Can carbon labels shift consumers towards sustainable food? Evidence from Chinese consumers

Xu, Yalin et al. Article — Published Version Can carbon labels shift consumers towards sustainable food? Evidence from Chinese consumers Sustainable Futures Provided in Cooperation with: Leibniz Institute of Agricultural Development in Transition Economies (IAMO), Halle (Saale) Suggested Citation: Xu, Yalin et al. (2024) : Can carbon labels shift consumers towards sustainable food? Evidence from Chinese consumers, Sustainable Futures, ISSN 2666-1888, Elsevier, Amsterdam, Vol. 8, pp. 1-13, https://doi.org/10.1016/j.sftr.2024.100363 , https://www.sciencedirect.com/science/article/pii/S2666188824002120 This Version is available at: https://hdl.handle.net/10419/306867 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc/4.0/ Can carbon labels shift consumers towards sustainable food? Evidence from Chinese consumers Yalin Xu a , Zhiwen Zhang a , Yanjun Ren a,b , Rong Yuan c , Yanan Wang a,* , Rui Li a , Shunan Zhao a , Lu Qiu a a College of Economics and Management, Northwest A&F University, Yangling 712100, China b Leibniz Institute of Agricultural Development in Transition Economies (IAMO), Theodor-Lieser-Str. 2, 06120 Halle (Saale), Germany c School of Economics and Business Administration, Chongqing University, Shazhengjie 174, 400040 Chongqing, China ARTICLE INFO Keywords: Carbon-labeled agricultural products Willingness to pay Theory of planned behavior Norm activation model China ABSTRACT Carbon labels are becoming an essential tool for policymakers in many countries to promote low-carbon consumption. To assess customers’willingness to pay for five carbon-labeled agricultural products (CAP), we used payment card to conduct a questionnaire survey among 641 respondents in Shanghai, Nanjing, Wuhan, and Xi’an, all located in China. This study quantitatively analyzes the influencing factors and interactive mechanisms of the publicʼs willingness to purchase CAP through the extended Theory of Planned Behavior and Norm Activation Model. The results show that consumers’willingness to pay a premium for carbon-labeled milk, corn, bananas, tomatoes, and eggs is 27.50 %, 29.73 %, 26.86 %, 26.51 %, and 24.26 % respectively. Perceived behavioral control has the strongest positive influence on purchase intention, followed by subjective norms, attitudes toward the behavior, and personal norms. There is a significant mediating effect between awareness of consequences and personal norms, which indirectly influences personal norms through the ascription of responsibility and subsequently affects purchase intention. Additionally, there is a gap between purchase intention and behavior, and risk perception negatively moderates the relationship between the two. Based on the research findings of this paper, practical and effective policy suggestions are proposed for the government to promote carbon labeling policies and reduce carbon emissions. 1. Introduction There is growing awareness of the impact of food choices on climate change. The food system is estimated to be responsible for 26–34 % of global greenhouse gas (GHG) emissions [1,2]. Recent modeling indicates that even if fossil fuel emissions were to cease immediately, current trends in the world’s food system would make it challenging to achieve the IPCC’s 1.5 ◦C objective. By the end of the century, these trends would threaten the 2 ◦C goal [3]. Though food producers should focus on lessening their environmental impact, consumer behavior changes can also impact production system improvements [2]. Shifting to low-carbon footprint diets has the potential to significantly decrease carbon emissions and alleviate the strain on the environment [4–6]. Accordingly, carbon label products are a good low-carbon consumption orientation, and customers can reduce their carbon footprint by purchasing food that is less harmful to the environment [7,8]. Among the policy tools used to promote the development of lowcarbon economy, carbon labels have been implemented in several countries, such as the United Kingdom, Germany, France, Sweden, and the United States [9]. Carbon labels refer to using quantitative measures on product labels to indicate the amount of greenhouse gas released during production [10]. According to Edwards-Jones et al. [11], carbon labels can help reduce greenhouse gas emissions. Regarding corporate emissions reduction, Shi [12] pointed out that carbon labels can bring economic benefits to companies, such as reducing emissions and achieving cost savings. At the consumer level, providing information to consumers about the carbon content of products through labels can assist them in making informed decisions about purchasing low-carbon products. This, in turn, contributes to the overall reduction of carbon emissions [13]. Despite the need for a significant change in personal food consumption, research indicates that people rarely consider their dietary choices when asked what they can do to help the environment [14] and often underestimate the environmental impact of their food [15,16]. Meanwhile, in some cases, consumers may be reluctant to pay more for carbon-labeled products. The general lack of awareness * Corresponding author at: College of Economics and Management, Northwest A&F University: NO.3 Taicheng Road, Yangling, Shaanxi, China. E-mail address: [email protected] (Y. Wang). Contents lists available at ScienceDirect Sustainable Futures journal homepage: www.sciencedirect.com/journal/sustainable-futures https://doi.org/10.1016/j.sftr.2024.100363 Received 13 June 2024; Received in revised form 30 August 2024; Accepted 1 November 2024 Sustainable Futures 8 (2024) 100363 Available online 3 November 2024 2666-1888/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC license ( http://creativecommons.org/licenses/bync/4.0/ ). regarding how dietary decisions affect the environment remains a major obstacle to reducing emissions of greenhouse gases caused by food [17]. Globally, there is a growing body of research on carbon labeling, particularly regarding how it affects consumer purchasing behavior [18, 19]. Several studies have been done on consumers from various geographic and cultural backgrounds [20]. Nevertheless, even if there have been some preliminary investigations of research on Chinese consumers, such as a study of carbon-labeled beef among urban Chinese customers [21], additional empirical research is required to obtain a deeper understanding of the preferences and behavioral patterns of various consumer groups. Currently, governments, producers, and merchants have not been able to receive comprehensive promotional advice or guidance. To achieve the transition to a low-carbon consumption pattern, China should integrate carbon labeling into a wider range of consumer product categories. To address the research gaps mentioned above, this paper aims to investigate Chinese consumers’ perceptions of carbon-labeled agricultural products (CAP), including their willingness to pay and the specific amount they are willing to pay for different types of CAP. The innovations are as follows: firstly, this study creates a more thorough explanatory framework, integrates the Norm Activation Model (NAM) and the Theory of Planned Behavior (TPB), and expands the original model by adding external variables (such as risk perception, low-carbon awareness, and personal knowledge). All these enhancements significantly improve the modelʼs capacity to explain the willingness of the CAP and purchasing behavior. In addition to examining the direct effects of respondents’subjective norms, personal norms, and perceived behavioral control on purchase intentions, the study sheds light on the mediating variables these factors use to affect purchase intentions and behavior. Additionally, this study confirms that risk perception moderates the relationship between buy intention and action, providing a more comprehensive explanation for the inconsistency between consumer intention and behavior when purchasing CAP. Secondly, information intervention is included in the experiment. This study uses an experimental design to explore how providing specific information on carbon labeling affects the publicʼs WTP for CAP and how the certainty that respondents will choose to improve the environment changes. Through empirical research in specific cultural and market backgrounds, we propose policies to promote the implementation of carbon labels at all levels, including government and enterprises, to encourage widespread participation in environmental protection and low-carbon activities, ultimately achieving “carbon reduction for all and benefits for all.” The remainder of the paper is structured as follows: Section 2 presents the theoretical framework and literature review, Section 3 describes the survey’s methodology and data collection, Section 4 presents the model’s results, Section 5 discusses the findings, and the final section raises policy implications. 2. Literature review and hypotheses 2.1. Carbon labeling concept There is increasing research on carbon labeling due to the urgent need to transform the global low-carbon economy. Carbon labels can successfully and effectively raise consumer understanding of the emissions of more environmentally hazardous foods, which can then help consumers choose more environmentally friendly food options [17]. Rating labels are more effective in guiding consumers to more sustainable products than positive and negative labels [22]. Edenbrandt & Lagerkvist [23] investigated consumer willingness to substitute high-emissions meat products with lower-emissions protein products, including blends of meat and vegetables. They found the traffic light carbon label affects choice behavior. In Canada and Argentina, younger consumers in the Americas exhibit differing perceptions of carbon labeling, with those who are more educated demonstrating a preference for the traffic light carbon label [24]. The greatest impact is observed when carbon footprint information is expressed in monetary units and color-coded in a manner analogous to the traffic signal system [25]. Traffic light labels are most effective in reducing carbon emissions, providing empirical support for the design of carbon labels [26,27]. However, there are also studies that carbon labels don’t succeed in attracting people’s attention when customers are unguided [28]. Carbon labeling has a positive but small impact on sustainable food choices [29]. As consumer trust in the label grows, carbon labeling boosts the perceived environmental sustainability of animal-based products but not that of plant-based products. Additionally, the “halo effect”of plant-based foods may lessen the effects of carbon labeling [30]. To change consumer behavior, consumers must develop trust in labels and the organizations behind them [31]. This trust and positive attitude remarkably influence the WTP and the positive effect of policy measures on consumers’WTP [32]. Interestingly, family and peer influences are more likely to affect consumers’ purchasing decisions than media exposure. This would appear to emphasize the importance of social dynamics in promoting environmentally friendly consumption [33,34]. By combining the results of these studies, we may conclude that carbon labeling may encourage sustainable consumption, but its effects may be affected by cultural, social, and psychological factors. In addition to enhancing our knowledge of the mechanisms underlying the effects of carbon labeling, integrating national and international research offers a solid scientific foundation for creating more focused environmental regulations and marketing strategies. Future research could further explore the best practices of carbon labeling in different cultural contexts and how to combine socio-cultural factors to improve its acceptance and effectiveness in the global market. 2.2. Theory of planned behavior The TPB model has its roots in psychological and sociological theories and is based on the Theory of Rational Behavior [35]. TPB, proposed by Ajzen [36], is the classic social psychological model for explaining people’s behavioral intentions and actions. It is a powerful and widely used tool for evaluating, modeling, and investigating people’s behavior about activities, products, or services [36]. According to the Theory of Planned Behavior, attitudes, subjective norms, and perceived behavioral control impact behavioral intentions. Attitudes toward the behavior refer to the degree to which a person has a favorable or unfavorable evaluation or assessment of the behavior under Nomenclature Abbreviations ATT Attitudes toward the behavior AR Ascription of responsibility AC Awareness of consequences CAP Carbon-labeled agricultural products CVM Contingent valuation method LA Low-carbon awareness NAM Norm Activation Model PBC Perceived behavioral control PN Personal norms PK Personal knowledge of CAP PB Purchase behavior PI Purchase intention RP Risk perception SEM Structural equation modeling SN Subjective norms TPB Theory of Planned Behavior WTP Willingness to pay Y. Xu et al. Sustainable Futures 8 (2024) 100363 2 discussion. Subjective norms represent the perceived social pressure to engage in a behavior or refrain from doing so. The third antecedent of intention is the degree of perceived behavioral control, which, as we’ve seen, refers to how easy or difficult the behavior is regarded to be to carry out. This antecedent is thought to reflect prior experience as well as anticipated impediments. The Theory of Planned Behavior suggests that behavioral performance is influenced by behavioral intentions, which are stronger if the individual has a more positive attitude, feels more pressure from external norms, or feels that he or she has more control over his or her behavior [36]. When a person can clearly understand the objective condition restrictions, such as resources and abilities, that he or she requires to conduct a given behavior, perceived behavioral control can also directly influence the occurrence of an individual’s behavior. The TPB model has been widely deployed to study sustainable food consumption behavior. For example, Vermeir and Verbeke [37] conducted a study with young people in Belgium. They found that personal attitudes, perceived social impacts, perceived consumer effects, and perceived availability were key factors influencing sustainable consumption intentions. In a separate study, Alam et al. [38] employed the extended TPB framework to ascertain the factors influencing sustainable food consumption behaviors in Malaysia. Their findings indicated that social norms, perceived value, and perceived consumer effects and attitudes significantly impacted consumption intentions. In contrast, perceived availability and perceived consumer effects and intentions significantly affected actual behavior. In China, Qi and Ploeger [39] revised and extended the TPB by incorporating face consciousness and group consistency into the model, replacing subjective norms to enhance the modelʼs applicability in Chinese consumers’intentions to purchase green food. They also included confidence and personal characteristics as model components to better reflect the current consumer environment and features. During the COVID-19 pandemic, Qi and Ploeger [40] extended the TPB to include the effects of ethical attitudes, health awareness, and COVID-19 to explain Chinese consumers’green food purchase intentions during the current and post-pandemic periods. According to the findings, Chinese consumers’ intentions to purchase green foods are better explained and predicted by the expanded TPB model than by the original TPB model. In conclusion, the TPB and its extended model provide a solid theoretical foundation for understanding consumers’sustainable food consumption behaviors under different cultural and environmental conditions. However, few studies have used the model to analyze the public’s willingness and behavior to purchase CAP and their links with each other. Based on these results, we propose the following hypotheses: H1. Attitudes toward the behavior (ATT) positively affect purchase intention (PI) in the purchase of CAP. H2. Subjective norms (SN) have a positive impact on PI for CAP. H3. Perceived behavioral control (PBC) has a favorable effect on PI in CAP purchasing. H4. PBC positively influences purchase behavior (PB) in CAP purchasing. H5. PI positively influences PB in the purchasing of CAP. 2.3. Norm activation model NAM is a behavioral theory that Schwartz [41] developed to explain and predict individual pro-social behavior. NAM contends that people’s pro-social/altruistic intentions and behaviors, such as volunteering their time and helping others, result from their norms, which are inspired by their awareness of problems and obligations [41]. Personal Norms (PN), Awareness of Consequences (AC), and Ascription of Responsibility (AR) are the three primary elements of NAM. The term PN denotes an individual’s self-expectations of behaviors in each situation, and AC is the propensity to be aware of the consequences of one’s actions on others [41]. The more likely individuals are to perceive situations in terms of the consequences of their actions on others, the more likely they are to attend to the values and norms associated with those interpersonal consequences, creating a sense of obligation to express those norms. AR is a person’s sense of whether they are responsible for the consequences of their actions [41]. NAM has become one of the most influential theories [42]. It has been used to study a variety of environmentally friendly behaviors, including drone food delivery services [43], litter reduction and litter picking behaviors [44], and environmentally friendly pest control adoption behaviors [45]. Steg and Groot [46] applied NAM to explain various pro-social and pro-environmental behavioral intentions. They discovered that the interaction between AC, AR, PN, and behavioral intentions is a chain-mediated model, meaning that AC activates PN through AR, leading to PI. According to some researchers, ATT, SN, and PN all indirectly impact consumers’intentions to recycle [47]. Song et al. [48] incorporated SN and environmental concern into an extended NAM, arguing that PN, established by consumers’AC and AR, significantly impacts their behavioral intention. In the context of the topic studied in this paper, consumers’AC and AR may increase their PN, thus leading to support for purchasing CAP. The study puts out the following hypothesis considering the analyses mentioned above. H6. AC has a favorable impact on AR in purchases of CAP. H7. AC has a positive influence on PN when purchasing CAP. H8. AR has a significant positive impact on PN when purchasing CAP. H9. When consumers buy CAP, PN has a positive effect on PI. H10. SN has a positive influence on PN when purchasing CAP. 2.4. Extension of TPB and NAM Applying and expanding theoretical frameworks are essential for a more profound comprehension of consumer behavior within sustainable food consumption. The TPB and NAM frameworks have been extensively utilized to predict and explain the environmental behavior of individuals [49]. Nevertheless, the extension and integration of these theories become especially important as social and environmental challenges become complex. Ajzen [36] states that TPB is an open theory to which additional variables can be added if they capture a significant portion of behavioral differences. Recent studies such as the TPB-NAM integration model have proven superior to the original TPB model when studying factors affecting Vietnamese farmers’intention toward organic agricultural production [50]. He and Sui [51] investigated the willingness of Chinese college students to consume green food by integrating the TPB with the NAM. They found that SN, ATT, and PN were the key factors influencing students’willingness to buy, with ATT having the strongest direct effect on willingness to buy. This work illustrates the function of NAM in promoting the formation of PN while also expanding the application of TPB. The empirical study by Vietnam Le and Nguyen [52] confirms the importance of ATT, social norms, and personal norms in organic food purchase intentions. The study offers new empirical evidence in support of NAM, emphasizing the significant impact of environmental awareness and knowledge about organic food on consumer purchase intentions through attitudes. Furthermore, Salmivaara et al. [53] provide novel insights into understanding sustainable food choices by delineating the distinction between descriptive and normative social norms. Their research reveals the importance of descriptive norms in actual and expected food choices, while SN fails to show the expected relevance. These findings challenge traditional theories and offer new strategies for influencing consumer behavior through descriptive norms. According to these researches, combining TPB and NAM offers a more thorough analytical framework and highlights the relative significance of various motivating factors in certain cultural and market contexts [54]. We can better understand consumer psychological and behavioral mechanisms when faced with Y. Xu et al. Sustainable Futures 8 (2024) 100363 3 sustainable food choices due to this theoretical extension and in-depth empirical study. This will enable us to develop more useful recommendations and tactics for encouraging sustainable consumption. Awareness is realizing or comprehending a situation or fact [55]. Resident acceptance and behavioral implementation are positively impacted by their general perception, satisfaction, and positive attitudes [56]. According to Xia et al. [57], raising consumer knowledge of low carbon emissions is advantageous for carbon reduction. Personal knowledge can be defined as the extent to which a consumer is aware of a specific product, service, or situation [58]. Consumers’level of personal knowledge regarding a product can influence their attitudes. For example, Li et al. [10] found that consumers with in-depth knowledge of low-carbon products are more likely to hold positive attitudes, which is consistent with the “ATT”component of the theoretical planned behavior (TPB) proposed by Ajzen [36]. In the context of sustainable consumption, Ding et al. [49] conducted a comprehensive review of the role of personal knowledge in influencing consumer choices towards environmentally friendly products. Studies by Xu and Lin [59] and Qi and Ploeger [39] have demonstrated that personal knowledge is a pivotal factor influencing consumer purchasing intentions, particularly in green and healthy products. In this study, “personal knowledge”describes the respondents’knowledge and understanding of CAP. This knowledge can potentially influence their attitudes, decision-making processes, and, ultimately, their purchasing behavior. According to research on consumer behavior and willingness to pay, risk perception and risk preference are significant factors of food acceptability, which has profound implications for consumer behavior and their willingness to pay [58,60,61]. Bhatti and Ur Rehman [62] examined the relationship between different factors, including perceived benefits, perceived risks, and online shopping behavior, with the mediating role of consumer purchase intention. The findings show that risk perception harms online shopping behavior and that these risks must be minimized to increase their sense of security. Based on the above theories, this paper integrates the original TPB and NAM models. It adds three external variables (i.e., personal knowledge, low-carbon awareness, and risk perception) to analyze the environmentally friendly behavior of the public’s behavioral willingness to purchase CAP. The following hypotheses have emerged, and the research framework is depicted in Fig. 1. H11. Consumers’ATT is positively impacted by personal knowledge (PK) of CAP. H12. Low-carbon awareness (LA) among consumers influences their AC to buy CAP. H13. The association between WTP and PB for CAP is moderated by risk perception (RP). 3. Material and methods 3.1. Contingent valuation method There are several approaches to elicit individuals’willingness to pay: contingent valuation, choice experiment, and experimental auction [63–65]. The contingent valuation method (CVM) was initially employed in 1958 to analyze non-market prices for recreational services in the Delaware River Basin region of the United States [66]. CVM has been utilized in various applications [59,67]. Typically, CVM employs questionnaires that let respondents explicitly express their preferences for specific goods in monetary terms [68]. The monetary value reported by respondents is generally expressed in terms of willingness to pay, i.e., the maximum monetary value that an individual would be prepared to pay for a hypothetical improvement program [68]. Alternatively, willingness to pay can be the minimum amount an individual will accept as compensation for the change [68]. Also, CVM has mostly been used to measure “preferences for goods or services for which a conventional market does not exist”[69]. The use of CVM is deemed appropriate, and hence, CVM has been chosen for this study. 3.2. Econometric model Structural Equation Modeling (SEM) is a statistical method for analyzing complex relationships among variables, combining factor and path analysis. Originating in the 1970s, SEM has become popular in economics [70]. It consists of two parts: the measured model, which links observed and latent variables, and the structural model, which connects latent variables. Observed variables are directly measurable, Fig. 1. Research framework. Y. Xu et al. Sustainable Futures 8 (2024) 100363 4 while latent variables cannot be measured directly and must be measured with the help of observed variables. The equation of the measurement model is as follows: X=Λxξ+δ(1) Y=Λy η + ε (2) The equation of the structural model is as follows: η =B η +Γξ+ζ(3) Table A1 displays the symbols and explanations for the three equations above. Eight latent variables influence consumers’purchase intention and purchase behavior, and a structural equation model is established to study this influence. 3.3. Questionnaire design Questionnaires are commonly utilized data collection survey tools within contingent valuation method studies [71]. The methodology measures the value consumers place on non-market products and enables researchers to determine attitudes and views. Consequently, this research uses a questionnaire as a data collection instrument. The questionnaire is split into four blocks. The first part is an introduction to the survey, which introduces the respondents to the topic of the questionnaire, the research organization, and the goal and relevance of the study. Step 1 asks respondents to complete individually a series of questions that captured their initial preferences and willingness to pay without entering any information. In Step 2, terms involved in the questionnaire are explained to improve the comprehensibility of the questionnaire. Based on the theoretical framework of TPB-NAM and combined the current market situation for agricultural products with past research findings ([72]; Oteng-Peprah et al. [73,74]), we have designed a total of 35 measurement questions. All questions were measured on a five-point Likert scale ranging from “completely disagree”to “completely agree”regarding low carbon awareness, ascription of responsibility, perceived behavioral control, perception of risk, and so on (Table 1). CVM is the main topic of the third section, which concentrates on how to evaluate consumers’willingness to pay and behavior. After examining and contrasting several WTP bootstrapping techniques, we choose to employ the payment card as a bootstrapping mechanism to prevent non-reflective bias in the questionnaire. We also create a survey with two questions to account for protest answers. It consists of sample selection questions and heuristic questions as follows. Regarding “carbon neutral”milk—currently offered for sale in China’s agricultural market—respondents are informed that labeling the goods would necessitate measuring greenhouse gas emissions during production and obtaining certification from a third-party organization, which would come at an extra expense to the manufacturer. Next, we asked respondents if they would pay more for CAP. If the answer is “yes”, the next leading question is to ask how much extra they/he would like to pay for each CAP. The poll indicates that there are currently carbon-labeled agricultural products on the market, such as “zero-carbon”vegetables and carbon-neutral milk. We choose the representative type of agricultural products among many varieties and talk with the participants in the interviews, considering factors like the types of agricultural products, the availability of each type in different regions, and the frequency of daily consumption by consumers. Using the price data already collected, we inform each respondent about the countryʼs average market price of several agricultural products. At the same time, based on the price, we give specific amounts in percentage increments that the respondents chose their largest willingness to pay. If a respondent answers “no”to the sample selection question, respondents who indicate they are unwilling to pay will be given a series of reasons to explore their intentions. The average willingness to pay in the payment card questionnaire can be determined for discrete variables using the Table 1 Definition and description of variables. Variable Code Item References Awareness of Consequences AC01 Purchasing high carbon dioxide emissions from agricultural products will cause serious pollution and environmental damage. [75] AC02 Purchasing carbon-labeled agricultural products can reduce environmental pollution.  AC03 Purchasing carbon-labeled agricultural products can ensure the quality and safety of agricultural products, which benefits us all.  Ascription of Responsibility AR01 As a consumer, I should bear some responsibility for reducing carbon dioxide emissions. [76] AR02 I feel responsible for environmental issues caused by not purchasing carbon-labeled agricultural products.  AR03 I would feel guilty if I didn’t buy agricultural items with carbon labels, contributing to increased carbon dioxide emissions.  AR04 I believe that every consumer bears some responsibility for the environmental and social problems caused by the production and consumption of agricultural products.  Attitudes toward the Behavior ATT01 I believe that purchasing carbonlabeled agricultural products is beneficial for the environment. [77] ATT02 I believe purchasing carbonlabeled agricultural products is a wise choice.  ATT03 Purchasing carbon-labeled agricultural products will make me feel physically and mentally pleasant.  Low-carbon awareness LA01 When shopping, I keep plastic shopping bags and reuse them. [78] LA02 When printing, I actively use both sides of each sheet of paper.  LA03 I often pay attention to articles or reports related to environmental issues.  Perceived Behavioral Control PBC01 The price of carbon-labeled produce significantly influences my decision to purchase it. [73] PBC02 I think that consumers find it difficult to purchase items with a carbon label because of the high prices.  PBC03 I am willing to buy carbon-labeled products when I have confidence in their environmental benefits.  Purchase Intention PI01 I am glad to purchase carbonlabeled agricultural products. [73,76] PI02 I am likely to purchase carbonlabeled agricultural products in the future.  PI03 I plan to buy more carbon-labeled agricultural products in the future.  PI04 I would recommend carbonlabeled agricultural products to my relatives and friends.  Personal knowledge of CAP PK01 I understand the concept of carbon-labeled agricultural products (such as “zero-carbon” vegetables, carbon-neutral milk, etc.). [79] (continued on next page) Y. Xu et al. Sustainable Futures 8 (2024) 100363 5 mathematical expectation formula. E(WTP) = ∑ n i=1 Pibi(4) where E(WTP) is the maximum average value of willingness to pay for each CAP; P i is the probability of respondents selecting each bid value; b i is the bid amount. The maximum average willingness of respondents to pay can be calculated by Eq. (4). In this paper, we combine the relevant literature [81] to measure consumers’actual purchase of CAP using two measures, i.e., whether they purchased CAP in the past year and the exact frequency of consumption. The last part is the socioeconomic characteristics of the respondents, including gender, age, current residence, education level, monthly income, etc. 3.4. Data collection To correct inaccurate or readily misconstrued content, a pilot poll was carried out with 60 randomly chosen participants in Oct. 2022. The formal survey was conducted from April to June 2023 in Shanghai, Nanjing, Wuhan, and Xi’an, China. The researchers are highly trained graduate and doctoral students. Target respondents are 18 years of age or older. A total of 641 respondents agreed to participate in the research. After excluding questionnaires with missing values or outliers based on the variables required for this research, 580 valid samples were recovered, with an effective recovery rate of 90.48 %. The demographic characteristics of the sample data are as follows. In terms of gender, there is an equal distribution of males and women (51.72 % and 48.28 %, respectively); in terms of age, the proportion of those aged 28 and below reaches 70.86 %; and in terms of education, 90.69 % of the samples have a high school diploma above. Table 2 presents a brief overview of demographic data. 4. Results 4.1. Willingness to pay for carbon-labeled agricultural products This section examines respondents’WTP for five distinct CAP. Descriptive statistics are then used to further explore disparities in respondents’WTP and possible reasons for consumers’reluctance to purchase CAP. Only 29.83 % of the 580 respondents are unwilling to pay the CAP premium, leaving 70.17 % eager to do so. This suggests that there is already some basis for moving forward with the carbon labeling system in terms of initial willingness to do so. As shown in Table 3, the results demonstrate that female respondents have slightly higher WTP on carbon-labeled milk and corn than male respondents. In contrast, male respondents have a higher premium WTP on carbon-labeled fruits, vegetables, and eggs than female respondents, indicating that WTP does not appear to be influenced by gender. Respondents under 40 have the highest WTP for most of CAP, and Table 1 (continued) Variable Code Item References PK02 I understand the quality characteristics of carbon-labeled agricultural products (such as “zero-carbon”vegetables, carbonneutral milk, etc.).  PK03 I am familiar with the costs associated with carbon-labeled agricultural products (such as “zero-carbon”vegetables, carbonneutral milk, etc.).  Personal Norms PN01 To safeguard the environment, I should limit the number of agricultural items I buy that release much carbon dioxide. [75] PN02 I feel morally obligated to contribute to reducing carbon dioxide emissions by purchasing carbon-labeled agricultural products.  PN03 I consider it a moral duty to society to purchase carbonlabeled agricultural products to reduce carbon dioxide emissions.  PN04 Everyone is responsible for considering the environmental impact when purchasing agricultural products.  Risk Perception PR01 I worry that I might be wasting money when buying carbonlabeled agricultural products. Martinho et al., 2022; [80] PR02 I am concerned that carbonlabeled agricultural products may not provide the necessary benefits.  PR03 I worry that carbon-labeled agricultural products may not perform as well as advertised.  PR04 Considering various factors, I think there are risks associated with purchasing carbon-labeled agricultural products.  Subjective Norms SN01 I prefer carbon-labeled agricultural products because my friends and family approve purchasing them. [77] SN02 My friends and family hope I will buy agricultural items with a carbon label, so I would like to do so.  SN03 I prefer to buy carbon-labeled agricultural products because the government encourages me.  SN04 I prefer to buy carbon-labeled agricultural products because of the environmentally friendly examples.  Table 2 Descriptive statistics of socio-economic characteristics of respondents. Characteristic Frequency Percent (%) Gender   Female 280 48.28 Male 300 51.72 Age   [18,28] 411 70.86 [29,39] 112 19.31 40 and above 57 9.83 Occupation   Vocational or blue-collar workers 30 5.17 Civil servant 50 8.62 Company staff 172 29.65 Student 264 45.52 Freelancer 35 6.04 Not working/Retired 29 5 Region   East 301 51.90 Middle 243 41.90 West 36 6.20 Monthly Income (CNY)   5000 or below 323 55.69 5001–10,000 107 18.45 10,001–20,000 101 17.41 20,000 above 49 8.45 Education   High School and below 54 9.31 Undergraduate 372 64.14 Master or above 154 26.55 Household Size   Two people or below 55 9.48 3–4 people 404 69.66 Over five people 121 20.86 Y. Xu et al. Sustainable Futures 8 (2024) 100363 6 respondents over 40 have the highest WTP regarding the premium for exclusively carbon-labeled eggs. This might be because different age groups have different opinions on the same issue. Eastern Chinese respondents have the greatest premium WTP, followed by central and western Chinese respondents. This may be because the eastern part of the country is economically developed, and its residents have relatively stronger financial resources and higher incomes. The household location is an important geographical factor for consumers, and it is related to the formation of their perceptions, beliefs, etc., which in turn are related to food consumption behavior. Meanwhile, respondents with monthly incomes of ¥20,000 or more have the highest premium WTP for 80 % of CAP mentioned in the survey. It is not difficult to understand that price is one of the main factors influencing consumers’decisions to buy products. According to the study, the family dimension appears to impact consumers’WTP, with respondents with smaller family sizes reporting higher WTP. A one-way analysis of variance (ANOVA) was performed to explore whether sociostatistical characteristics have a role in residents’WTP for CAP. Six independent variables, such as gender, age, and education, and one dependent variable, willingness to pay, were examined. The analysis’s findings reveal that three variables—respondents’place of residence, degree of education, and occupation—substantially impact their WTP for CAP. In contrast, the remaining variables failed to pass the test. This implies that an individual’s inclination to buy a CAP is somewhat influenced by their place of residence’s traits, level of education, and line of work. Among the respondents who are not willing to pay a premium for CAP, the most common answer (almost 20 %) is that they cannot pay for personal financial reasons but would be willing to pay if their income increased. This is followed by a reluctance to spend more money (16 %) and a belief that the government should shoulder more responsibility for reducing emissions than the respondents should (11 %). More than 9 % of participants feel that agricultural products had no significant environmental impact at any point along the supply chain, from production to consumption, and another 8 % states that the reason for their refusal is that they are unwilling to try new things. Some participants offer “others”justifications for declining to pay a premium for CAP, with a significant portion citing their belief that agricultural products are necessary for our daily lives, that they shouldn’t be excessively priced, and that we should look for ways to cut costs. Currently, there aren’t enough options on the market, and the carbon labeling market should be expanded. Some respondents also note that their decisions would be impacted if they didn’t know the authenticity and safety of CAP available on the market. 4.2. Measurement model testing As seen through the results in Table 4, all the factor loadings have values in the range of 0.631–0.913, which meets the criteria proposed by Hair et al. [82]. After analyzing the factor loadings, the reliability and validity of the questionnaire were analyzed. Cronbach’s alpha was used to confirm the reliability of the factors, and 0.7 was chosen as the standardized critical value, as indicated by Nunnally and Bernstein [83]. The overall Cronbach’s alpha for all latent variables was 0.960, and the estimates for the factors’Cronbachʼs alpha ranged from 0.722 to 0.928, all of which were above 0.7. This suggests that the respondents’internal consistency in assessing the observed variables reached an acceptable level, as did the reliability of the factors. Additionally, the study conducted KMO and Bartlett’s tests on the factors (Table 5). KMO estimations greater than 0.7 implied good adequacy of the sample collected [84]. The average extracted analysis of variance (AVE) ranged from 0.479 to 0.766, with nearly all of them being greater than 0.5, and the composite reliability (CR) for all latent variables ranged from 0.733 to 0.929, all of which were greater than 0.7. The study demonstrates good validity and reliability in examining the proposed hypotheses as all the indicators met the criteria [82]. 4.3. Structural model assessment Several key metrics for determining model fit are the NC value (CMIN/DF), root mean square of the approximation error (RMSMA), goodness-of-fit index (GFI), comparative fit index (CFI) and normative fit index (NFI). In general, a reasonable fit can be defined as having NC values between 1 and 5, RMSEA <0.08, and GFI, CFI, and NFI greater than 0.8 [85–87]. Table 6 shows that, following a few model modifications, all fitness indicators perform well, demonstrating that the model of the purchasing behavior of consumers of CAP and the collected data fit well. The study’s hypotheses are put to the test based on the structural modeling. Fig. 2 displays the specific visualization findings. The study Table 3 Willingness to pay for carbon-labeled agricultural products. Variables Category WTP-Milk (%) WTP-Corn (%) WTP-Banana (%) WTP-Tomato (%) WTP-Egg (%) Total 27.50 29.73 26.86 26.51 24.26 Gender Male 27.06 29.59 27.15 27.33 24.91 Female 27.96 29.88 26.55 25.63 23.56 Age [18,28] 27.83 29.59 26.74 26.22 24.01 [29,39] 27.22 31.36 27.55 27.60 24.55 40 and above 25.70 27.65 26.63 26.57 25.62 Occupation Vocational or blue-collar workers 31.45 30.46 30.19 31.07 29.12 Civil servant 24.78 27.04 39.52 26.80 23.92 Company staff 27.83 25.70 27.03 26.84 24.29 Student 26.18 27.98 25.65 24.82 22.67 Freelancer 32.46 35.59 32.70 32.68 31.43 Not working/Retired 32.09 33.38 29.53 27.21 25.50 Region East 28.85 31.53 27.74 27.24 25.01 Middle 26.38 28.17 26.39 26.24 23.76 West 23.71 25.20 22.71 22.16 21.34 Monthly Income (CNY) 5000 or below 26.57 28.02 26.00 25.13 23.21 5001–10,000 29.66 32.44 27.86 27.61 25.25 10,001–20,000 27.66 29.99 27.59 28.49 25.56 20,000 above 28.55 34.57 28.86 29.09 26.38 Education High School and below 28.70 31.69 26.73 28.57 28.45 Undergraduate 27.74 29.80 27.01 26.45 23.94 Master or above 26.48 28.87 26.57 25.92 23.56 Household Size Two people or below 30.48 33.18 28.61 28.44 26.06 3–4 people 27.28 29.55 27.05 26.50 24.50 Over five people 26.85 28.76 25.46 21.35 22.64 Y. Xu et al. Sustainable Futures 8 (2024) 100363 7 results indicate that ATT significantly and positively influences PI (β=0.219, P<0.001) within the theoretical framework of TPB, thus supporting H1. SN positively influences PI (β=0.404, P<0.001) and H2 is supported. H3 is confirmed, with a significant positive effect of PBC on PI (β=0.408, P<0.001). Among them, the standardized path coefficient of PBC is significantly higher than that of ATT and SN, indicating that the main factors influencing consumers’WTP about CAP are their control over relevant resources like money, time, knowledge, and skills, as well as their access to information about CAP. PBC has a significant positive effect on PB (β=0.261, P=0.002), and H4 is valid. This suggests that PBC can also directly influence the occurrence of a consumer’s behavior when he or she can perceive the objective constraints, such as resources and capabilities, that he or she needs to perform a certain behavior. The significance test refuses H5 with the result that PI harms PB (β=− 0.307, P<0.001). In the framework of NAM, consumers’AC of purchasing CAP positively affects their AR (β=0.683, P<0.001) and PN (β=0.435, P<0.001), supporting H6 and H7. Consumers’AR positively affects PN (β=0.389, P<0.001), implying that a consumer’s sense of responsibility for the adverse consequences of failing to purchase CAP activates his or her sense of moral obligation to purchase them, and H8 is validated. The positive effect of consumers’PN on PI passes the significance test (β=0.113, p=0.002); thus, H9 holds, meaning that people with greater ethical obligations have a higher CAP purchase intention. In the extended TPB-NAM theoretical framework, the hypothesized relationships of SN positively affecting PN (β=0.164, P<0.001), consumers’perceptions of CAP affecting their ATT (β=0.6, P<0.001), and consumers’low-carbon awareness affecting their AC (β=0.903, P< 0.001) all pass the significance test, and H10, H11, and H12 are confirmed, indicating that pressure from society positively affects their sense of moral obligation towards the act of purchasing CAP and that the level of consumers’knowledge of the background and existential significance of the proposed CAP significantly and positively affects their ATT. At the same time, the more environmentally conscious customers themselves are, the more they are aware of the harm that not buying CAP can do to the environment. 4.3.1. Moderating effects The SEM results indicate that purchase intention negatively influences purchase behavior. To understand the mechanism of RPʼs influence in the process of PI’s effect on consumers’CAP purchase, structural equation modeling with latent variable interaction terms was constructed using AMOS 26.0, and the moderating effect of the intention-behavior gap was analyzed based on the theory of likelihood. Table 4 Validity and reliability of the study. Variable Coding Factor loading Cronbach’s alpha Composite reliability Average variance extracted (AVE) Awareness of Consequences AC01 0.631 0.796 0.804 0.581 AC02 0.795    AC03 0.845    Ascription of Responsibility AR01 0.649 0.851 0.857 0.602 AR02 0.831    AR03 0.826    AR04 0.784    Attitudes toward the Behavior ATT01 0.855 0.890 0.892 0.735 ATT02 0.889    ATT03 0.826    Low-carbon awareness LA01 0.660 0.730 0.733 0.479 LA02 0.735    LA03 0.678    Perceived Behavioral Control PBC01 0.668 0.722 0.762 0.519 PBC02 0.66    PBC03 0.822    Purchase Intention PI01 0.859 0.928 0.929 0.766 PI02 0.87    PI03 0.891    PI04 0.88    Personal Knowledge of CAP PK01 0.795 0.874 0.8772 0.705 PK02 0.913    PK03 0.806    Personal Norms PN01 0.822 0.886 0.889 0.666 PN02 0.86    PN03 0.805    PN04 0.776    Risk Perception PR01 0.686 0.877 0.879 0.647 PR02 0.85    PR03 0.855    PR04 0.816    Subjective Norms SN01 0.907 0.905 0.909 0.715 SN02 0.912    SN03 0.82    SN04 0.73    Table 5 KMO and Bartlett’s test results of factor analysis. KMO and Bartlett’s Test Kaiser–Meyer–Olkin Measure of Sampling Adequacy. 0.958 Bartlett’s Test of Sphericity Approx. Chi-Square 16,148.911 Degree of freedom 630 Significance 0.000 Table 6 Results of model fitness test. Measure Threshold Estimate Interpretation CMIN/DF (NC) 1<NC <5 4.957 Acceptable RMSEA <0.1 0.083 Acceptable GFI >0.8 0.810 Acceptable CFI >0.8 0.878 Acceptable NFI >0.8 0.852 Acceptable PCFI >0.5 0.765 Acceptable PNFI >0.5 0.742 Acceptable Y. Xu et al. Sustainable Futures 8 (2024) 100363 8