Plato Yip, 2022).

OriginalGlobally, many EV batteries are being scrapped, and by 2035, approximately 10 million will be unable to use (Plato Yip, 2022). The global EV battery recycling market is forecast to increase from a market value of $26.9 billion in 2023 to $54.3 billion in 2030, a growth of 10.5% (MarketsandMarkets, 2023).

OriginalOf these, China is the largest market for EV battery recycling, handling around 500,000 metric tons followed by the United States and Europe with 200,000 metric tons each (Bruna Alves, 2024). However, about 80% of decommissioned batteries went to the black market in China resulting in severe environmental pollution (Xinhuanet, 2021).

OriginalSuch a market is also known as "informal recycling", where small workshops without recycling qualifications and unprofes sional dismantling practices profit through high-priced recycling and lower-priced disposal (Tang et al., 2023; Tian et al., 2024).

OriginalThe coun terpart is "formal recycling, " carried out by government-issued EV battery recycling white-listed companies and regulated by the govern ment. Improving formal recycling of retired EV batteries has therefore * Corresponding author.

OriginalSchool of Management Science and Engineering, Chongqing Technology and Business University, Chongqing, PR China. E-mail addresses: xjhuang@cqupt.edu.cn (X. Huang), leisong@ctbu.edu.cn (S.

OriginalLei), liufengcqu@163.com (F. Liu), 2514691564@qq.com (Y. Li), deepbreath329@ outlook.com (F. Zhou), ming.lim@glasgow.ac.uk (M.K. Lim). Contents lists available at ScienceDirect Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman Received 7 March 2024; Received in revised form 3 August 2024; Accepted 8 August 2024 Journal of Environmental Management 370 (2024) 122180 Available online 9 September 2024 0301-4797/© 2024 Elsevier Ltd.

OriginalAll rights are reserved, including those for text and data mining, AI training, and similar technologies. become ever more pressing. Policy development is also accelerating globally, aiming to stimulate the development of the EV battery recycling industry and regulate the market.

OriginalFor example, the California Lithium Ion Battery Recycling Advisory Group has launched a support program for the disposal of endof-life EV batteries and the recycling of critical materials for clean en ergy vehicle batteries (Plato Yip, 2022).

OriginalChina has also issued four policies for EV battery recycling (Huang et al., 2022b). However, aca demic research on EV battery recycling is relatively limited, especially in understanding formal and informal recycling behavior.

OriginalExisting studies mainly concentrate on the supply chain management of EV battery recycling, which includes aspects like recycling pricing (Ding et al., 2020; Wang et al., 2022; Zhang et al., 2023) and recycling outlet se lection (Zhu et al., 2020; Zhang et al., 2021; Lin et al., 2023).

OriginalA few studies empirically explore the underlying mechanism of EV battery recycling behavior (Tripathy et al., 2022; Li et al., 2023; Tang et al., 2023), but these studies revolve latent variables (such as consumer perceptions of ambiguity), lacking quantitative analysis of observed attributes.

OriginalIn response, our study proposes a discrete choice model to identify the differences between formal and informal recycling behaviors and their significant influencing factors.

OriginalThe model includes nine observed attributes that have never been empirically analyzed before, such as recycling price, market penetration rate of formal recycling outlets, trade-in policy, carbon credit reward policy, recycling subsidy policy, traceability of recycling information, recycling entity, and recycling

Originalmethod. We collected 1190 samples in Chongqing to validate the

Originalfeasibility of our framework. To the best of our knowledge, our study is the first to employ the discrete choice model to investigate formal and informal recycling preferences for EV batteries.

OriginalThe proposed frame work complements existing literature on understanding EV battery recycling behavior. Additionally, our study addresses the following questions: (1) What factors affect formal and informal recycling for EV batteries, especially policy preferences?

Original(2) How does consumer het erogeneity impact their recycling behaviors? (3) How can we facilitate the transition of the EV battery recycling market from informal to formal recycling?

Original2. Literature review

OriginalOur study is related to two main areas: the public's recycling pref erences for EV batteries and existing recycling policies, which identify potential research gaps.

Original2.1. Public recycling preferences for EV batteries

OriginalMany factors usually influence public choice behavior, and factor identification is the basis for revealing behavioral mechanisms and a prerequisite for consumer demand guidance or behavior intervention.

OriginalFactors influencing choice behavior could be divided into perceptual (attitudinal) indicators, socio-demographic characteristics, and observ able attributes (Swait, 1994).

OriginalIn this section, we focus on perceptual indicators and observable attributes, but after a literature review, we found that existing research on EV battery recycling mainly focuses on perceptual factors, lacking research on observable attributes.

OriginalTherefore, EV battery recycling behavior research is in dire need of identifying and quantifying observable attributes. The recycling field has extensive literature on the impact of perceptual indicators on recycling behavior.

OriginalFor example, Hage et al. (2009) found that facility convenience and moral norms matter in driving Swedish households to recycle packaging waste like paper, glass, plastic and metal.

OriginalIzagirre-Olaizola et al. (2015) argued that perceived consumer effectiveness significantly improves recycling behavior by university students, followed by environmental knowledge.

OriginalGeiger et al. (2019) indicated that recycling self-identity, personal norms towards recycling and perceived recycling behavioral control are significant predictors of recycling behavior.

OriginalBai and Lin (2022) found that classified knowledge and publicity help the public pay more for waste recycling. Chao et al. (2023) indicated that environmental concern (EC) has a significant positive impact on recycling behavior among university students.

OriginalChang et al. (2023) found that EC and information security risk perception significantly impact consumers' recycling of old phones. For retired EV battery recycling, factors influencing recycling behavior in other areas also have similar effects.

OriginalFor instance, Dong and Ge (2022) suggested that perceived behavioral control, subjective norms and value-belief-norm causal chains are the most significant positive facilitators among EV battery recycling behavior.

OriginalLi et al. (2023) found that perceived convenience, policy incentives, environmental con sciousness, and subjective norms are essential drivers influencing young consumers to choose formal EV battery recycling.

OriginalTang et al. (2023) identified economic incentive, subjective norm, perceived behavior control and self-identity as the main antecedent variables predicting users' EV battery recycling behavior.

OriginalOn the other hand, Tripathy et al. (2022) used a hybrid Delphi method to define perceptual indicators of EV battery recycling in terms of technology, organization and environment.

Original2.2. Existing recycling policies related to EV batteries

OriginalPolicy attributes refer to "market conditions and constraints" (Swait, 1994), and are often effective tools used by governments to incentivize pro-environmental behavior (Huang et al., 2022b).

OriginalRegarding recycling behavior, policy incentives can encourage active participation in waste recycling by reducing prices or increasing recycling convenience, thereby improving recycling rates (Mak et al., 2019; Tang et al., 2023).

OriginalAt a broader level, Taghipour et al. (2022) indicated that in a circular economy, governments can support recycling companies in four areas: facilities, barriers, enablers and EC.

OriginalThese supports include investing directly or indirectly in facilities, providing funds for new technologies or processes for companies, providing facilities or financial support for logistics and so on for companies, and setting targets for environmental development for companies.

OriginalFor example, Dou et al. (2023) mentioned that the widespread application of the scrappage subsidy policy can promote product recycling. Zhou et al. (2021) suggested that govern ment subsidies help businesses and individuals recycle agricultural waste.

OriginalSimilar to other recycling policies, EV battery recycling policies also focus on the impact of recycling subsidies. For example, Li et al. (2023) found that subsidy incentives are a notable factor driving young con sumers to choose formal recycling outlets for EV batteries.

OriginalTang et al. (2023) also suggested that economic incentives, such as cash rewards, bonuses, subsidies, and tax exemptions, could encourage widespread participation in EV battery recycling.

OriginalLiu and Wang (2021) analyzed the impact of subsidizing different players in the EV supply chain on EV battery recycling. In addition to recycling subsidies, EV battery recycling involves many other policies.

OriginalFor instance, Huang et al. (2022b) pro posed a recycling information traceability policy and a trade-in policy, in which the recycling information traceability policy is divided into four stages: destination untraceable, traceable to the recycler, traceable to the battery dismantling stage, and traceable to the battery cascade utilization stage.

OriginalTheir study demonstrated that such policies affect EV adoption, but it remains unclear whether they impact EV battery recycling.

Original2.3. Aims and gaps

OriginalOur study addresses gaps in the existing literature by investigating public preferences for formal and informal recycling of EV batteries. Despite the growing importance of EV battery recycling, previous studies have primarily neglected quantitative assessments of observable attributes, focusing instead on perceptual indicators.

OriginalAdditionally, the potential application of discrete choice models has been overlooked in X. Huang et al. Journal of Environmental Management 370 (2024) 122180 this field. To address this, our study proposes a discrete choice model, integrating nine observable attributes to analyze public recycling pref erences comprehensively.

OriginalTo the best of our knowledge, our study is the first to apply the discrete choice model to elucidate the public's pref erence for formal versus informal recycling options for EV batteries, contributing novel insights to the literature.

OriginalIn addition, existing studies predominantly focus on recycling subsidy policies, with limited analysis of other potential policy impacts. In response, our study extends the scope of analysis to include trade-in policy, carbon credit incentive policy, and recycling information traceability policy.

OriginalBy exploring these additional policy dimensions, we aim to provide new insights into government policy implications for EV battery recycling.

Original3. Methodology

OriginalIn this section, we will focus on the design of the choice experiment, the choice model specification, and the data collection process utilizing the method flowchart depicted in Fig. 1.

Original3.1. Discrete choice experiment assumptions

OriginalRandom utility theory underpins discrete choice experiments, assuming that decision-makers latent preferences, or latent utilities, are associated with all considered options and are present in all individuals (Louviere et al., 2010).

OriginalThese preferences consist of an observed component and a random component. The observed component en compasses all observable factors influencing the decision-maker's choices, such as covariates (e.g., age and income) and attributes (e.g., recycling cost and convenience).

OriginalThe random component includes all unobserved variables affecting the decision-maker choices but are not observable to the researcher, such as unobserved attributes, measure ment errors, or instrumental variables (Louviere et al., 2010; Lerusse and van de Walle, 2021).

OriginalAlthough discrete choice experiments are well suited to the analysis of individual behavior, they have certain inherent limitations of their own. One significant limitation is that participants evaluate a restricted number of attributes.

OriginalSome participants might base their assessments on an additional set of "unseen" factors, leading to potential endogeneity issues. To address this, we instructed participants to assume that any unmentioned characteristics were identical for both recycling outlets before making their evaluations.

OriginalAlthough this method does not completely eradicate the potential endogeneity problem caused by un mentioned attributes, it effectively reduces the risk of omitted variable bias (Lancsar and Louviere, 2006; Lerusse and van de Walle, 2021).

OriginalSecondly, participants may exhibit social desirability bias when evaluating choice tasks, selecting the socially preferred alternative rather than the option they would choose in real life (Lerusse and van de Walle, 2021).

OriginalThis social desirability bias can lead to inflated coefficient estimates. Given this inherent limitation of discrete choice experiments, we emphasize that our research focuses on stated preferences rather than actual behavior.

OriginalTherefore, we interpret our conclusions with caution.

Original3.2. Choice experiments design

Original3.2.1. Attributes and attributes levels

OriginalOur study considers nine attributes to reveal the public's preferences for formal and informal recycling of EV batteries. Table 1 illustrates all the variables and their encoding methods used in this study through linear, dummy, and label encoding.

OriginalLinear encoding is employed for continuous variables, dummy encoding for non-ordered categorical variables, and label encoding for ordered variables. Regarding the service attributes of EV battery recycling, we include recycling price, the market penetration rate of formal outlets, recycling entity, and recycling method.

OriginalRecycling pricing significantly impacts the return that EV owners receive when recycling end-of-life EV batteries and is regarded as the primary factor affecting recycling behavior.

OriginalIn the Chinese EV market, the recycling price for a retired EV battery generally ranges between 7500 and 20,000 RMB, depending on the recycling outlet and the battery's weight.

OriginalPrices at informal recycling outlets are usually higher than those at formal outlets (Huang et al., 2022b; Li et al., 2023; Su et al., 2024). Therefore, we set the recycling price levels of formal outlets as 7500 and 10,000 RMB, and prices at informal outlets are set as 15,000 and 20,000 RMB.

OriginalSome studies have found there exists a strong link between the number of recycling facilities and the distance EV owners are willing to Fig. 1.

OriginalMethod flow chart.

OriginalHuang et al. Journal of Environmental Management 370 (2024) 122180 travel to reach them; the more facilities there are, the more willing EV owners are to recycle (Hage et al., 2009; Geiger et al., 2019).

OriginalIn our study, the formal outlet market penetration rate, defined as the pro portion of formal recycling outlets to the total number of recycling outlets in the market, indicates the distribution of formal recycling outlets within the market.

OriginalAn increase in market penetration rate im plies an increase in formal recycling outlets, thereby reducing the travel distance to reach recycling facilities. Therefore, we believe that the formal outlet market penetration rate may influence EV owners' de cisions to choose formal outlets for recycling.

OriginalSafety has been explicitly identified as a major factor influencing the recycling of EV batteries. Safety responsibilities are primarily borne by the entity responsible for recycling, making the recycling entity a crucial factor affecting battery recycling (Thompson et al., 2020).

OriginalFollowing the classification by Zhang et al. (2021), we categorize the recycling entities into three types: battery manufacturers, automobile manufacturers, and third-party recycling companies.

OriginalIn the battery recycling process, con venience and privacy are factors that EV owners commonly consider, and different recycling entities have varying practicality in these aspects (Wang et al., 2011; Miliute-Plepiene et al., 2016; Cai et al., 2023; Chang et al., 2023).

OriginalInspired by Wang et al. (2018) on the emerging door-to-door recycling method, we combine this new method with the traditional methods of postal delivery and self-delivery to recycling designated outlets.

OriginalWe consider these three recycling methods as factors influencing retired EV battery recycling and incorporate them into our study. Regarding policy attributes, we examine carbon credits, end-of-life, and subsidy policies.

OriginalCarbon emission trading is a specialized approach aimed at maximizing the reduction of carbon emissions (Li et al., 2020). Carbon credits are a form of individual carbon trading scheme that requires governments to establish a market based on carbon credits and set trading rules, including setting initial credits and asso ciated penalties, to ensure that each offset their energy consumption in each compliance period (Starkey, 2012).

OriginalThe impact of individual car bon trading schemes on the adoption of EVs has been verified, with findings suggesting that, in some cases, this policy can be a substitute for subsidized car purchases (Li et al., 2019, 2020).

OriginalTherefore, we seek to understand whether carbon credits impact the recycling of retired EV batteries, hence including carbon credits in the scope of policy attribute examination.

OriginalOur study introduces carbon credit policies, assuming one carbon credit equals 5 RMB, and incorporates 100, 300 and 500 carbon credits into the discrete choice experiment to explore the impact of different carbon credit levels on retired EV battery recycling at autho rized recycling outlets.

OriginalThe trade-in policy is a method for recycling EV batteries. This approach has been widely adopted in the mobile phone and computer fields but is rarely implemented in EV manufacturers in China.

OriginalThe classified levels of the trade-in policy for each EV battery are 10%, 20% and 40% of the cost of a new battery (Huang et al., 2022b). Regarding the traceability of recycling information policy, Huang et al. (2022b) categorized the traceability of batteries into four stages: no recycling information traceability, traceable to the battery cascade utilization stage, traceable to the outlet stage and traceable to the battery dismantling stage, where this study also leverages the same categorizations.

OriginalSubsidy policies have effectively promoted mass recycling behavior (Mak et al., 2019; Tang et al., 2023). The Chinese government has enacted subsidy policies to stimulate the formal recycling industry.

OriginalStandard subsidy policies include subsidies based on the remaining ca pacity of the battery and one-time fixed subsidies (Chen et al., 2022). The current subsidy standard for battery remaining capacity in the Guangxi region is approximately 20 RMB/kWh.

OriginalTherefore, our study considers three subsidy levels: 10 RMB/kWh, 20 RMB/kWh, and 30 RMB/kWh (Huang et al., 2022b). Regarding one-time fixed subsidies, the standard in Shanghai is approximately 1000 RMB per vehicle, pro vided to recycling companies, and the actual subsidy amount given to EV owners may be correspondingly reduced.

OriginalHence, our study considers three one-time fixed subsidy values: 500 RMB, 600 RMB, and 700 RMB.

Original3.2.2. Choice task generation

OriginalUsing a complete factorial design would require generating 34,992 virtual selection tasks. However, handling many virtual tasks is impractical in practice, so an orthogonal factorial design is recom mended (Hidrue and Parsons, 2015).

OriginalOur study utilized this method to reduce the complexity of the choice task, resulting in the generation of 53 profiles. These profiles were divided into 20 blocks, each containing six choice tasks.

OriginalUpon consenting to participate in the survey, partici pants were randomly assigned to one of these blocks. Each choice task included two hypothetical recycling center profiles and a "None" option.

OriginalFig. 2 illustrates an example of a choice task.

Original3.3. Model specification

OriginalOur study conducted a discrete choice experimental survey in which respondents were required to choose the optimal option among multiple virtual retired EV battery outlets based on their preferences.

OriginalDuring this process, respondents were portrayed as rational individuals who maxi mize their utilities by carefully evaluating preferences for various recycling attributes.

OriginalOur survey aids us in better understanding public behaviors in recycling choices within a simulated scenario (Li et al., 2020; Huang et al., 2022b). Specifically, within the given choice set (C),

OriginalTable 1 Variable descriptions and assignment.

OriginalVariables Encoding type Variable assignment Recycling price (RMB Linear 0.75, 1, 1.5, 2 Market penetration rate of formal recycling outlets Linear 10%, 20%, 30% Carbon credits Linear 100, 300, 500 Retired EV battery trade-in policy (X% of new battery price) Linear 10%, 20%, 40% One-time fixed subsidy (RMB) Linear 500, 600, 700 Capacity subsidy (RMB/ kWh) Linear 10, 20, 30 Traceability of recycling information level Dummy Destination untraceable = (0,0,0); Traceable to recycling outlets = (1,0,0); Traceable to battery dismantling = (0,1,0); Traceable to battery cascade utilization = (0,0,1) Recycling entity Dummy EV battery producers = (1,0,0); Automotive manufacturers = (0,1,0); Third-party recycling companies = (0,0,1) Recycling method Dummy Individuals personally delivering the batteries to recycling outlets = (1,0,0); Individuals mailing the batteries to recycling outlets = (0,1,0); Recycling outlets offering doorstep battery collection = (0,0,1) Gender Dummy Male = 1; Female = 0 Age Label [18, 25) = 1; [25,35) = 2; [35,45) = 3; [45,55) = 4; [55,65) = 5; [65, +∞) = Monthly income Label (0, 3000] = 1; (3000, 6000] = 2;

OriginalCommuting cost Label (0, 300] = 1; (300, 600] = 2; (600, Education Label Senior middle school or lower = 1; Bachelor degree or Associate degree = 2; Master's degree = 3; Doctoral degree = 4 Household car ownership Label 0 = 0; 1 = 1; 2 = 2; 3 or more = 3 Driver's license Label Yes = 1; No = 0 X.

OriginalHuang et al. Journal of Environmental Management 370 (2024) 122180 the selected option has the highest utility. For respondent (n), the utility of option (Uin) encompasses an observable or systematic explanatory variable component (βXin) and an unobservable random component or error term (εin), as expressed in Equation (1).

OriginalUin = βXin + εin (1) Equation (2) represents the probability, denoted as Pi, of choosing alternative i over alternative j within the choice set C. Pi = Pr ( Vi + εi≥ Vj + εj, i ∈ C, i ∕ = j ) = Pr ( Vi − Vj ≥ εj − εi ) (2) Concerning willingness-to-pay (WTP), when the utility Uin is repre sented by equation (2), the partial derivatives of utility for the k-th attribute Xkn and the cost attribute Xcn can be formulated as dUin = βkdXkn + βcdXcn.

OriginalSetting this expression to 0 and solving for y results in equation (3). dXcn dXkn = WTPk = − βk βc (3)

Original3.4. Data collection

OriginalOur study takes Chongqing, China, as the study area due to its pio neering role in EV promotion and major status in EV production base (Huang et al., 2021, 2022a, 2022b).

OriginalIn 2022, Chongqing experienced a remarkable surge in EV production, manufacturing 365,200 units - a 140% increase from the previous year. This surge also resulted in an installed capacity of approximately 8.6 GWh for EV batteries.

OriginalPro jections suggest that by 2030, the retired volume of EV batteries in Chongqing will exceed 50,000 tons, with a capacity of 5 GWh, making the city as a key for studying EV battery recycling.

OriginalThe stated preference questionnaire consisted of three parts: an

Originalintroduction to the background of EV battery recycling and some atti

Originaltudinal indicator questions, a discrete choice experimental part, and demographic characteristics. To ensure questionnaire validity, we also set six filter rules: (1) only residents living in Chongqing could complete the study.

Original(2) Participants could participate only once using the same device or IP address. (3) "The number of cars owned" was placed at the beginning and end to detect logical errors, while inconsistent responses were flagged as invalid.

Original(4) Participants unfamiliar with retired EV battery recycling or the establishment rules for formal recycling outlets after reading the introduction were considered invalid responses.

Original(5) Participants without a driver's license were deemed invalid. (6) Those who select "None" for all tasks are invalid responses. As for the data collection itself, we executed two steps: (1) set six filtering rules to identify valid questionnaires; (2) mobilized all members of the team to collect data by distributing questionnaires using an online platform.

OriginalNote that in order to ensure the randomness of the sample and try to remove the influence of convenience samples, we prioritized distrib uting the questionnaires on platforms without social networks, such as Weibo, TikTok, Xiaohongshu, and forums, followed by WeChat.

OriginalAs an incentive, those who provided valid responses would receive a reward in the form of a WeChat red envelope (5 yuan, about $0.69). Following the outlined steps, we surveyed Chongqing from May 8, 2023, to May 15, 2023, utilizing the Sojump platform. Finally, we collected 1190 valid questionnaires.

OriginalTheir spatial and geographical distribution is shown in Fig. 3.

Original4. Result

OriginalIn this section, we present a sample descriptive analysis, model estimation and WTP, with model estimation results including analysis of formal versus informal recycling behaviors, multi-group analysis, and comparisons of the two behaviors.

Original4.1. Sample description and analysis

OriginalThe description and analysis of the data involve three steps: 1) sta tistical description of the overall data; 2) employing Exploratory Factor Analysis (EFA) to identify latent variables; 3) using the k-means clus tering technique to group responses based on the concavity of latent variables.

Original4.1.1. Sample description

OriginalThe demographic characteristics of the respondents are presented in Table 2. In our sample, males slightly outnumber females, constituting 55%, while females account for 45%. The age distribution is predomi nantly between 18 and 64 years old, which is similar to the gender distribution in the actual census for Chongqing (49.27% males, 50.73% females).

OriginalSpecifically, 41% of the respondents were aged between 18 and 34, 48.7% were between 35 and 44, and the remaining 10.3% were between 45 and 64. In terms of education, 75.5% have a bachelor's degree, 12.6% have a graduate degree, and the remaining respondents have a high school education or lower.

OriginalRegarding monthly income, 60.6% of respondents earn below 10,000 RMB, 35.6% earn between 10,000 and 20,000 RMB, and the remainder earn more than 20,000

OriginalRMB.

OriginalAmong respondents with cars at home, 78.7% of households have one car, and only 1.1% of respondents' households have three or more cars. Additionally, 68.9% of respondents report spending between 300 and 900 RMB on commuting each month.

OriginalApproximately 91.4% of re spondents are willing to choose formal retired EV battery recycling outlets, whereas only 55.3% are prepared to choose informal retired EV battery recycling outlets.

OriginalThis trend aligns with the development di rection of formal recycling outlets for discarded EV batteries. Overall, the demographic characteristics of our sample correspond with the usage of EVs in the Chongqing region.

Original4.1.2. Cluster analysis

OriginalK-means clustering is an iterative algorithm that divides a dataset into several groups based on specific features of its samples. This method aims to identify subgroups with high similarity within each subgroup while ensuring significant differences between groups (Hastie et al., 2009; Hifinger et al., 2017). In model selection, clustering was per formed first, followed by regression analysis to reveal differences in influencing factors between groups (Hifinger et al., 2017; Shoabjareh et al., 2021).

OriginalIn K-means clustering, the researcher typically undertakes Fig. 2. Example task used in the survey (translated from the Chinese questionnaire). Huang et al. Journal of Environmental Management 370 (2024) 122180 the task of determining the number of clusters.

OriginalAccording to the vali dation results from the literature (Lezhnina and Kismihók, 2022), the Bayesian Information Criterion (BIC) elbow value and the maximum Average Silhouette Width (ASW) value were found to be more accurate than determining the number of clusters based on the minimum BIC values alone.

OriginalOur study used EC and retired EV battery recycling knowledge (REVBRK) as the primary latent factors influencing EV owners' recycling behavior (Geiger et al., 2019; Chang et al., 2023; Li et al., 2023).

OriginalA seven-point Likert scale was employed to measure re spondents' EC and REVBRK (for detailed information on measurement issues, refer to Table A.1 in Appendix A1). We conducted the EFA, and the results are presented in Table 3.

OriginalThese dimensions collectively ac count for 59.477% of the total variance. We utilized the Keiser-Meyer-Olkin (KMO) test to assess the adequacy of principal component analysis.

OriginalThe KMO value (0.879) indicates excellent sample adequacy (Kaiser, 1974). Furthermore, Bartlett's sphericity test yielded a statistic of 20731.687 (degrees of freedom = 36, p = 0.00 < 0.005), confirming the data's factorability.

OriginalThe formulas for the BIC and ASW are given by Equations (4) and (5), respectively: BIC = − 2logL + plogn (4) where p represents the number of free parameters in the model, n de notes the number of observations, and L represents the maximum like lihood function of the model.

OriginalWhen the number of observations n is large, minimizing the BIC is equivalent to maximizing the posterior probability of the model. S(i) = b(i) − a(i) max{a(i), b(i)} (5) where, a(i) represents the average dissimilarity between observation ii Fig. 3.

OriginalMap of the spatial and geographical distribution of the sample.

OriginalTable 2 Demographic characteristics.

OriginalVariables Categories Total Sample Count Percentage Gender Male 654 55% Female 536 45% Age 18–25 47 3.9% 55 or older 7 0.6% Education Senior middle school or lower Bachelor's degree or Associate degree Master's degree 137 11.5% Doctoral degree 13 1.1% Monthly income Under 3000 RMB 49 4.1% 3001-6000 RMB 187 19.8% 6001-10000 RMB 485 40.8% 10001-20000 RMB 424 35.6% Over 20000 RMB 45 3.8% Total monthly weekday commuting costs Under 300 RMB 236 19.8% 300-600 RMB 504 42.4% 600-900 RMB 315 26.5% Over 900 RMB 135 11.3% Total number of family cars 0 49 0.7% 3 or more 13 1.1% Willingness to choose formal EV recycling outlets Very unlikely 4 0.3% Unlikely 99 8.3% Likely 334 28.1% Very likely 753 63.3% Willingness to choose informal EV recycling outlets Very unlikely 363 30.5% Unlikely 168 14.1% Likely 229 19.2% Very likely 430 36.1% Driver's license Owned 1190 100% Table 3 Indicator and rotated factor loadings.

OriginalIndicators REVBRK EC REVBRK01 0.748 REVBRK02 0.795 REVBRK03 0.751 REVBRK04 0.795 REVBRK05 0.783 EC01 0.719 EC02 0.743 EC03 0.776 EC04 0.819 X. Huang et al. Journal of Environmental Management 370 (2024) 122180 and the other points within its cluster, while b(i) denotes the average dissimilarity between observation ii and all observations in the nearest but different cluster.

OriginalThe ASW value ranges from − 1 to 1, where higher positive values indicate well-defined clusters characterized by intracluster cohesion and inter-cluster separation. Conversely, ASW values close to 0 or negative suggest insufficient clarity in the separation be tween clusters.

OriginalTable 4 shows the settlement results of BIC and AWS.

OriginalFig. 4 illus trates the relationship between BIC, ASW, and the number of clusters, with the vertical line indicating the optimal number of clusters (Lezhnina and Kismihók, 2022).

OriginalIt can be observed that the optimal number of clusters is 2. Finally, two groups were obtained using the k-means method and the optimal number of clusters, as shown in Table 5.

OriginalThe first group comprises respondents with high EC and a high

OriginalREVBRK.

OriginalThe second group of respondents has a lower EC and a lower REVBRK.

Original4.2. Model estimation

OriginalThis section employs the multinomial logit model (MNL) for the sample data. The model estimation process utilized Biogeme for model construction and result estimation (Bierlaire and Ortelli, 2023).

OriginalTables 6 and 7 present the estimated coefficients of the variables for selecting formal versus informal recycling outlets, for both the two groups and the total sample, respectively.

Original4.2.1. Formal recycling estimation result

OriginalRegarding formal outlet service attributes, the MNL results (see Table 6) indicate that the coefficients of recycling price, formal outlet market penetration rate, recycling entity, and recycling method exhibit positive effects at a significance level of 1%.

OriginalSpecifically, the first two attributes (recycling price and formal outlet market penetration rate) positively influence respondents' choices to recycle retired EV batteries at formal outlets.

OriginalRegarding preferences for recycling entities, with third-party recycling companies as the reference, respondents expressed a greater inclination to choose EV battery manufacturers over automo bile manufacturers or third-party recycling companies.

OriginalConcerning preferences for recycling methods, our study finds that, compared to delivering items to designated outlets or mailing them to specified lo cations, potential EV owners are more inclined to choose recycling outlets that offer doorstep collection services.

OriginalOur study indicates that two of the three policy incentives (trade-in and two types of subsidy policies) significantly and effectively promote potential EV owners' choices to recycle retired EV batteries at formal outlets at a 1% significance level.

OriginalHowever, the carbon credit policy does not significantly impact the adoption of formal outlets for recycling retired EV batteries. Finally, regarding socio-demographic characteris tics, only age and gender exhibit significant associations at a 1% level, indicating that as age increases, potential EV owners are more willing to recycle retired EV batteries at formal outlets.

OriginalHowever, individuals with higher education levels are less inclined to engage in EV battery recycling.

Original4.2.2. Informal recycling estimation result

OriginalRegarding informal outlet service attributes, the MNL estimation

Originalresults (see Table 7) indicate that the coefficients of recycling price,

Originalrecycling entity, and recycling method mostly exhibit positive effects at a 1% significance level. Specifically, recycling price positively impacts consumers' choices to recycle retired EV batteries at informal outlets.

OriginalConcerning consumers' preferences for recycling entities, with thirdparty recycling companies as the reference, respondents expressed a greater inclination to choose third-party recycling companies over automobile manufacturers or EV battery manufacturers.

OriginalAs for con sumers' preferences for recycling methods, our study found that, compared to providing doorstep collection services or mailing items to specified locations, potential EV owners are more willing to choose selfdelivery to designated outlets as a recycling method.

OriginalRegarding sociodemographic characteristics, our findings indicate that older in dividuals hold a positive attitude toward informal outlet recycling of retired EV batteries, while individuals with higher education levels ex press a negative attitude.

Original4.2.3. Multi-group analysis results

OriginalThrough cluster analysis, we categorized the entire sample into two groups based on two latent variables, EC and REVBRK. The first group comprises individuals with high EC and REVBRK, while the second group comprises those with relatively lower EC and REVBRK.

OriginalThese groups exhibit differences in the utility coefficient for selecting EV battery recycling outlets, reflecting distinct attitudes toward recycling outlet preferences.

OriginalConcerning formal recycling outlet attributes (see Table 6), both groups share the same preferences for recycling price and type of sub sidy policy, showing acceptance of higher recycling prices and a pref erence for capacity subsidies over one-time fixed subsidies.

OriginalOther than that, Group 1 has more choice preferences, and Group 2 has very few. Specifically, market penetration and trade-in policies can incentivize Group 1 to opt for formal recycling, while Group 2 remains indifferent.

OriginalRegarding the recycling entity, both groups are more likely to choose EV battery manufacturer recycling for formal recycling than third-party recycling companies. However, Group 1 is less likely to choose formal recycling if the car manufacturer recycles, and Group 2 doesn't care.

OriginalIn terms of the recycling method, if it was mailed to the designated outlet, Group 1 would be less inclined to choose formal recycling, and Group 2 would not care. As for the information traceability policy, if it is trace able to the battery cascade utilization stage, Group 1 will be less inclined to opt for formal recycling, and Group 2 will not care.

OriginalRegarding de mographic characteristics, age is a significant factor influencing the choice of formal recycling for Group 1 but not for Group 2. As for informal recycling outlet attributes (see Table 7), both groups still share the same preferences for recycling price and type of subsidy policy.

OriginalRegarding recycling entities, Group 2 is less likely to choose informal recycling if it is EV battery recycling, and Group 1 is not Table 4 The cluster results of the individuals.

OriginalCluster BIC ASW Fig. 4. Graphical output of BIC and ASW results. Huang et al. Journal of Environmental Management 370 (2024) 122180 concerned. Regarding the recycling method, Group 1 is less inclined to select informal recycling if it involves mailing to the designated outlet, while Group 2 is not significant.

OriginalRegarding demographic characteristics, age and education influence the informal recycling choices of Group 1 but not Group 2.

Original4.2.4. Comparison of formal and informal recycling behavior

OriginalThere are significant differences between potential EV users' preferences for formal versus informal recycling of vehicle batteries, and in this section, we focus more on the differences and how to facilitate the conversion of the user market from informal to formal recycling.

OriginalRegarding recycling service attributes, the price of recycling is a sig nificant factor. The recycling price coefficient of formal recycling (1.99, see Table 6) is greater than that for informal recycling (1.32, see Table 7), and an increase in the recycling price is more conducive to facilitating the conversion of informal recycling to formal recycling.

OriginalTable 5 The cluster results of the individuals.

OriginalCluster Label Group1 (High EC and high REVBRK) Group2 (Low EC and low REVBRK) Size Item Value Item Value Average index of latent variable attributes REVBRK 01 6.59 REVBRK 01 5.54 REVBRK 02 6.42 REVBRK 02 4.85 REVBRK 03 6.51 REVBRK 03 5.41 REVBRK 04 6.46 REVBRK 04 5.06 REVBRK 05 6.47 REVBRK 05 5.26 EC01 6.65 EC01 5.99 EC02 6.56 EC02 5.84 EC03 6.59 EC03 5.56 EC04 6.65 EC04 5.84 Table 6 The model estimation results of selecting formal outlets for groups.

OriginalVariables MNL Cluster analysis-MNL Estimate Std. Err (t-test) G1 G2 Estimate Std. Err (t-test) Estimate Std. Err (t-test) Recycling outlet attribute Rrecycling price 1.99a Market penetration 0.477a Recycling entity EV battery producers 0.228a Automotive manufacturers − 0.148b Third-party recycling companies Based Based Based Based Based Based Recycling method Recycling outlets offering doorstep battery collection 0.325a Mail to the designated outlets − 0.171a Self-deliver to the designated outlets Based Based Based Based Based Based Policy attribute Trade-in policy 0.628a Carbon credit 0.000146 0.000154 (0.948) 0.000171 0.000183 (0.938) 9.38e-05 0.00029 (0.324) Subsidy based on remaining battery capacity 0.0195a

OriginalOne-time fixed subsidy 0.000473a Information traceability policy Traceable to the battery dismantling stage − 0.105 0.0743 (− 1.41) − 0.101 0.0882 (− 1.14) − 0.107 0.14 (− 0.764) Traceable to the outlet stage 0.329a Traceable to the battery cascade utilization stage − 0.175b No recycling information traceability Based Based Based Based Based Based Demographic attributes Gender − 0.109 0.183 (− 0.598) − 0.173 0.236 (− 0.733) 0.0487 0.3 (0.162) Age 0.27b Education − 0.37b Commuting cost 0.00281 0.101 (− 0.0279) − 0.0279 0.126 (− 0.221) 0.0241 0.18 (0.134) Monthly income − 0.0643 0.109 (− 0.589) − 0.0202 0.141 (− 0.144) − 0.0992 0.179 (− 0.544) Number of cars 0.117 0.2 (0.583) 0.257 0.281 (0.915) 0.0473 0.306 (0.151) Constants Formal recycling outlet 0.851 0.665 (1.28) 0.19 0.898 (0.212) 1.05 1.04 (1) None Based Based Based Based Based Based Number of observations 6873 4848 2025 Log-likelihood − 4960.287 − 3454.47 − 1466.488 Rho-square 0.0757 0.205 0.000131 Adjusted rho-square 0.343 0.351 0.341

OriginalStandard errors are in parentheses. a p < 0.01. b p < 0.05. c p < 0.1. Huang et al. Journal of Environmental Management 370 (2024) 122180 Regarding the recycling entity, users who choose informal recycling prefer third-party companies to recycle EV batteries, while those who choose formal recycling prefer EV battery companies to recycle.

OriginalTherefore, recycling of EV batteries by EV battery companies is signifi cantly better than recycling by EV manufacturers and recycling by thirdparty companies and is more conducive to facilitating the trans formation of the market from informal to formal recycling.

OriginalIn terms of recycling methods, users who chose informal recycling preferred to selfdeliver their batteries to designated recycling outlets, while those who chose formal recycling preferred door-to-door collection.

OriginalTherefore, door-to-door recycling is essential in stimulating the formal recycling market for EV batteries.

Original4.3. Willingness to pay

OriginalTable 8 presents the WTP for recycling outlet and policy attributes, calculated based on the coefficients obtained from Tables 6 and 7. These WTP values are determined by the ratio of the respective attribute co efficients to the purchase price coefficient.

OriginalFirst, it is observed that in dividuals exhibit a higher WTP for policy attributes, while the WTP for recycling outlet attributes is relatively lower, as detailed in Table 8.

OriginalWhen selecting formal recycling outlets, individuals demonstrate the highest WTP for the trade-in policy. An increase of 10% in the trade-in policy incentive leads individuals to accept a reduction of 3156 RMB in recycling returns.

OriginalSecond, regarding recycling methods, Individuals are willing to accept a 1633 RMB reduction in recycling returns for the convenience of doorstep collection by recycling merchants compared to self-delivery to designated outlets.

OriginalThird, the market penetration rate of formal outlets is one of the crucial factors individuals consider. Ac cording to Table 7 with every 10% increase in the market penetration rate of formal outlets, individuals are willing to accept a decrease of

Original2397 RMB in recycling returns.

OriginalWhen choosing informal recycling outlets, individuals exhibit a significantly higher WTP for the trade-in policy than other attributes. An increase of 10% in the trade-in policy incentive leads individuals to accept a reduction of 8939 RMB in recycling returns.

OriginalSecondly, con cerning the recycling entity, choosing automotive manufacturers over third-party recycling companies increases the recycling returns by 3038 RMB, indicating that individuals are more willing to pay for the service provided by third-party recycling companies.

OriginalFinally, regarding recy cling methods, selecting recycling outlets offering doorstep battery collection over self-delivering to designated outlets increases the acceptable recycling returns by 3416 RMB.

OriginalTable 7 The model estimation results of selecting informal outlets for groups.

OriginalVariables MNL Cluster analysis-MNL Estimate Std. Err (t-test) G1 G2 Estimate Std. Err (t-test) Estimate Std. Err (t-test) Recycling outlet attribute Recycling price 1.32a Recycling entity EV battery producers − 0.111c Automotive manufacturers − 0.401a Third-party recycling companies Based Based Based Based Based Based Recycling method Recycling outlets offering doorstep battery collection − 0.451a Mail to the designated outlets − 0.178a Self-deliver to the designated outlets Based Based Based Based Based Based Policy attribute Trade-in policy 1.18a Demographic Factors attribute Gender 0.164 0.182 (0.9) 0.222 0.236 (0.943) 0.0286 0.289 (0.096) Age 0.229c

OriginalEducation − 0.385b Commuting cost 0.0278 0.101 (0.277) − 0.0273 0.126 (− 0.217) 0.0943 0.178 (0.529) Monthly income 0.0169 0.109 (0.155) 0.0803 0.14 (0.572) − 0.0567 0.178 (− 0.319) Number of cars − 0.00736 0.2 (− 0.0368) 0.131 0.281 (0.462) − 0.0897 0.304 (− 0.295) Constants Informal recycling outlet 1.22c None Based Based Based Based Based Based Number of observations 6873 4848 2025 Log-likelihood − 4960.287 − 3454.47 − 1466.488 Rho-square 0.0757 0.205 0.000131 Adjusted rho-square 0.343 0.351 0.341 Standard errors are in parentheses.

OriginalTable 8 Marginal WTP for changes in vehicle and policy attributes.

OriginalRecycling outlets type FMRO (RMB) IFMRO (RMB) Recycling outlet attribute Market penetration − 2397 – Recycling entity EV battery producers − 1146 841 Automotive manufacturers 744 3038 Third-party recycling companies Based Based Recycling method Recycling outlets offering doorstep battery collection − 1633 3416 Mail to the designated outlets 859 1348 Self-deliver to the designated outlets Based Based Policy attribute Trade-in policy − 3156 − 8939 Subsidy policy Capacity subsidy − 98 – One-time fixed subsidy − 2 – Information traceability policy Traceable to the outlet stage − 1653 – Traceable to the battery cascade utilization stage 884 – No recycling information traceability Based – X.

OriginalHuang et al. Journal of Environmental Management 370 (2024) 122180

Original5. Discussion

Original5.1. Theoretical contributions

OriginalOur study contributes to the existing knowledge base in three main aspects. First, our study proposes a discrete choice model to identify the choice preferences of formal and informal recycling for EV battery recycling.

OriginalTo the best of our knowledge, our study is the first to be used to reveal observable attribute preferences for EV battery recycling. Second, we introduce new attributes from the recycling service and policy perspectives, such as recycling entities, recycling methods, tradein policies, carbon credit policies, subsidy policies, and information traceability policies, to quantify the market's choice preferences and policy influence.

OriginalThese attributes have not been analyzed in existing studies, yet they are essential for an in-depth understanding of the EV battery recycling market. Finally, we reveal consumer heterogeneity in the EV battery recycling market by using multi-group analysis and comparing the differences.

OriginalFinally, our results show that most of the proposed attributes are significant for formal recycling options for EV batteries, except for carbon credit policies, which are not significant.

OriginalRegarding demographic variables, we found that only age and education affect EV battery recycling.

Original5.2. Managerial implications

OriginalOur study also provides valuable management insights for govern ment policy-making and demand guidance for the EV battery recycling industry. The market penetration of formal recycling outlets is essential for the formal recycling of EV batteries, especially for potential EV users with high EC and REVBRK.

OriginalA higher market penetration implies that formal recycling outlets are more widely available, allowing consumers to choose them for battery recycling conveniently. The shortage of formal recycling outlets has consistently been a significant obstacle to the recycling of retired EV batteries in China, with statistics indicating that the market share of formal recycling outlets is currently less than 25% (Li et al., 2023). Therefore, the government and enterprises should focus on promoting more formal recycling outlets and, simultaneously, publicizing Group 1 users to increase the amount of recycling from formal recycling outlets.

OriginalRegarding the recycling entity, the formal EV battery recycling market favors recycling by EV battery manufacturers over third-party recyclers, while recycling by car manufacturers is the least effective

Originalmethod. This finding may be because EV battery manufacturers are

Originalmore specialized in battery handling and are more likely to be trusted by the market, whereas many car manufacturers are not responsible for producing batteries and therefore do not enjoy the same level of trust.

OriginalTherefore, supporting EV battery companies to recycle should be the priority in promoting the formal recycling industry. On the other hand, for the informal recycling market, it is their least favorite that car manufacturers are responsible for recycling decommissioned batteries.

OriginalThis phenomenon may be because recycling networks dominated by car manufacturers are more developed and offer greater convenience, reducing the likelihood of informal recycling.

OriginalTherefore, supporting vehicle manufacturers in recycling EV batteries is more advantageous in curbing the informal recycling industry. Regarding the recycling

Originalmethod, door-to-door recycling of EV batteries is the favorite of formal

Originalrecycling users and, simultaneously, the least preferred way by informal recycling users. Door-to-door recycling offers the greatest convenience and facilitates the transition of the EV battery market from informal to formal recycling.

OriginalTherefore, companies should prioritize the promotion of door-to-door recycling services. As for the recycling policies, the trade-in and subsidy policies are conducive to promoting the formal recycling market for EV batteries.

OriginalIn particular, the EV battery recycling market has a much higher WTP for the trade-in policy than for the subsidy policy (see Table 8). This phe nomenon suggests that the essence of battery recycling for potential EV users is to renew and reduce battery replacement costs, not to obtain a subsidy.

OriginalTherefore, actively promoting trade-in policies may have a vast stimulating effect on the recycling market. Certainly, subsidy policies are still effective for the pro-environmental behavior of EV battery recycling, and this conclusion is also consistent with the existing liter ature (Chen et al., 2022; Tang et al., 2023).

OriginalWe also find that the recy cling market is more interested in capacity subsidy policies than one-off subsidy policies (see Table 8). Therefore, the Government should consider a capacity subsidy policy, which can give the market the most significant stimulus, rather than a one-off subsidy policy across the board.

OriginalWhile capable of promoting the formal recycling market, the recy cling information traceability policy's influence is relatively weak, and the carbon credit policy does not show a significant effect.

OriginalSpecifically, battery traceability in the recycling market is limited to ensuring that batteries are delivered to recycling outlets, with minor concern for the subsequent flow of batteries for laddering.

OriginalThis makes sense as ensuring that batteries go to the correct recycling outlets is enough to meet the needs of most users, especially those with a high level of EC and REVBRK.

OriginalTherefore, it is necessary to build a traceability system for retired batteries, but it is enough to only record the flow of batteries to recycling outlets. Regarding demographic characteristics, age positively impacts the formal recycling market for EV batteries, while education level exhibits a negative influence.

OriginalThis finding may be because older users care more about the battery recycling itself than the final desti nation of the battery, stemming from a sense of self-risk aversion, whereas those with a higher level of education tend not to recycle, which may be related to the imperfections of the current recycling system.

OriginalBoth behavioral tendencies are influenced by biases, so the government and industry should actively promote and educate individuals in the recy cling market to formally increase their motivation to recycle EV batteries.

Original6. Conclusion and policy implications

OriginalThe surging EV market is giving the EV battery recycling market a huge challenge. Existing studies on the understanding of EV battery recycling behavior have been limited to latent variable analysis, neglecting the analysis of observable attributes and the assessment of related policy impacts.

OriginalTo address this gap, our study proposes a new discrete choice model including nine observable attributes and validates the feasibility of the proposed framework using a sample of 1190 con sumers from Chongqing, China, contributing knowledge to the existing literature.

OriginalAdditionally, we classified consumers into two groups based on potential variables (EC and REVBRK) to identify priority areas for recycling policies and outlet services and quantify their impact on formal outlet adoption.

OriginalTo our knowledge, this study is the first to employ a discrete choice model to investigate preferences for formal versus informal recycling options for EV batteries. Our study obtains several important findings.

OriginalRecycling prices significantly affect the EV battery recycling market, both for formal and informal recycling users, and the recycling price impact coefficient for formal recycling (1.99, see Table 6) is greater than that for informal recycling (1.32, see Table 7).

OriginalFor the formal recycling market, the tradein policy and market penetration have the highest WTPs (− 3156 for trade-in policy and − 2397 for market penetration, see Table 8), which is the favorite of formal recycling users.

OriginalThe recycling entity is also important in determining the EV battery recycling business model. Specifically, formal recycling users prefer EV battery producers to recycle, while the informal recycling market is least likely to be subject to the responsibility of vehicle producers for recycling.

OriginalIn terms of recycling methods, informal recycling users prefer to self-deliver to designated recycling outlets. Regarding the recycling subsidies, the ca pacity subsidy policy is preferred to the one-time fixed subsidy policy (− 98 for the capacity subsidy and − 2 for the one-off fixed subsidy; see Table 8).

OriginalAs for the information traceability policy, traceability to the X. Huang et al. Journal of Environmental Management 370 (2024) 122180 outlet stage is necessary and sufficient compared to non-retroactivity, whereas traceability to the battery cascade utilization stage is unnec essary.

OriginalFinally, older people are more inclined to support recycling, while those with more education tend not to participate. Our study also provides several important implications.

OriginalRecycling subsidies are essential for the government policy side, where capacity subsidies should be prioritized over one-time fixed subsidies. The gov ernment should also pay active attention to the density of formal recy cling outlets for EV batteries and construct a recycling traceability system, especially the traceability stage for recycling outlets.

OriginalThey should intervene and incentivize EV formal recycling market develop ment effectively. Increasing the recycling price is an important step for companies to effectively increase the amounts of batteries recycled by formal companies, especially to capture the recycling demand from some informal recycling options.

OriginalEV battery recyclers should be actively involved in EV battery recycling, which can capture more formal recy cling demand, and automakers should be actively engaged in EV battery recycling, which can capture more informal recycling demand.

OriginalIn terms of recycling methods, promoting door-to-door services is important for businesses to increase the demand for formal recycling and block the informal recycling market.

OriginalSeveral limitations are also involved in our study. First, our study only considered a sample of four policies and may have overlooked other relevant legislation. Future studies could attempt to examine the policy framework more comprehensively.

OriginalSecondly, despite our efforts to mitigate endogeneity issues in the experiment, it is impossible to completely eliminate the potential presence of measurement errors and residual endogeneity.

OriginalFuture research could try to suggest better ways to address endogeneity. Thirdly, our study focuses primarily on observable variables, and it is unclear how latent and observable variables jointly affect the EV battery recycling market.

OriginalFuture research could construct hybrid discrete choice models to analyze this further. Finally, how these policies and service attributes affect the evolution of the recycling market is also an important research question.

OriginalFuture research could try to construct an agent-based simulation model for further analysis. CRediT authorship contribution statement Xingjun Huang: Writing – original draft, Visualization, Investiga tion, Conceptualization.

OriginalSong Lei: Methodology, Formal analysis, Conceptualization. Feng Liu: Visualization, Investigation. Yan Li: Software, Resources, Funding acquisition. Fuli Zhou: Writing – review & editing, Supervision, Funding acquisition.

OriginalMing K. Lim: Writing – review & editing, Investigation, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

OriginalData availability Data will be made available on request.

OriginalAcknowledgements This study is supported by the Natural Science Foundation of Chongqing (grant no. CSTB2023NSCQBHX0084), the China Post doctoral Science Foundation (grant no.

Original2023M732389), and the Hu manities & Social Sciences Project from the Ministry of Education in PRC (grant no.

Original22YJC630220).

OriginalAppendix Table A.1 Measures of Constructs.

OriginalConstruct Adapted Items References Environmental concern (EC) (EC1) I disagree that protecting the environment is everyone's responsibility. Acemoglu et al. (2022) (EC2) I don't think we should use clean energy more daily.

Original(EC3) I won't consider saving electricity or reducing carbon dioxide emissions. (EC4) I am not willing to contribute to improving environmental issues. Retired EV battery recycling knowledge (REVBRK) (REVBRK1) I am aware that there are various recycling entities for retired EV batteries.

OriginalGeiger et al. (2019) (REVBRK2) I am familiar with the retired EV battery recycling process, including cascaded utilization and disassembly. (REVBRK3) I know that China is entering a "retirement wave" of power batteries, indicating a substantial scale in the recycling market.

Original(REVBRK4) I understand the scale and prospects of the retired EV battery recycling market. (REVBRK5) I know that China is improving its battery recycling system, including establishing a diversified recycling network and a traceability platform for recycling and utilizing power batteries.