Assessing Consumers’ Willingness to Pay for Secondary Utilization of Retired Battery Products: The Role of Incentive Policy, Knowledge, and Perceived Risks

Abstract

The rapid development of the new energy vehicle industry has resulted in a large number of retired power batteries. Creating products from second-use retired batteries (SURB) is crucial for sustainability by extending the batteries’ operational life, which, in turn, conserves resources and protects the environment. Consequently, this paper establishes a structural equation model (SEM) based on an interpretive structural model (ISM). It investigates consumers’ willingness to pay (WTP) for secondary utilization of retired batteries (SURB) products by extending the theory of planned behavior (TPB)with incentive policy, knowledge, and perceived risk. The study reveals that incentive policies and knowledge are fundamental factors, while subjective norms, perceived risk, and perceived behavioral control exert moderate influence. Attitude emerges as the most significant predictor, directly affecting consumers’ WTP, with perceived behavioral control also playing a key role. Incentive policies and knowledge have an indirect influence through perceived behavioral control and perceived risk. Finally, this paper discusses the theoretical and practical significance of the findings and provides relevant policy recommendations.

1. Introduction

Electric vehicles (EVs) have gained more attention as a sustainable and eco-friendly mode of transportation. The Chinese government has introduced several policies to promote the development of the new energy industry [1]. However, the power batteries used in new energy vehicles have a relatively short lifespan and become unsuitable for vehicle propulsion once their capacity falls below 80% [2]. According to the New Energy Battery Recycling Committee forecast, by 2027, the cumulative retired power batteries on new energy vehicles will reach 1.14 million tons. If these retired power batteries are not properly disposed of, they may cause potential waste of resources and environmental pollution [3]. According to Zhang et al. [4], the secondary use of retired SURBs in new energy vehicles has high market potential in China. Experts predicted that SURBs will gradually enter the field of low-speed vehicles, electric bicycles, and other mass consumer goods. Therefore, consumer acceptance of SURB-based products is crucial for the sustainable development of the secondary-use industry chain in this sector.

Numerous countries worldwide have actively promoted research and practical applications in the field of SURBs. Both domestic and international research primarily concentrated on key technologies for battery recycling, national policy development, and applied implementation. For example, Rautela et al. [5] reviewed current recovery technologies for lithium-ion batteries, emphasizing the industrialization potential of the recycling process. Wang et al. [6] proposed a treatment framework for retired batteries and systematically outlined methods for performance evaluation, classification, and reorganization. Ahmadi et al. [7] analyzed the life cycle management of electric vehicle lithium-ion battery packs and underscored the importance of environmental protection and resource efficiency in promoting SURB. They also emphasized the need for battery health assessment to ensure safety and effectiveness in second-life applications.

Research on consumer behavior regarding pro-environmental practices has also become increasingly relevant. Prior studies explored individuals’ willingness to voluntarily offset carbon emissions, purchase circular products, engage in municipal solid waste sorting, and adopt electric vehicles [8,9,10,11]. When analyzing consumers’ WTP for green products, researchers frequently adopted the Theory of Reasoned Action (TRA) or the TPB, extending the models by integrating contextual or situational variables. For instance, Nketiah et al. [12,13,14] employed TRA and TPB to evaluate WTP for renewable energy and solid waste–to–energy projects. Tan et al. [15] assessed residents’ WTP for battery swapping stations in Hangzhou using the TPB framework. Dong and Ge [16] incorporated factors such as recycling experience, policy incentives, and after-sales service to examine consumer intent to recycle retired EV batteries. Lin and Qiao [17] examined how policy design impacts public acceptance of green electricity. Gong et al. [18] explored the influence of perceived fiscal subsidies on WTP for clean heating. Zheng et al. [19] analyzed WTP for electric vehicles in China, introducing additional variables including performance expectancy, information load, and perceived risk.

While numerous countries are actively promoting research and applications in the technological development and policy design for SURB, there remains a notable gap in the literature regarding consumer acceptance and willingness to pay for SURB products. This distinction highlights the novelty of our work, which focuses on consumer behavioral perspectives rather than technical or policy aspects. Existing studies do not analyze the mutually complex relationship between the influencing factors. Finally, few studies have explored the causal pathways and mechanisms between incentive policies and WTP. This paper utilizes a combination of ISM and SEM to investigate this relationship, integrating the variables of incentive policy, knowledge, and perceived risk into the TPB model. This study justifies its use of both ISM and SEM by leveraging their complementary strengths. ISM was used first to map the hierarchical relationships between the factors, providing a clear and simplified structure for the subsequent SEM analysis. This sequential approach, supported by recent research, ensures that the SEM is both theoretically sound and empirically testable by clarifying complex relationships and refining hypothesized pathways [20,21].

Recent developments in metaheuristic optimization algorithms such as Particle Swarm Optimization (PSO), is inspired by the social behavior of birds flocking or fish schooling and optimizes a problem by iteratively improving a population of candidate solutions based on their individual and collective experience; Genetic Algorithms (GA), it mimics the process of natural selection and evolution, applying crossover, mutation, and selection operators to evolve a population toward optimal solutions [22]; and Grey Wolf Optimization (GWO), which is based on the leadership hierarchy and hunting strategy of gray wolves in nature, balancing exploration and exploitation effectively to avoid local optima, have significantly improved the efficiency, convergence stability, and robustness of parameter estimation in SEM [23]. Unlike traditional estimation methods, these algorithms are capable of efficiently exploring complex, non-convex solution spaces and are particularly useful in optimizing model fit and structural weights [22,24]. For instance, Yang et al. [23] applied an enhanced metaheuristic strategy within an SEM framework and demonstrated that it outperformed traditional estimation methods in terms of convergence and accuracy. Building on this, several high-impact studies successfully integrated such algorithms into hybrid ISM–SEM frameworks to improve factor weight estimation and model validation [22,23,24]. These advancements point to a promising direction for future research aiming to integrate intelligent optimization techniques within the ISM-SEM modeling process.

This study explores the factors influencing consumers’ WTP for SURB products. Experts in the field of new energy batteries were invited to evaluate the relationship among these factors. The ISM method was employed to construct a hierarchical structure of the factors. The SEM method examined how internal and external situational factors affect consumers’ WTP. The findings of this study will offer valuable insights for the advancement of SURB.

The paper is structured as follows: Section 2 develops the theoretical model and formulates the relevant research hypotheses. Section 3 outlines the research design, followed by the empirical analysis and results in Section 4. Section 5 is a discussion of the analysis results. Finally, Section 6 provides the conclusions and policy implications of the paper.

The main contributions of this study are threefold. First, it extends TPB by incorporating incentive policy, knowledge, and perceived risk, offering a more comprehensive framework for explaining consumer willingness to pay for SURB’s products. Second, it employs a combined ISM–SEM approach, enabling both the identification of hierarchical relationships among influencing factors and the empirical validation of causal pathways. Third, it provides practical policy insights into how governments and enterprises can reduce perceived risks and enhance consumer acceptance of SURB, thereby promoting sustainable battery use.

2. Theoretical Model and Research Hypotheses

2.1. Variable Selection

This study employs the TPB model to examine the factors influencing consumers’ willingness to pay for SURB products. Ajzen [25] proposed the Theory of TPB, which has since been widely adopted to explain human behavioral intentions. Stern [26] further emphasized the TPB’s applicability in environmental psychology, especially for studying pro-environmental behaviors.

The TPB posits that an individual’s behavior is determined by their intentions, which are influenced by their attitudes, subjective norms, and perceived behavioral control [25]. However, the factors that influence consumers’ WTP are complex and diverse. Therefore, this study considers the internal and external situational factors to comprehensively assess the factors that influence consumers’ willingness to pay for SURB products.

This study uses three extended variables: incentive policy, perceived risk, and knowledge. As a key external factor for government intervention in consumers’ green purchasing behavior, incentive policies positively impact the formation of consumers’ willingness to pay for green products [17]. Consumers often consider policy orientation when making purchasing decisions, making it essential to understand their perception of such policies [27]. Additionally, consumers consider the environmental performance of products and prioritize product quality and safety. As the development of SURB technology is not yet mature, the perceived risk of the product to the consumer is also an indispensable factor. Finally, consumers’ knowledge reserve is crucial for correctly judging such products and directly or indirectly affects their WTP.

2.2. Interpretative Structural Modeling

This section uses the ISM method to establish a hierarchical structural model for systematically decomposing the factors influencing the willingness to pay for SURB products. The aim is to identify the intrinsic connections between various influencing factors and to establish the relationship between them.

(1)

Establishment of the initial direct matrix

This study invited six experts in the field of new energy power batteries to use the Delphi method to score the mutual influence relationship between the indicators of willingness to pay for SURB products. To validate the consistency of expert responses in the Delphi process, Inter-Rater Reliability Analysis (Kendall’s W) was calculated across 42 pairwise comparisons. The result, W = 0.74, indicates a strong agreement among the six experts, exceeding the accepted threshold of 0.70 for reliability [28,29] (see

Table A5. Kendall’s W Inter-Rater Reliability Summary.

Number of Experts (n)	Number of Items (m)	Kendall’s W Interpretation	Threshold for Consensus	Degrees of Freedom (df)	p-Value (p)
6	42	0.74	≥0.70	9	<0.001

). This confirms the robustness of the expert input used in the ISM modeling.

The scores range from 1 to 5, indicating five grades of mutual influence relationship: none, weak, weaker, stronger, and strong. ${\alpha }_i$ denotes the $i$ indicator, $i$ = 1~7, where ${\alpha }_1~{\alpha }_7$ respectively denote subjective norms, perceived risk, knowledge, incentive policies, attitudes, WTP, and perceived behavioral control. The initial direct matrix, denoted by $C^m$, is expressed as $C^m=({\beta }_{ij}^m{)}_{n\times n}$, where ${\beta }_{ij}^m$ indicates the degree of direct influence evaluation of the indicator ${\alpha }_i$ on ${\alpha }_j$ made by the experts. Note that ${\alpha }_{ij}=0$ when $i=j$ as it will not influence itself. Let n be the number of experts, and the scores are aggregated using the averaging method according to Formula (1) to obtain the initial direct matrix ‘C’. Note: ${\beta }_{ij}$ denotes the direct influence of factor i on factor j. Superscripts (${\beta }_{ij}^m$) are only used when indicating the m-th iteration or derived matrices (reachability, transitivity) and are clearly defined where applicable.

$$\begin{gathered} {\beta }_{ij}={\displaystyle \frac{1}{n}}\sum _{m=1}^n {\beta }_{ij}^m, m=\mathrm{1,2},\dots ,n \\ C=\left(\begin{array}{ccccccc}0& 2.14& 2.49& 2.64& 3.28& 3.24& 1.64\\ 2.44& 0& 2.44& 2.89& 2.72& 3.19& 1.64\\ 2.95& 2.99& 0& 3.02& 3.61& 3.73& 1.83\\ 2.67& 3.21& 3.02& 0& 3.36& 3.19& 2.04\\ 2.48& 2.68& 2.93& 2.73& 0& 3.06& 1.81\\ 2.55& 2.89& 2.68& 2.82& 3.13& 0& 1.39\\ 1.58& 1.28& 2.03& 1.60& 2.94& 3.06& 0\end{array}\right) \end{gathered}$$

(1)

(2)

Establish the adjacency matrix and reachability matrix

The threshold λ = 3.01 was selected based on sensitivity analysis and aligned with ISM methodological standards (see

Table A2. Adjacency Matrix for λ = 3.0.

	1	2	3	4	5	6	7
1	0	1	0	0	0	0	0
2	0	0	1	0	1	0	1
3	0	1	0	1	0	0	1
4	1	0	1	0	1	0	0
5	0	1	0	0	0	1	1
6	0	1	1	0	1	0	1
7	0	1	0	0	1	0	0

,

Table A3. Adjacency Matrix for λ = 3.01.

	1	2	3	4	5	6	7
1	0	1	0	0	0	0	0
2	0	0	1	0	1	0	1
3	0	1	0	1	0	0	1
4	1	0	1	0	1	0	0
5	0	1	0	0	0	1	1
6	0	1	1	0	1	0	1
7	0	1	0	0	1	0	0

and

Table A4. Adjacency Matrix for λ = 3.1.

	1	2	3	4	5	6	7
1	0	1	0	0	0	0	0
2	0	0	0	0	1	0	1
3	0	1	0	1	0	0	1
4	1	0	1	0	1	0	0
5	0	1	0	0	0	1	1
6	0	1	1	0	1	0	1
7	0	1	0	0	1	0	0

). On a 5-point Delphi scale, this value distinguishes moderate-to-strong influences and avoids inclusion of weak or ambiguous links. This approach is consistent with previous studies [30,31] and ensures both conceptual clarity and structural robustness in the ISM hierarchy. Set the threshold $\lambda =3.01$ based on the initial direct matrix. Then, establish the adjacency matrix $\varphi$ using Equation (2), where 1 indicates a direct influence of the indicator. ${\alpha }_i$ on indicator ${\alpha }_j$, and 0 indicates no direct influence of the indicator ${\alpha }_i$ on indicator ${\alpha }_j$.

$$\begin{gathered} {\varphi }_{ij}=\left\{\begin{array}{c}1, {\beta }_{ij}\ge \lambda ,\\ 0, \beta ij<\lambda .\end{array}\right. \\ \varphi =\left(\begin{array}{ccccccc}0& 0& 0& 0& 1& 1& 0\\ 0& 0& 0& 0& 0& 1& 0\\ 0& 0& 0& 1& 1& 1& 0\\ 0& 1& 1& 0& 1& 1& 0\\ 0& 0& 0& 0& 0& 1& 0\\ 0& 0& 0& 0& 1& 0& 0\\ 0& 0& 0& 0& 0& 1& 0\end{array}\right) \end{gathered}$$

(2)

After obtaining the adjacency matrix, we find the maximum number of passes $K$ using Equation (3) in MATLAB R2023b (9.15) software. The unit matrix $I$ is used to obtain the reachability matrix $M$, as shown in Equation (4).

$$\begin{array}{c}(\varphi +I)\ne (\varphi +I{)}^{K-1}\ne (\varphi +I{)}^K=(\varphi +I{)}^{K+1}\end{array}$$

(3)

$$\begin{gathered} \begin{array}{c}M=(\varphi +I{)}^K\end{array} \\ M=\left(\begin{array}{ccccccc}1& 0& 0& 0& 1& 1& 0\\ 0& 1& 0& 0& 1& 1& 0\\ 0& 1& 1& 1& 1& 1& 0\\ 0& 1& 1& 1& 1& 1& 0\\ 0& 0& 0& 0& 1& 1& 0\\ 0& 0& 0& 0& 1& 1& 0\\ 0& 0& 0& 0& 1& 1& 1\end{array}\right) \end{gathered}$$

(4)

(3)

Regional division of the reachability matrix

The reachability matrix is regionally divided, i.e., the influencing factors are divided into reaches set $A\left({\alpha }_i\right)$, prior set $B\left({\alpha }_i\right)$, and a common set $C\left({\alpha }_i\right)$ to further study the hierarchical relationship among the indicators, as shown in

Table 1. Reachable set, prior set, and a common set of impact factor indicator systems.

Indicators	$\mathrm{Reachable}\mathrm{Set}A\left({\alpha }_i\right)$	$\mathrm{Prior}\mathrm{Set}B\left({\alpha }_i\right)$	$\mathrm{Common}\mathrm{Set}C\left({\alpha }_i\right)$
${\alpha }_1$	1, 5, 6	1, 2	1
${\alpha }_2$	2, 5, 6	2, 3, 4	2
${\alpha }_3$	2, 3, 4, 5, 6	3, 4	3, 4
${\alpha }_4$	2, 3, 4, 5, 6	3, 4	3, 4
${\alpha }_5$	5, 6	1, 2, 3, 4, 5, 6, 7	5, 6
${\alpha }_6$	5, 6	1, 2, 3, 4, 5, 6, 7	5, 6
${\alpha }_7$	5, 6, 7	7	7

.

(4)

Establishment of an interpretive structural model

Based on the reachable and standard sets obtained in Table 1, a hierarchical decomposition of the influencing factors ${\alpha }_i(i=1~7)$ is performed. The steps are as follows: firstly, identify the factors that correspond to $A\left({\alpha }_i\right)=C\left({\alpha }_i\right)$ and consider them as the first layer in the ISM. Secondly, remove the columns and rows corresponding to the factors in the first layer from the matrix to create a new reachable matrix. Finally, determine the factors included in the second layer based on the same principle. And so on, until each factor is categorized into the corresponding level. This paper classifies the seven factors influencing the willingness to pay for SURB products into three levels based on the layering principle. The top-level factors are ${\alpha }_5$ WTP and ${\alpha }_6$ attitude. The mid-level factors are ${\alpha }_1$ subjective norms, ${\alpha }_2$ perceived risk, and ${\alpha }_7$ perceived behavioral control. The bottom-level factors are ${\alpha }_3$ knowledge and ${\alpha }_4$ incentive policy. The results are presented in Figure 1.

As can be seen from Figure 1, knowledge and incentive policies are the bottom-level factors that directly affect the three mid-level influences of subjective norms, perceived risk, and perceived behavioral control. Therefore, starting with knowledge and incentives is important to explore certain perceived factors affecting consumers’ willingness to pay. The results of the ISM analysis differ from TPB, but they can still be interpreted to some extent. The difference between the ISM and the TPB lies in the placement of attitudes. In the ISM, attitudes are at the same level as WTP, while in the TPB model, attitudes are at a lower level and directly influence WTP. According to the TPB model, attitudes affect WTP despite being at the same level. Additionally, numerous studies have confirmed that subjective norms impact attitudes in the TPB, which is further supported here.

2.3. Preliminary Theoretical Model Based on ISM

The hierarchical structure between the factors influencing the willingness to pay for the SURB is revealed through ISM analysis. This paper constructs a preliminary SEM based on the results of the ISM analysis, assuming the validity of all relationships in the hierarchical structure diagram. The model shown in Figure 2 lays the foundation for the validation analysis of structural relationships. Based on this, we propose additional research hypotheses that consider relevant literature and the practical background of SURB. That will help us create a formal SEM, as shown in Figure 3.

2.4. Research Hypotheses

2.4.1. The Theory of Planned Behavior

The TPB model has been utilized to measure pro-environmental intentions and behaviors, such as the willingness towards voluntary carbon offsetting (VCOs), paying for circular products, sorting municipal solid waste, and adopting electric vehicles [8,9,10,11]. These studies found that attitudes, subjective norms, and perceived behavioral control positively affect green behavioral intentions. Furthermore, Reference [32] confirmed that environmental subjective and personal norms significantly influence intentions to rent electric vehicles.

Studies above indicate that consumers’ willingness to pay for eco-friendly products is influenced by their attitudes, subjective norms, and perceived behavioral control. Consumers with higher subjective norms are more likely to be influenced by others. If they have a positive attitude and believe it is easy to engage in, they are more likely to have a positive intention to perform the behavior. SURB products promote environmental benefits and are considered green products. According to the TPB, similar studies, ISM results, and the practice background of SURB products, the following hypothesis is developed:

H1a: 

Consumers’ positive attitudes have a positive impact on their willingness to pay for SURB products.

H1b: 

Consumers’ subjective norms have a positive effect on their willingness to pay for SURB products.

H1c: 

Consumers’ perceived behavioral control has a positive effect on their willingness to pay for SURB products.

There may also be some connection between subjective norms, perceived behavioral control, and attitudes. According to reference [33], the strength of the association between the variables in the TPB suggests that subjective norms may influence attitude. This indicates that when individuals develop their attitudes towards a behavior, they consider the expectations and perceptions of others. Several studies have also confirmed the positive influence of subjective norms on attitudes [32,33,34]. In addition, the more substantial consumers’ perceived behavioral control over the purchase of SURB products, the more likely they are to adopt a positive attitude towards these products, which suggests that perceived behavioral control may positively influence attitudes. According to the ISM results, as well as the practice context of SURB products, we propose the following hypotheses:

H1d: 

Consumers’ subjective norms have a positive effect on attitude.

H1e: 

Consumers’ perceived behavioral control has a positive effect on attitude.

2.4.2. Perceived Risk

Perceived risk is the uncertainty that arises when a person performs a specific behavior. This paper defines perceived risk as “the degree to which consumers perceive that using a SURB product may result in potential losses.” Studies have shown that perceived risk is a major barrier for consumers to purchase green products and is negatively correlated with consumers’ attitudes [35,36] and WTP [27,34,36,37,38]. Although SURB products are eco-friendly and innovative, their development is still immature, and consumers’ perceived risks for SURB products usually come from price, product characteristics, safety, and interpersonal interactions. These shortcomings may trigger consumers’ perceived risks, create negative attitudes towards SURB products, or lead to unwillingness to purchase these products. According to similar studies (pro-environmental behaviors), ISM results, and the practice context of SURB products, we propose the following hypotheses:

H2a: 

Consumers’ perceived risk negatively affects attitudes toward SURB products.

H2b: 

Consumers’ perceived risk for SURB products negatively affects WTP.

2.4.3. Incentive Policies

The Chinese government has introduced a series of policies to support the development of the battery secondary market. But at present, China’s policies for SURB products are still in the publicity stage, mainly regarding industry standards and technical specifications of the SURB. The specific economic policies, such as subsidies and tax reductions, have yet to be formulated. The incentive policies in this study refer to economic incentives and market standardization policies.

Most studies have shown that government incentive policies directly impact consumers’ WT. Liao [11] found that both fiscal and non-fiscal incentive policies had a positive influence on consumers’ willingness to adopt electric vehicles. Ajzen [25] also noted that a favorable perception of policy could increase residents’ acceptance of and compliance with related behaviors. Wang et al. [10] and Liao [11] supported the idea that stronger policy incentives were associated with greater behavioral willingness.

In addition, some studies have shown that government incentive policies indirectly affect consumers’ WTP. Dong and Ge [16] concluded that, in the context of China, battery recycling policies did not have a significant direct impact on consumers’ willingness to recycle. Some studies have pointed out that subsidy policies do not directly affect consumers’ green purchasing intentions but indirectly affect purchasing intentions through psychological factors such as knowledge and attitudes [39]. Therefore, by formulating relevant policies, the government can enhance the reliability of certain behaviors and improve society’s confidence in such products. Furthermore, introducing the policy will give the consumers more control over such purchasing behaviors, thus enhancing their perceived behavioral control. In addition, the better the government’s publicity policy is, the more information consumers learn about the product. Enriching product knowledge helps reduce consumers’ perceived risk, thereby enhancing consumers’ willingness to pay [40]. In this paper, based on the results of ISM analysis and existing research, the following hypotheses are proposed:

H3a: 

The secondary utilization incentive policies negatively affect consumers’ perceived risk.

H3b: 

The secondary utilization incentive policies positively affect consumers’ subjective norms.

H3c: 

The secondary utilization incentive policies positively affect consumers’ perceived behavioral control.

H3d: 

The secondary utilization incentive policies positively affect consumers’ WTP.

2.4.4. Knowledge

In this study, knowledge refers to understanding the characteristics of SURB products. Knowledge plays two roles in pro-environmental behaviors: firstly, it raises consumers’ awareness of environmental protection; Secondly, it provides consumers with scientific expertise in solving environmental problems [41]. And those who know about a product are more likely to purchase it. Hu et al. [42] confirmed that consumers’ perceptions of environmental benefits, affordability, and safety significantly accelerated the adoption of shared electric vehicles. In addition, in the study of whether the public supports the construction of battery swap stations, the more the public knows about such behaviors, the stronger their willingness to pay [34].

It has been shown that people with environmental knowledge are better able to understand the purpose of an organization’s implementation of an intervention, and they are more likely to perceive external expectations for engaging in environmental behaviors [43]. So, knowledge may positively influence consumers’ subjective norms. Consumers’ knowledge of external information will increase an individual’s beliefs about controlling certain behaviors, enhancing consumers’ perceived behavioral control, and thus influencing behavioral intentions [44].

In addition, consumers with insufficient product knowledge will be skeptical of their consumption behavior. Wang and Hazen [45] reported that consumers’ product knowledge was negatively associated with perceived risk. Liu et al. [46] further showed that perceived risk mediated the relationship between knowledge and the public’s acceptance of carbon capture, utilization, and storage (CCUS) technologies. Product knowledge reduces perceived risk and uncertainty, thereby increasing consumers’ WTP. Therefore, this paper proposes the following hypotheses based on the results of ISM analysis and existing research:

H4a: 

Knowledge related to SURB products negatively affects perceived risk.

H4b: 

Knowledge related to SURB products positively influences subjective norms.

H4c: 

Knowledge related to SURB products positively influences perceived behavioral control.

H4d: 

Knowledge related to SURB products positively influences WTP.

2.5. Formal Theoretical Model

Based on the preliminary structural equation model and the above hypotheses, the formal theoretical model of this study is determined, as shown in Figure 3, to prepare for the subsequent empirical analysis.

3. Methodology

To investigate the factors influencing consumers’ willingness to pay for SURB products, we conducted an empirical study using a questionnaire survey. We designed the survey scale and arranged personnel to conduct field surveys and data recovery based on the theoretical framework. We then pre-processed the recovered data to extract valid samples for descriptive statistics. As attitude, perceived behavioral control, subjective norms, perceived risk, incentive policy, knowledge, and WTP are psychological latent variables that cannot be measured directly, corresponding measurement questions must be set up for each latent variable to form measurement scales. This study employed the five-point Likert scale method, which ranges from ‘strongly disagree’ to ‘strongly agree’, with a score from 1 to 5. In this section, firstly, the questionnaire was designed based on existing literature; secondly, descriptive statistical analyses were also carried out to understand the sample data’s primary characteristics.

3.1. Questionnaire and Data Collection

The questionnaire is divided into two parts. The first part comprises six questions on the socio-economic attributes of the respondents. These include five single-choice questions on gender, age, education level, family size, and monthly income level. Additionally, there is one open-ended question on the respondent’s industry. The second part of the study presents a scale of 23 psychological latent variables that affect consumers’ willingness to pay for SURB products. These variables include attitude, perceived behavioral control, subjective norms, perceived risk, incentive policy, knowledge, and willingness to pay. The measurement variables were adapted from existing literature to suit the subject of this study, which focuses on consumers’ willingness to pay for SURB products, as shown in

Table 2. Questionnaire Scale.

Variable	Question Item	Source of Item
Perceived Risk (PR)	PR1: I feel that using the SURB products will cause me economic losses.	Reference [19] Reference [38]
	PR2: When I use the SURB products, I feel that they may not be safe.
	PR3: When I use the SURB products, I am worried that people around me may not understand my behavior.
Knowledge (KN)	KN1: I know the environmental benefits of battery secondary utilization, such as reducing the generation of harmful substances.	Reference [14]
	KN2: I know the economic benefits of battery secondary utilization, such as reducing resource consumption and saving costs.
	KN3: I know that the cost of the SURB product is lower and cheaper than a new battery product.
Incentive Policy (IP)	IP1: I would be more willing to buy it if I could get state subsidies for purchasing the SURB product.	Reference [12]
	IP2: The government’s policies in the economy and other aspects will boost the consumption of the SURB product.
	IP3: The government should give tax incentives and other financial support to companies that produce SURB products.
	IP4: The government should standardize the recycling network of power batteries.
Attitude (ATT)	ATT1: My evaluation of SURB is positive.	Reference [11]
	ATT2: I think battery secondary technology is very interesting.
	ATT3: I think the development prospects of SURB are broad.
Subjective norm (SN)	SN1: Society expects me to contribute economically to the development of SURB.	Reference [19]
	SN2: I feel that I have a moral obligation to support the development of SURB by purchasing related products.
	SN3: IF my parents, friends, and other significant individuals around me use SURB products, I also tend to choose such products.
Perceived behavior control (PBC)	PBC1: I have the ability to undertake SURB products (resources, time, and opportunities).	Reference [43]
	PBC2: As long as I am willing, I can obtain SURB products.
	PBC3: Whether or not to pay for SURB products depends entirely on me.
Willingness to pay (WTP)	WTP1: When I need it, I am willing to pay for SURB products.	Reference [43]
	WTP2: Considering energy conservation and environmental protection, I will use SURB products.
	WTP3: I will recommend SURB products to those around me.
	WTP4: Compared to other types of battery products, I will prioritize using SURB products.

.

The online questionnaire survey was conducted using Wenjuanxing Star. The sampling frame consisted of adult Chinese consumers reachable through established online networks, particularly within WeChat groups that included participants from various regions, industries, and socioeconomic backgrounds. Wenjuanxing Star (https://www.wjx.cn/, accessed on 14 August 2025) is a widely used online survey platform in China, comparable to internationally recognized platforms such as Qualtrics or SurveyMonkey. It is frequently adopted in academic research for questionnaire distribution, data collection, and statistical analysis [25,33]. Similarly, WeChat is China’s most popular social media and communication platform, with over 1.3 billion active users, making it a reliable and representative channel for reaching diverse demographic groups. Both tools have been extensively used in prior social science and consumer behavior studies in the Chinese context. Inclusion criteria required that respondents be at least 18 years old, have basic knowledge of electric vehicles or battery-powered products, and be willing to complete the survey independently. To enhance demographic representativeness, the research team monitored respondent distributions in real-time, ensuring balanced representation across key variables such as age, gender, education level, and income. Where discrepancies were observed during data collection, targeted outreach within underrepresented demographic segments was undertaken to correct sampling imbalances. This approach contributed to a more heterogeneous and reflective sample of the broader consumer base relevant to the context of secondary utilization of retired batteries. The survey link was shared in the WeChat group, and participants received red packets upon completion. A total of 432 questionnaire results were obtained. It employed a simple random sampling approach to ensure each potential respondent had an equal chance of being included in the sample, thereby enhancing the internal validity of the results. Participants were randomly selected from a pool of individuals accessible through established WeChat communities, representing diverse demographic and occupational backgrounds. To mitigate the risk of self-selection bias commonly associated with online and incentivized surveys, several measures were taken. First, the survey invitation emphasized the academic nature of the study rather than the incentive itself, minimizing participation motivated solely by financial gain. Second, responses were screened to exclude duplicate, incomplete, or patterned submissions, thereby improving data reliability. Third, demographic quotas were monitored during data collection to maintain proportional representation across age, gender, education, and income groups, enhancing the sample’s alignment with broader population characteristics. These steps collectively helped reduce bias and improve the generalizability of the findings within the study’s contextual limitations. To ensure the accuracy and reliability of the questionnaire data, incomplete and illogical questionnaires and those with duplicate responses were excluded. A total of 406 valid questionnaires were obtained, resulting in an effective rate of 93.98%.

According to the sample size Formula (5), a 95% confidence level was selected. The standard deviation SD was generally chosen to be 0.5. The margin of error ME was 0.05, and the Z (Z-scores) was 1.96, which was substituted to obtain sample size N = 385. The sample size of this study was 406, which meets the minimum sample size requirement. In the study of consumers’ willingness to pay for circular products [9], N = 395. In the study of residents’ willingness to pay for energy from anaerobic digestion in Jiangsu Province [14], N = 400. The sample sizes of many studies are around 400, so the sample sizes selected for this study meet the requirements.

$$N={\displaystyle \frac{Z\times SD\times \left(1-SD\right)}{M{E}^2}}$$

(5)

3.2. Demography of Respondents

Table 3. Descriptive statistical analysis of the sample’s basic characteristics.

Sample	Category	Frequent	Percentage
Gender	Male	265	65.2%
	Female	141	34.8%
Age	Under 20	1	0.3%
	21–30	144	35.5%
	31–40	221	54.5%
	41–50	33	8.2%
	Above 50	6	1.5%
Educational Level	High School and Below	7	1.8%
	Junior College	43	10.6%
	Bachelor	227	55.8%
	Master’s	119	29.4%
	PhD	10	2.4%
Family Size	2 or fewer	52	12.7%
	3	147	36.1%
	4–5	158	38.8%
	Above 5	50	12.4%
Monthly Income Level	Below 10,000 RMB	176	43.3%
	10,001–15,000 RMB	112	27.6%
	15,001–20,000 RMB	56	13.9%
	Above 20,000 RMB	62	15.2%
Industry	Open-ended question

shows the descriptive statistical analysis results of the data from 406 respondents using SPSS software. The respondents consisted of males (65.2%) and females (34.8%), and the age of the respondents was mainly concentrated in the age group of 21–40 years (90%), which was relatively young reflecting the core demographic of digitally engaged, early- to mid-career consumers in China; among all the respondents, only 7 (1.8%) had a high school degree and below, 43 (10.6%) had a college degree, and 356 (87.6%) had a bachelor’s degree or higher, with a higher overall level of education. 176 (43.3%) had a monthly income of less than RMB 10,000, while 62 (15.2%) had a monthly income of more than RMB 20,000. The last open-ended question about industry could not be described by SPSS software. Manual analysis of the completed data showed that the respondents were involved in a wide range of industries, such as manufacturing, Internet, computer software, energy, and finance, with manufacturing accounting for the highest percentage, followed by the computer industry.

4. Data Analysis and Results

4.1. Measurement Model

In this study, SPSS (Version 26.0) and AMOS (Version 24.0) were used to analyze the data, and the quality of the scale was verified before the hypothesis testing of the SEM, including reliability and validity tests.

Reliability gauges the consistency of measurement data, determining whether the measurement indices effectively capture the underlying variables they intend to measure. The factor loading coefficients, Cronbach’s alpha (Cronbach’s α), and Composite Reliability (CR) are usually used to evaluate the measurements. A scale with good reliability should have factor loading coefficients, Cronbach’s α, and CR greater than 0.7 [47].

Validity analysis is used to test the validity and accuracy of the scale design, with higher validity indicating better accuracy. In layperson’s terms, to prove validity, the measurement variable must achieve the intended purpose of the measurement based on the data collected. When the AVE exceeds 0.5, the latent variables have better convergent validity [47]. At the same time, if the square root of the AVE is greater than the correlation coefficient between the latent variables, it indicates that the scale has better discriminant validity.

Table 4. Validity and reliability test results.

Variable	Question Item	Standardized Factor Loading	Cronbach’s α	Composite Reliability	Average Variance Extracted
Perceived Risk (PR)	PR1	0.732	0.785	0.787	0.554
	PR2	0.677
	PR3	0.817
Knowledge (KN)	KN1	0.77	0.821	0.821	0.604
	KN2	0.79
	KN3	0.771
Incentive Policy (IP)	IP1	0.783	0.842	0.843	0.574
	IP2	0.763
	IP3	0.694
	IP4	0.786
Attitude (ATT)	ATT1	0.739	0.774	0.773	0.532
	ATT2	0.743
	ATT3	0.705
Subjective norm (SN)	SN1	0.741	0.769	0.770	0.528
	SN2	0.72
	SN3	0.718
Perceived behavior control (PBC)	PBC1	0.729	0.766	0.767	0.524
	PBC2	0.688
	PBC3	0.753
Willingness to pay (WTP)	WTP1	0.761	0.843	0.841	0.570
	WTP2	0.734
	WTP3	0.795
	WTP4	0.729

presents the test results of the reliability and validity of the scales. Most of the factor loading coefficients for each measurement variable were greater than 0.7, Cronbach’s α and CR were greater than 0.7, and the AVE was greater than 0.5, which indicated that the collected data had strong reliability and convergent validity.

Table 5. Discriminant validity test results.

Variable	AVE	PR	KN	IP	ATT	SN	PBC	WTP
PR	0.544	0.744
KN	0.604	−0.311	0.777
IP	0.574	−0.275	0.345	0.758
ATT	0.532	−0.216	0.169	0.144	0.729
SN	0.528	−0.055	0.086	0.178	0.273	0.727
PBC	0.524	−0.174	0.389	0.413	0.169	0.080	0.724
WTP	0.570	−0.388	0.235	0.254	0.478	0.300	0.366	0.755

Note: Diagonal elements are the square roots of AVE, while nondiagonal elements are the correlation coefficients between latent variables.

shows the results of the test of the scale’s discriminant validity. The square root of the AVE (diagonal elements) is greater than the absolute value of the nondiagonal elements, indicating that the scale has good discriminant validity.

The bootstrap resampling analysis revealed that most relationships in the ISM were retained with frequencies ranging from 0.36 to 0.43 across 1000 iterations. This range indicates a moderate level of structural stability, suggesting that the ISM hierarchy is reasonably robust to expert variability (see

Table 6. Bootstrap Mean Matrix of ISM Hierarchy Stability.

Elements	GS	TI	EA	EI	RB	EC	LP
Government Support (GS)	1.00	0.40	0.40	0.39	0.41	0.38	0.39
Technological Infrastructure (TI)	0.38	1.00	0.40	0.40	0.40	0.41	0.41
Environmental Awareness (EA)	0.41	0.42	1.00	0.38	0.41	0.40	0.43
Economic Incentives (EI)	0.38	0.42	0.41	1.00	0.37	0.41	0.36
Recycling Behavior (RB)	0.39	0.40	0.39	0.40	1.00	0.37	0.39
Educational Campaigns (EC)	0.37	0.38	0.39	0.38	0.38	1.00	0.36
Legislative Pressure (LP)	0.40	0.41	0.41	0.39	0.40	0.39	1.00

).

4.2. Structural Model

4.2.1. Goodness of Fit

To ascertain the degree of fit of the hypothetical model, the SEM was analyzed with the help of fit indices, including the chi-square to degrees of freedom ratio (CMIN/DF), the goodness fit index (GFI), the adjusted goodness fit index (AGFI), the value-added fit indices (NFI, TLI), the standardized residual mean squared sum square root (SRMR), the root mean squared error of approximation (RMSEA), and the parsimony fit indices (PCFI, PGFI).

Table 7. Model fittest results.

Fit Index	Benchmark Values	Model	Model Fit Judgment
CMIN/DF	1~3	1.217	Y
GFI	>0.9	0.938	Y
AGFI	>0.9	0.920	Y
NFI	>0.9	0.911	Y
TLI	>0.9	0.980	Y
SRMR	<0.08	0.063	Y
RMSEA	<0.05	0.026	Y
PCFI	>0.5	0.835	Y
PGFI	>0.5	0.731	Y

presents some of the model fit metrics and the benchmark values. As can be seen from the table, the results of the hypothesized model fit test are all above the benchmark values, indicating that the hypothesized model can explain reality and can be accepted.

4.2.2. Path Analysis

The structural model of this study comprises six latent variables: incentive policy, knowledge, perceived risk, subjective norms, perceived behavioral control, attitude, and WTP. The causal relationship between these variables has been hypothesized in the previous section. To verify the hypotheses, examining the path coefficients and their respective significance levels is essential.

Figure 4 presents the results of path analysis. The solid line represents a statistically significant relationship, while the dotted line represents a non-significant relationship except for the following hypotheses: “H1e, H3d, H4b, H4d”. Other paths are significant, and the hypotheses are valid.

Table 8. Path Analysis Results.

Hypothesis	Path	Estimate	S.E.	C.R.	p	Results
H1a	ATT→WTP	0.345	0.073	4.870	0.000 ***	Y
H1b	SN→WTP	0.204	0.069	2.940	0.003 **	Y
H1c	PBC→WTP	0.326	0.082	3.979	0.000 ***	Y
H1d	SN→ATT	0.260	0.074	3.521	0.000 ***	Y
H1e	PBC→ATT	0.121	0.075	1.619	0.105	N
H2a	PR→ATT	−0.165	0.064	−2.563	0.01 *	Y
H2b	PR→WTP	−0.276	0.065	−4.255	0.000 ***	Y
H3a	IP→PR	−0.200	0.062	−3.217	0.001 **	Y
H3b	IP→SN	0.146	0.057	2.542	0.011 *	Y
H3c	IP→PBC	0.284	0.056	5.088	0.000 ***	Y
H3d	IP→WTP	−0.105	0.059	−1.762	0.078	N
H4a	KN→PR	−0.246	0.066	−3.706	0.000 ***	Y
H4b	KN→SN	0.030	0.060	0.507	0.612	N
H4c	KN→PBC	0.248	0.058	4.290	0.000 ***	Y
H4d	KN→WTP	0.007	0.060	0.121	0.904	N

Significant at: p < 0.05 *, p < 0.01 ** and p < 0.001 ***.

lists the paths corresponding to the 15 hypotheses, as well as the path coefficients and p-values. ATT (β = 0.345, p < 0.001), SN (β = 0.204, p < 0.01), and PBC (β = 0.326, p < 0.001) have a positive effect on WTP, confirming H1a, H1b, and H1c. SN (β = 0.260, p < 0.001) had a positive impact on ATT, while PBC (β = 0.121, p = 0.097) had an insignificant positive impact on attitude, thus validating H1d and rejecting H1e.

PR showed a negative effect on ATT (β = −0.165, p < 0.05) and WTP (β = −0.276, p < 0.001), affirming H2a and H2b. IP had a negative effect on PR (β= −0.200, p < 0.01) and a positive effect on SN (β = 0.146, p < 0.05) and PBC (β = 0.284, p < 0.001), validating H3a, H3b, and H3c. There was no direct significant relationship between IP and WTP (β = −0.105, p = 0.078), rejecting H3d. KN indicates a negative effect on PR (β = 0.246, p < 0.001), an insignificant positive effect on SN (β = 0.030, p = 0.612) and WTP (β = 0.007, p = 0.904), and a positive effect on PBC (β = 0.248, p < 0.001), confirming H4a and H4c, rejecting H4b and H4d. By comparing the absolute value of the direct effect on WTP, it can be obtained that the largest direct effect on WTP is ATT, followed by PBC, PR, and SN.

4.3. Mediating Effects Analysis

This study includes four mediating variables. Firstly, ATT as a mediating variable is analyzed. Table 8 shows the results of the mediating effect test of ATT, which demonstrates the estimated values of the direct and indirect effects and the corresponding p-values. The total impact of all three paths was significant. There is a partial mediating effect of ATT between PR and WTP, which is negative, suggesting that PR can inhibit WTP by influencing ATT. ATT also has a partially mediated effect between SN and WTP, and this mediation effect is positive, indicating that SN can form positive attitudes and thus promote WTP. Although the direct effect of PBC on WTP is significantly positive, the indirect impact on WTP through ATT is not significant, and there is no mediation effect.

The mediating effect between IP\SN→WTP is then analyzed. As previously indicated in

Table 9. Mediation affects test results (1).

Path	Total Effects		Direct Effects		Indirect Effects		Result
	Estimate	p	Estimate	p	Estimate	p
PR→ATT→WTP	−0.334	0.000 ***	−0.284	0.000 ***	−0.060	0.005 **	Partial
SN→ATT→WTP	0.273	0.000 ***	0.188	0.003 *	0.085	0.000 ***	Partial
PBC→ATT→WTP	0.330	0.000 ***	0.292	0.000 ***	0.038	0.103	No

Significant at: * p < 0.05, ** p < 0.01 and *** p < 0.001.

, neither IP nor KN has a significant direct effect on WTP. Consequently, it can be concluded that this is a fully mediated effect. It is necessary to investigate the indirect and total effects of IP and KN on WTP separately to determine (1) whether IP and KN have an indirect effect on WTP significantly and (2) whether the four mediating variables: PR, SN, PBC, and ATT have mediating effects (3) if they do, which mediating effect accounts for the most significant proportion and explains the most.

Table 10. Mediation effects test results (2).

Path	Estimate	Confidence Interval	p	Percentage
IP→PR→WTP	0.061	(0.020, 0.117)	0.001 **	25.85%
IP→SN→WTP	0.033	(0.005, 0.082)	0.013 *	13.98%
IP→PBC→WTP	0.102	(0.047, 0.177)	0.000 ***	43.22%
IP→PR→ATT→WTP	0.013	(0.003, 0.033)	0.005 **	5.51%
IP→SN→ATT→WTP	0.015	(0.004, 0.035)	0.008 **	6.36%
IP→PBC→ATT→WTP	0.013	(−0.002, 0.032)	0.088	5.51%
IP→⋯→WTP	0.236	(0.145, 0.344)	0.000 ***	100%
KN→PR→WTP	0.072	(0.027, 0.139)	0.000 ***	37.11%
KN→SN→WTP	0.007	(−0.021, 0.041)	0.555	3.61%
KN→PBC→WTP	0.086	(0.038, 0.153)	0.000 ***	44.33%
KN→PR→ATT→WTP	0.015	(0.004, 0.036)	0.003 **	7.73%
KN→SN→ATT→WTP	0.003	(−0.010, 0.017)	0.560	1.55%
KN→PBC→ATT→WTP	0.011	(−0.002, 0.030)	0.081	5.67%
KN→⋯→WTP	0.194	(0.106, 0.293)	0.000 ***	100%

Significant at: * p < 0.05, ** p < 0.01 and *** p < 0.001.

presents the results of the indirect and mediated effects tests for IP\SN→WTP. These results indicate that both IP (β = 0.236, p < 0.001) and KN (β = 0.194, p < 0.001) have an indirect effect on WTP. Subsequently, the analysis was continued using the bias-corrected percentile bootstrap method, whereby the sampling was repeated 2000 times, and 95% confidence intervals were calculated. This method does not require that the samples obey a normal distribution and has a large amount of data, and therefore, represents the most appropriate method for studying the mediating effect. Table 9 lists 12 mediating paths and the corresponding mediating effect estimates, confidence intervals, and p-values. The last column presents the proportion of each mediating path. It can be observed that four mediating paths, namely “IP→PBC→ATT→WTP”, “KN→SN→WTP”, “KN→SN→ATT→WTP”, and “KN→PBC→ATT→WTP”, were found to be insignificant. The mediating effect of PBC was the most significant, accounting for 43.22%/44.33%, followed by PR, which indicates that PBC and PR were the most essential mediating variables between IP/KN and WTP.

4.4. Multigroup Analysis

To ascertain whether there are differences between different groups, this paper draws on existing studies and divides the overall sample into various groups. Except for gender, the other four characteristic factors were found to be distributed in a scattered manner. To facilitate analysis, they were regrouped using the statistical software package SPSS. Age was divided into two groups: low (35.8%, below 30 years old) and high (64.2%, above 30 years old). Education level was divided into low (68.2%, bachelor’s degree and below) and high (31.8%, master’s and doctoral degree). Family size was divided into two groups: few (48.8%, three persons and below) and large (51.2%, more than three persons). Finally, the monthly income level was divided into low (43.3%, less than 10,000) and high (56.7%, more than 10,000). Subsequently, the results of the AMOS analysis were obtained separately for each group, as shown in

Table 11. Multigroup analysis results.

Hypothesis	Gender		Age		Educational Level		Family Size		Monthly Income Level
	Male	Female	Low	High	Low	High	Few	More	Low	High
H1a	0.323 ***	0.354 ***	0.474 ***	0.341 ***	0.346 ***	0.389 **	0.359 **	0.323 ***	0.314 **	0.377 ***
H1b	0.107 *	0.211 **	0.302 **	0.111	0.19 *	0.211	0.17	0.236 *	0.23 *	0.171
H1c	0.305 **	0.321 ***	0.377 **	0.26 *	0.292 **	0.424 ***	0.378 **	0.303 **	0.293 *	0.35 ***
H1d	0.285 **	0.257 ***	0.223	0.313 ***	0.209 *	0.414 **	0.123	0.356 **	0.249 *	0.303 **
H1e	0.181	0.12	−0.029	0.239 *	0.108	0.157	0.234 *	0.002	0.067	0.134
H2a	−0.141	−0.169 **	−0.217 *	−0.132	−0.152	−0.153	−0.199 *	−0.147	−0.142	−0.194 *
H2b	−0.333 ***	−0.27 ***	−0.342 ***	−0.218 **	−0.197 *	−0.45 ***	−0.351 ***	−0.249 **	−0.313 ***	−0.24 **
H3a	−0.209 *	−0.179 **	−0.298	−0.139 *	−0.237 *	−0.034 **	−0.193 **	−0.198 *	−0.196 **	−0.178 *
H3b	0.09	0.127 *	0.165	0.142	0.117	0.164	0.186	0.119	0.097	0.227 *
H3c	0.219 **	0.253 ***	0.495 ***	0.174 *	0.191 **	0.465 ***	0.149	0.367 ***	0.276 ***	0.269 **
H3d	−0.161 *	−0.108	−0.067	−0.128	−0.054	−0.19	−0.078	−0.129	−0.162	−0.042
H4a	−0.212 *	−0.242 **	−0.293	−0.219 *	−0.213	−0.276 *	−0.169	−0.321 **	−0.236	−0.248 **
H4b	−0.014	0.028	0.065	−0.002	0.004	0.081	0.069	−0.015	0.165	−0.09
H4c	0.227 **	0.254 ***	0.313 **	0.208 **	0.21 **	0.27 *	0.275 **	0.211 *	0.276 **	0.218 **
H4d	0.006	0.02	−0.034	0.057	0.057	−0.116	0.089	−0.099	0.077	−0.047

Significant at: * p < 0.05, ** p < 0.01 and *** p < 0.001.

. It can be observed that the individual characteristics of consumers exert a slight influence on the path coefficients, yet the positivity and negativity of the path coefficients remain largely unaffected. The path coefficient for SN→WTP is most affected by the type of consumer, and there are differences in significance between the same consumers in the paths of IP→PR and KN→PR.

4.4.1. SN→WTP

The analysis results revealed that, except for the gender group, only the groups with low age, low education level, large family size, and low monthly income were significant. The following factors may elucidate this outcome. First, individuals in the demographic characterized by younger age and lower monthly income tend to opt for cost-effective products due to their limited economic capacity. Moreover, they are more susceptible to peer influence, particularly from those who have previously purchased SURB products. Second, individuals with lower education levels often lack comprehensive product knowledge, leading them to rely on others’ opinions rather than making independent decisions. Third, consumers from more prominent families are inclined toward collective decision-making, influenced by the dynamics of their upbringing in environments with multiple family members.

Therefore, it would be more effective for the groups to implement suitable measures to strengthen their subjective norms to enhance their WTP because these groups are more vulnerable to external influences. Relevant authorities can cultivate a social atmosphere conducive to making purchases.

4.4.2. KN/IP→PR

Except for the gender group, the impact of KN on PR (H4a) was only significant in the high age group, high education level group, large family size group, and high monthly income group. In contrast, the influence of IP on PR (H3a) is significant across nearly all groups, suggesting that the effectiveness of the incentive policy factor in reducing consumers’ perceived risk exceeds that of knowledge. This could be attributed to the fact that policies are already established and relatively easy to understand, while knowledge about secondary use remains limited and complex. Therefore, it is crucial and imperative to disseminate information about policies related to secondary use to the public.

5. Discussion

This study expands upon the TPB model by incorporating perceived risk, incentive policy, and knowledge to explore consumers’ willingness to pay for SURB products. The initial step is constructing a preliminary research model by building an ISM. Then, a formal theoretical framework was designed, and research hypotheses were proposed based on the literature review findings.

ISM indicates that incentive policy and knowledge exert a bottom-level influence on consumers’ WTP, while consumers’ perceived risk, perceived behavioral control, and subjective norms exert a mid-level influence. The fit indicators of the extended TPB model are all within the acceptable range, thereby confirming the predictive validity of the theoretical framework proposed in this paper.

The results of the SEM analysis indicate that the remaining six factors have a significant direct or indirect effect on WTP. First, Consumers’ attitudes have the most important direct impact on WTP, followed by perceived behavioral control. Furthermore, consumers’ subjective norms also positively and significantly affect WTP. This is because individuals tend to consider the opinions of significant others in their lives and societal expectations when making purchasing decisions. This is consistent with Chinese consumers’ willingness toward voluntary carbon offsetting [8]. Moreover, the study indicates that the greater the perceived risk, the lower the consumers’ willingness to pay for SURB products, consistent with previous studies [15]. It can be concluded that reducing the uncertainty associated with the products is an important factor in promoting consumers’ WTP.

While the findings offer valuable insights into Chinese consumers’ willingness to pay for SURB products, their applicability to other countries or markets may be constrained by several contextual factors. Culturally, consumer attitudes toward environmental sustainability, trust in recycled products, and the influence of social norms vary significantly across regions, potentially altering behavioral responses. Regulatory environments also differ; China’s top-down policy incentives and government-led market interventions may not be mirrored in countries with more liberal or decentralized regulatory structures. Economically, income levels, purchasing power, and the maturity of the electric vehicle and recycling infrastructure can influence the feasibility and attractiveness of SURB products. Therefore, while the theoretical model may hold conceptual relevance, the specific behavioral patterns and policy implications observed in this study may not fully generalize to markets with different socio-cultural or institutional contexts.

Next, enhancing consumers’ subjective norms is facilitated by policy incentives, consistent with the findings of willingness to pay for renewable green electricity [12]. The involvement of the government and the formulation of incentive policies are conducive to the formation of social awareness. Furthermore, the incentive policy can also reduce consumers’ perceived risk, which is consistent with the findings of the study on the willingness to pay for a new energy vehicle consumption promotion policy [11]. The formulation of the policy has led to an enhancement of the reliability of SURB products. Moreover, consumers’ knowledge about the product significantly impacts their perceived risk, which aligns with the study’s findings on the willingness to pay for circular products [9]. Therefore, the more an individual understands the products, such as the environmental and economic benefits brought to society, the less risk they will perceive.

6. Conclusions and Policy Implications

6.1. Conclusions

This study offers three key contributions that advance both theoretical understanding and practical application in the context of consumer behavior toward SURB products. First, it enriches TPB by integrating perceived risk, knowledge, and incentive policy factors, which are particularly relevant in emerging green product markets. This extended framework captures the complexity of consumer decision-making more effectively than traditional TPB models. Second, the sequential application of ISM followed by SEM provides a novel methodological contribution. ISM revealed the hierarchical structure among variables, enabling a theoretically grounded and empirically robust SEM, an approach that is increasingly being validated in sustainability research. Third, the study generates actionable insights for policymakers and industry stakeholders: while attitudes and perceived behavioral control are the strongest direct drivers of willingness to pay, incentive policies and knowledge exert significant indirect effects by reducing perceived risk and enhancing consumers’ sense of control. These findings offer a roadmap for designing more effective policy interventions and communication strategies to promote the adoption of SURB products. Together, these contributions advance the theoretical development of pro-environmental behavior models and support evidence-based policy-making in circular economy initiatives.

6.2. Policy Implications

The study’s findings provide important guidance for advancing the adoption of SURB products through strategic policy interventions and public engagement. First, government support should focus on strengthening incentive mechanisms such as targeted subsidies, tax benefits, and investment in recycling infrastructure to stimulate industry participation and lower consumer entry barriers. Regulatory efforts must also prioritize the standardization of recycling channels, certification of technicians, and third-party product quality verification to enhance trust and ensure safety in the SURB market.

Second, improving consumer knowledge is critical for reducing perceived risk and increasing the willingness to pay. Government agencies, in collaboration with enterprises and professional associations, should actively promote SURB products through public campaigns, digital media, and community events. Highlighting the environmental and economic benefits of these products, while involving consumers in the feedback and co-design processes, can foster greater product familiarity and acceptance.

Finally, cultivating a culture of green consumption is essential. Public messaging should emphasize the societal and environmental significance of choosing SURB products. At the same time, enterprises should invest in product quality, innovation, and reliable after-sales service to address consumer concerns and reinforce confidence in the value of sustainable battery solutions.

6.3. Limitations

This paper provides theoretical support and practical guidance for the publicity and promotion of consumers’ willingness to pay for SURB products. Nevertheless, due to the intricate nature of consumer purchasing behavior, this paper is still not exhaustive in elucidating the complex buying behaviors of consumers. Also, the limited spread of values in the direct matrix is acknowledged as a potential limitation. Future research could benefit from involving a more diverse expert panel to improve input variation and reduce potential response bias. It also does not consider the various types of incentive policies and the suppliers of SURB products. In future research, factors such as SURB technology, corporate brand preference, and categorization of policies into public infrastructure policies, manufacturing subsidy policies, and consumer subsidy policies can be employed to delve deeper into analyzing the determinants impacting consumers’ willingness to pay for SURB products.