The future of sustainable food consumption in China
Funding information
No funding was received to support this research or manuscript.
Abstract
Food production is one of the main contributors to greenhouse gas emissions and climate change. China, as a rapidly developing economy, contributes to an unsustainable food system as its consumption of animal products and meat has continued to grow in recent decades. Using the extended theory of planned behavior as the conceptual framework, this paper examines factors influencing consumers' intention to purchase sustainable food in China. To this end, a population-based face-to-face survey was conducted with 2422 respondents in five provinces spanning the north and south of China. The results showed that the traditional constructs of behavioral attitude, subjective norms, perceived behavioral control, and the additional construct of perceived quality are significant in inducing such intentions. This paper suggests that to enhance consumers' willingness to shift to sustainable food consumption, appropriate regulation and monitoring framework is needed to increase consumers' trust toward sustainable food. The government can also cooperate with the media, experts, and social media opinion leaders to ensure that messages on sustainable development are promoted in effective ways.
1 INTRODUCTION
Food sustainability is one of the priorities of the global sustainability agenda, with United Nations' Sustainable Development Goals (SDG) highlighting “food security” and “sustainable agriculture” (United Nations, 2019). However, the growing size of the population is constantly increasing the demand for food which makes the guarantee of the availability of food challenging. By 2050, global food production is expected to increase by as much as 110%, with most of this demand coming from developing countries (Garnett, 2013). At the same time, while plant-based diets are more beneficial to the environment as their production tends to consume fewer resources and produce lower greenhouse gas (GHG) emissions compared with animal-based products (Lacour et al., 2018), as countries become richer, people's diets tend to become more animal protein-based. The growing demand for beef is of particular concern because bovine meat is associated with higher GHG emissions and intensive use of agricultural resources compared with other meat types including chicken, fish, and pork (Poore & Nemecek, 2018). For example, 100 g of protein from beef herd produces about eight times more GHG emissions than 100 g of chicken or fish protein (Poore & Nemecek, 2018). This dietary shift and hence changes in food consumption behavior can potentially cause significant environmental damage (Herrero & Thornton, 2013; Poore & Nemecek, 2018).
China, as a rapidly developing economy and the most populous country in the world, without exception is encountering challenges to sustainable food production. According to the Food and Agriculture Organization Statistics (FAOSTAT, 2018), land use associated with agricultural activities in China accounts for 55% of the country area. While water resources available for agriculture purposes in China remain scarce (Yang et al., 2021), pollution problems have imposed a serious threat to agriculture production (Lu et al., 2015). Furthermore, China's continuous economic and population growth has induced an increasing trend of animal product consumption (He et al., 2018; Herrero & Thornton, 2013), with the rapid increase in meat consumption of particular concern as a recent study suggests that meat-intensive diets in China are closely related to higher ammonia emissions from agriculture (Liu et al., 2021).
To reduce the negative environmental impacts of agriculture, strengthened efforts to maintain a sustainable food system and a change in consumer habits such as a shift to a lower animal-based diet have been proposed (Garnett, 2013; Ivanova et al., 2020). Since the last decade, a new approach focusing on sustainable food consumption has emerged (Bilali et al., 2018; Lang & Barling, 2012). Sustainability, as defined by Solow (1991), is an obligation to conduct ourselves responsibly so that we leave to the future the option or the capacity to be as well off as we are. In relation to food consumption, this concept encompasses the responsibility to protect the environment and public health and to provide continued and reliable access to food (Capone et al., 2014). For example, a diet that is low in animal protein, high in vegetable protein, and within the limits of GHG emissions is considered sustainable (Carlsson-Kanyama, 1998). The concept of sustainable food also encompasses a variety of ideas such as economic viability, social justice principles (Gao et al., 2016), animal health and welfare, and labor rights (von Meyer-Höfer et al., 2015). To narrow the scope of the study, this study adopts a narrower definition of sustainable food as environmentally friendly food produced in a way that protects the environment and natural resources, such as air, water, and soil, and consumes less energy.
To devise viable solutions to food insecurity and unsustainability, two ideas proposed by Garnett (2013) are relevant to this paper. The first idea is demand constraint, which is the “conviction that excessive consumption, particularly of high-impact foods such as meat and dairy products, is a leading cause of [our] environmental crisis” (p. 33). This perspective focuses on the choice of consumers—if consumers can shift to sustainable food, environmental problems can be better addressed. The second idea is food system transformation, which supports the notion that food sustainability can only be realized “by changing the socio-economic governance of the food system” (Garnett, 2013, p. 34). To this end, various “hard” and “soft” measures, such as regulations, fiscal instruments, and education, are proposed to incentivize society to consume sustainable food. However, it remains uncertain whether these measures can motivate consumers to adjust their sustainable food consumption intention. This study aims to fill this identified gap in the literature.
This study adds to the understanding of sustainable food consumption behavior by investigating the factors that motivate individuals' willingness to consume sustainable food. The theoretical framework underpinning this research is the extended theory of planned behavior (TPB) (Ajzen, 1985; 1991; Wang et al., 2020), which captures the influence of attitudes, subjective norms (SNs), perceived behavior control, and perceived quality (PQ) in people's intentions toward performing a behavior. The research question about consumers' intention to shift to sustainable food consumption relates to Garnett's (2013) “demand constraint” in the sense that it seeks to understand consumers' motivation to consume more sustainable or less unsustainable food, and “food system transformation” (Garnett, 2013) that the effectiveness of government policies and business activities are studied. In addition, the study contributes to the literature by examining whether the variables in TPB affect sustainable food purchase intention (Liu et al., 2012; Wang et al., 2020; Zhao et al., 2014), and expanding the scope of the literature on food preferences by examining the extent to which Chinese consumers are willing to purchase sustainable food and determining the factors that lead to this willingness.
2 LITERATURE REVIEW AND THE CONCEPTUAL FRAMEWORK
The TPB is a well-developed social psychological model to explain the antecedents of behavior (Dowd & Burke, 2013). TPB posits that intentions are the determinants of behaviors, while intentions are predicted by attitude toward the behavior (BA), perceived behavioral control over the behavior (PBC), and SNs (Ajzen, 1985, 1991). Based on TPB, previous studies have examined the green consumption behaviors of Chinese consumers (Sheng et al., 2019; Wang et al., 2021), which refers to an ecological consumption model that seeks to minimize the negative impacts of individual behaviors on the ecological environment while meeting human needs (Carlson & Adriano, 1993; Pieters, 1991). Previous studies have suggested that an individual who has a positive attitude toward green consumption (Wang & Wang, 2013), aligns with social norms that support green consumerism (Sheng et al., 2019), and is conscious about its positive factors (Zhang & Li, 2017) are more likely to have a positive intention toward green consumption.
Although TPB is widely used in different contexts including consumers' willingness and decision to purchase certified food, organic food, or green products (Rezai et al., 2012; Vassallo et al., 2016), there is a lack of coverage of sustainable food. For example, Kalafatis et al. (1999) analyzed purchases of environmentally friendly products in the United Kingdom and Greece using TPB and found the framework to be more appropriate in well-established markets. Yadav and Pathak (2016) argued that people's environmental concerns could be used to predict the intention of young Indian consumers to purchase green products. In the Chinese context, Liu (2008) used the TPB framework to explain green behaviors; Chan and Lau (2002) found that both SNs and PBC influence Chinese consumers more than they influence their American counterparts in green purchases. As the essence of intention to purchase sustainable food is a kind of environmental behavior; therefore, we perceive TPB as an appropriate model to analyze consumers' sustainable food purchase decision.
2.1 TPB constructs and the hypotheses
To measure the latent variables in the model, previous studies have been taken into reference (Ajzen et al., 2004; Conner & Norman, 1996; Dowd & Burke, 2013; Nurse Rainbolt et al., 2012; Vermeir & Verbeke, 2008; Wang et al., 2018, 2020, 2021; Yadav & Pathak, 2016) to develop various manifest variables which could be directly observed or measured.
With respect to the three traditional constructs of TPB, first, behavioral attitude (BA) is associated with the positive or negative attitude of an individual toward a specific idea, object, or behavior. In this study, BA refers to one's attitude toward sustainable food, which is measured by three manifest variables related to environmental issues, namely concerns about the deforestation of agricultural land, concerns about land and water resources, and concerns about drying lakes and rivers. The more positive the consumers' attitude toward sustainable food, the higher the possibility of forming sustainable food consumption intention. The following hypothesis is thus proposed:
H1.Consumers' behavioral attitude has a positive impact on their willingness to pay for sustainable food.
Second, SN refers to the interpersonal or social pressure that an individual is exposed to when deciding whether to consume sustainable food. In this study, this variable is defined as the influence from others including the government, media, experts, international organizations, and social media opinion leaders. The higher the influence from others, the greater an individual's intention to proceed with sustainable food consumption behavior. The study thus proposes the following hypothesis:
H2.Subjective norm has a positive impact on consumers' willingness to pay for sustainable food.
Third, perceived behavioral control is defined as the extent to which individuals perceive themselves as being able to take an action. In this research, we interpret and measure the participants' perceived behavioral control with factors such as the cost of purchase, the level of convenience in purchasing, recognizability of the certifications, and conditions of purchase. The following hypothesis is proposed:
H3.Consumers' perceived behavioral control has a positive impact on their willingness to pay for sustainable food.
Previous research has also suggested that the addition of the construct of PQ strengthens the predictive power of TPB (Wang et al., 2020), and that the construct serves as an antecedent of satisfaction and behavioral intention (BI) and potentially leads to a purchase decision (Cronin et al., 2000). PQ refers to both the consumers' subjective perception and the inherent quality of a product or service in question, which appeals to the consumers and eventually influences their purchase decision (Jin & Gu, 2005; Kotler et al., 1990). In this study, PQ refers to consumers' judgment on the nutritional benefits and values of sustainable food. Notably, trust is a key factor affecting consumers' PQ, which is related to the degree to which the quality of the product is trustworthy or performs consistently well (Wang et al., 2020). In view of the close connection between PQ and trust, this study will also measure the trust level of consumers toward sustainable food certified by different certifying bodies including international organizations, the Chinese central government, and local universities and scientific institutions. The following hypothesis is further proposed:
H4.Consumers' perceived quality has a positive impact on their willingness to pay for sustainable food.
TPB also assumes that there may be interactions among exogenous variables. Wang et al. (2020) showed that the four constructs interact with each other in relation to consumers' willingness to buy certified pork. Separately, Sweeney and Soutar (2001) suggested that the perceived values of certain products lead to positive and proactive BA for consumer buying behavior. The role of SNs varies. In the field of green products, it is a mediator between consumers' attitude and perceived value (Li & Mattsson, 1995). Furthermore, consumers' purchase intention depends on their consumption culture and perceptions, with their consumption attitude being influenced by PBC (Dong et al., 2010). In summary, although interrelated, the constructs distinctly determine consumers' purchase intention and behavior.
Figure 1 summarizes the conceptual framework of the research. The next sections will proceed to test the hypotheses proposed in the study.
3 MATERIALS AND METHODS
3.1 Data collection
Data for this study were collected using a face-to-face survey in five Chinese cities between July and August 2019. These cities, namely Beijing, Nanchang, Xian, Taiyuan, and Shenyang, are located in five provinces spanning the north and south of China. These cities represent the different income levels of Chinese cities and China's economic disparities. In 2019, Taiyuan, Beijing, Xian, Nanchang, and Shenyang had GDPs per capita of CNY88,272 (approximately USD12,642), CNY140,211 (approximately USD20,081), CNY85,114 (approximately USD12,190), CNY95,825 (approximately USD13,724), and CNY75,766 (approximately USD10,851), respectively (National Bureau of Statistics of China, 2020). Using stratified sampling, one rural and one urban district were selected in each city, with the rural district located approximately 50 km from the urban center of the same city. To represent the population distribution of the country, the number of respondents from urban districts was double that from the rural districts.
The respondents were selected as follows to ensure a representative sample. First, we randomly chose the respondents based on their addresses. The respondents were located through local officials, such as village chiefs, officials in charge of the communities, and heads of local health authorities. The officials concerned had a list of villagers or residents that lived in their areas. Based on the address list, we randomly selected our respondents. Second, when there were inadequate respondents recruited by the first approach, we visited public areas frequently visited by residents and randomly invited them to participate.
Each interview lasted 20 to 25 min. Before the start of the questionnaire, the researchers verbally familiarized the respondents with the concepts of sustainable food and production, and the questionnaire provided a definition of sustainable food and/or environmentally friendly food choices. The researchers explained the definition of “sustainable food” as “food produced in a way that protects the environment and natural resources such as air, water and soil, and consumes less energy.” The survey consists of two parts: The first part asks about different dimensions related to the latent variables BA, SN, PBC, PQ, and BI, followed by some demographic questions in the second part. As shown in Table 1, BA, SN, PBC, PQ, and BI were measured by different items or manifest variables.
| Latent variable | Manifest variable | Mean | Standardized mean | Standard deviation |
|---|---|---|---|---|
| PQ | Q15_1 Ingredients will influence my food choice | 4 | 4 | 1.05 |
| Q15_6 Health benefits (e.g., lowers cholesterol and is good for bones) will influence my food choice | 3.78 | 3.78 | 1.03 | |
| Q15_7 Nutritional facts/value (e.g., protein and fat content) will influence my food choice | 3.79 | 3.79 | 1.06 | |
| Q15_8 Nutritional benefits (e.g., low fat, reduced salt, and low calories) will influence my food choice | 3.75 | 3.75 | 1.04 | |
| Q15_10 The brand will influence my food choice | 3.77 | 3.77 | 1.02 | |
| BA | Q12_1 I am concerned about deforestation for agricultural land | 3.1 | 3.1 | 1.16 |
| Q12_4 I am concerned about the damage and use of land and water resources in agricultural food production | 3.16 | 3.16 | 1.14 | |
| Q12_6 I am concerned about drying lakes and rivers due to agricultural food production | 3.15 | 3.15 | 1.18 | |
| SNs | Q28_3 Government promotions can influence my decision to purchase certified food | 3 | 3 | 1.02 |
| Q28_4 Information provided in the media can influence my decision to purchase certified food | 2.89 | 2.89 | 1.01 | |
| Q28_5 Information provided by experts and academic organizations can influence my decision to purchase certified food | 2.97 | 2.97 | 1.03 | |
| Q28_7 Information provided by social media opinion leaders on social media can influence my decision to purchase certified food | 2.86 | 2.86 | 1.01 | |
| PBC | Q1 My home is far from the shops/stores where I buy vegetables and fruits | 4.5 | 3.75 | 1.62 |
| Q2 My home is far from the shops/stores where I buy meat | 4.41 | 3.68 | 1.59 | |
| Q3 My home is far from the shops/stores where I buy fast-food/hot meals | 4.27 | 3.56 | 1.71 | |
| BI | Q21 I am willing to pay more for environmentally friendly food, even if both environmentally friendly and ordinary food products come from the same origin, with similar packages and appearance | 1.67 | 1.39 | 1.42 |
| Q24_1 I am willing to pay more for environmentally friendly food certified by the government | 2.24 | 1.87 | 1.36 | |
| Q24_2 I am willing to pay more for environmentally friendly food certified by international organizations | 2.18 | 1.82 | 1.31 | |
| Q24_3 I am willing to pay more for environmentally friendly food certified by food safety accreditation organizations (e.g., research institutes and universities) | 2.2 | 1.83 | 1.34 |
Of the collected 2492 questionnaires, we identified 70 invalid questionnaires, which were discarded, and 2422 valid questionnaires, which were included in the analysis of this study.
3.2 Instrumentation
The questionnaire was designed to reflect the theoretical basis of this study. Based on previous studies (Ajzen et al., 2004; Conner & Norman, 1996; Dowd & Burke, 2013; Nurse Rainbolt et al., 2012; Vermeir & Verbeke, 2008; Wang et al., 2018, 2021, 2020; Yadav & Pathak, 2016), we have identified various manifest variables as indicators to measure the five latent variables in the model through direct measurement or observation. Each manifest variable was measured using a 5-point Likert scale (e.g., 1: strongly disagree; 2: moderately disagree; 3: neutral; 4: moderately agree; 5: strongly agree).
For BI which indicates a respondent's intention to pay, it was measured by whether a respondent is willing to pay more for sustainable food certified by different certifying bodies. Behavioral attitude was measured by the positive or negative attitude of consumers toward agricultural food production, toward the effects of agriculture on the environment, and toward the use of pesticides, chemicals, and carbon emissions in agriculture. SNs were measured by questions concerning one's social circles, covering information provided by, and influence of the government, media, experts, social media opinion leaders, sales, and international organizations. Perceived behavioral control was measured by respondents' convenience to purchase food including the distance between one's home and shops selling different food items, and the costs of food. Manifest variables utilized to measure the latent variable of PQ include nutritional value, nutritional benefits, being healthy, brand-attitude, ingredients, expiry date, and origin of food.
Table 1 summarizes the mean, standardized mean, and standard deviation of the manifest variables included in the model. Some manifest variables that we thought might measure the four constructs are excluded from Table 1 and the model because of the result of the exploratory factor analysis (EFA).
4 RESULTS
4.1 Descriptive statistics
Table 2 shows the respondents' demographic information, namely gender, age, living area, education level, employment status, family size, family income, and the ages of any young family members. The sample of this study largely corresponded to the Chinese population, with the exceptions of gender ratio and employment status. In the sample, 35.2% of the respondents were men and 64.8% were women, resulting in a gender ratio of 54.3 men to 100 women. However, the gender ratio in China was about 105 men to 100 women in 2017 (Textor, 2020). The discrepancy could be explained by the fact that women are more likely to take responsibility for housework in China (Lu et al., 2000; Zhang, 2017), who are categorized as unemployed or retired in the survey. For the same reason, unemployed women, who had a higher chance to stay at home or go for shopping, were more likely to be selected for responding to our face-to-face survey. After careful consideration, we perceive this discrepancy to be acceptable, as the responsibility for food purchasing in the household usually falls on women or the retired, whose perspectives, intentions, and behaviors of food purchase are highly valued by our study compared with the remaining groups. In general, the sample was representative of Chinese consumers across the country.
| Variable | Category | Frequency | Percentage (%) |
|---|---|---|---|
| Gender | Male | 806 | 35.2 |
| Female | 1485 | 64.8 | |
| Age | Below 18 years old | 4 | 0.2 |
| 18–30 years | 641 | 26.4 | |
| 31–45 years | 650 | 26.8 | |
| 46–60 years | 666 | 27.5 | |
| 60+ | 465 | 19.2 | |
| Residence | Urban area | 1580 | 69.1 |
| Rural area | 705 | 30.9 | |
| Education level | Middle school or below | 882 | 38.5 |
| High school (vocational senior high schools included) | 453 | 19.8 | |
| College diploma | 420 | 18.3 | |
| Undergraduates | 433 | 18.9 | |
| Postgraduate or above | 103 | 4.5 | |
| Employment status | Full-time employed | 1060 | 46.3 |
| Part-time employed | 92 | 4 | |
| Self-employed | 61 | 2.7 | |
| Employer | 32 | 1.4 | |
| Unemployed/retired | 1042 | 45.6 | |
| Family size | 1 member | 51 | 2.2 |
| 2 members | 349 | 15.2 | |
| 3 members | 690 | 30.1 | |
| 4 members | 534 | 23.3 | |
| 5 members or above | 667 | 29.1 | |
| Children under 18 | < 6 years | 623 | 27.2 |
| 6 >=< 12 years | 452 | 19.7 | |
| 13 >=< 18 years | 373 | 16.3 | |
| No children < 18 years | 1219 | 53.2 | |
| Family income (RMB) | 4000 or below | 351 | 15.3 |
| 4001–7000 | 354 | 15.5 | |
| 7001–10,000 | 314 | 13.7 | |
| 10,001–14,000 | 287 | 12.5 | |
| 14,001–20,000 | 138 | 6 | |
| 20,001–30,000 | 76 | 3.3 | |
| 30,001–50,000 | 38 | 1.7 | |
| 50,001–80,000 | 20 | 0.9 | |
| 80,001–100,000 | 15 | 0.7 | |
| Above 100,000 | 14 | 0.6 | |
| Prefer not to answer | 680 | 29.7 |
- Note: 1 RMB = 0.157 USD.
Trust, as a key factor influencing consumers' PQ, affects consumers' intention to purchase sustainable food. Once trust does not exist between consumers and certification bodies, respondents will make different purchase choices. Table 3 indicates that the respondents showed a fairly high level of trust in food certification bodies. The three types of bodies presented to the respondents were international organizations, China's central government, and local universities and scientific institutions. The percentages of respondents who found them trustworthy or highly trustworthy were similar in each category, namely 54%, 56.4%, and 51.2%, respectively.
| International organizations (%) | China's central government (%) | Local universities and scientific institutions (%) | |
|---|---|---|---|
| Highly untrustworthy | 2.5 | 2.2 | 2.5 |
| Untrustworthy | 7.7 | 7.7 | 8.4 |
| No opinion | 28.4 | 26.2 | 30.5 |
| Trustworthy | 33.7 | 32.9 | 31.4 |
| Highly trustworthy | 20.3 | 23.5 | 19.8 |
The results of this study are in line with the current trend (Rupprecht et al., 2020; Wang et al., 2020). Indeed, although the difference in trust between the different agencies is small, Chinese consumers regard China's central government as the most trustworthy food certification body; in other words, this agency has received an increasing level of trust from consumers in recent years (Wang et al., 2020). This trust distribution is also consistent with the findings of other studies. For example, Zhang et al. (2016) found that the government is the most trustworthy institution in terms of food safety; Ortega et al. (2011) suggested that Chinese consumers are more willing to pay for food certified by the government than for food certified by other bodies, and Bai et al. (2013) showed that consumers value certificates issued by the government for traceability more than those issued by third parties (Bai et al., 2013).
However, in comparison to consumers in other countries, Chinese consumers in general have a lower level of trust in food certification. For instance, in Greece, Krystallis and Chryssohoidis (2005) found a 74.2% consumer trust level in institutions certifying food products, and in Malaysia, Nawi and Nasir (2014) found that 86% of consumers trust the quality of government-certified food for its cleanliness and safety.
4.2 Exploratory factor analysis
To examine the structure and internal reliability of the measures, we first conducted EFA. This consists of organizing the data to obtain a smaller set of summary variables. To achieve this, we used varimax rotation. The results are presented in Table 4, which shows that the proposed variables corresponded statistically to their associated manifest variables. This confirmed that our grouping of manifest variables in the five constructs under study was appropriate.
| Factor loading | |||||
|---|---|---|---|---|---|
| Factor 1 | Factor 2 | Factor 3 | Factor 4 | Factor 5 | |
| Q12_1 | 0.168 | 0.048 | 0.072 | −0.041 | 0.8 |
| Q12_4 | 0.133 | 0.091 | 0.078 | −0.021 | 0.873 |
| Q12_6 | 0.152 | 0.097 | 0.097 | −0.021 | 0.871 |
| Q15_1 | 0.712 | 0.013 | 0.056 | −0.037 | 0.149 |
| Q15_6 | 0.847 | 0.084 | 0.062 | 0.008 | 0.121 |
| Q15_7 | 0.889 | 0.083 | 0.1 | −0.018 | 0.111 |
| Q15_8 | 0.879 | 0.076 | 0.07 | −0.024 | 0.133 |
| Q15_10 | 0.657 | 0.102 | 0.086 | 0.013 | 0.023 |
| Q21 | 0.149 | 0.736 | 0.06 | 0.022 | 0.077 |
| Q24_1 | 0.052 | 0.938 | 0.056 | 0.063 | 0.065 |
| Q24_2 | 0.068 | 0.931 | 0.087 | 0.07 | 0.064 |
| Q24_3 | 0.07 | 0.933 | 0.074 | 0.059 | 0.053 |
| Q28_3 | 0.092 | 0.045 | 0.842 | −0.021 | 0.066 |
| Q28_4 | 0.083 | 0.051 | 0.882 | −0.032 | 0.073 |
| Q28_5 | 0.107 | 0.082 | 0.886 | −0.007 | 0.069 |
| Q28_7 | 0.076 | 0.095 | 0.882 | −0.029 | 0.065 |
| q_1 | −0.025 | 0.071 | −0.035 | 0.934 | −0.025 |
| q_2 | 0.001 | 0.063 | −0.015 | 0.927 | −0.032 |
| q_3 | −0.024 | 0.055 | −0.031 | 0.905 | −0.028 |
- Note: Extraction method: principal component analysis. Rotation method: Varimax with Kaiser normalization. Rotation converged in five iterations.
4.3 Reliability and validity tests
To test the applicability of the variables, we performed reliability and validity tests. As shown in Table 5, Cronbach's α values for the five latent variables exceeded 0.83, falling in the range of acceptable reliability values and indicating good internal consistency between the variables.
| Latent variable | Manifest variable | Standardized factor loading | Cronbach's α | CR | AVE |
|---|---|---|---|---|---|
| PQ | Q15_1 | 0.53 | 0.873 | 0.89911 | 0.643793 |
| Q15_6 | 0.641 | ||||
| Q15_7 | 0.681 | ||||
| Q15_8 | 0.665 | ||||
| Q15_10 | 0.506 | ||||
| BA | Q12_1 | 0.631 | 0.839 | 0.88521 | 0.720257 |
| Q12_4 | 0.707 | ||||
| Q12_6 | 0.696 | ||||
| SNs | Q28_3 | 0.602 | 0.906 | 0.92771 | 0.762452 |
| Q28_4 | 0.639 | ||||
| Q28_5 | 0.63 | ||||
| Q28_7 | 0.648 | ||||
| PBC | q_1 | 0.671 | 0.914 | 0.94453 | 0.850236 |
| q_2 | 0.652 | ||||
| q_3 | 0.64 | ||||
| BI | Q21 | 0.512 | 0.917 | 0.93703 | 0.789698 |
| Q24_1 | 0.63 | ||||
| Q24_2 | 0.618 | ||||
| Q24_3 | 0.617 | ||||
| Overall | 0.809 |
Furthermore, Table 5 shows that the standardized factor loadings of most of the manifest variables were above 0.5. Table 5 also shows that all of the composite reliability (CR) values were above 0.8 and that the average variance extracted (AVE) values exceeded 0.6. These results showed the strong association of the manifest variables with the latent variables. The latent variables therefore had good convergent validity.
To further confirm construct validity, we performed discriminant validity tests. Table 6 shows the square root of the AVE of the latent variables and reveals that the values on the diagonal exceeded the maximum coefficients in the associated rows and columns, indicating good discriminant validity.
| Variable | PQ | BA | SNs | PBC | BI |
|---|---|---|---|---|---|
| PQ | 0.802 | ||||
| BA | 0.321** | 0.849 | |||
| SNs | 0.210** | 0.198** | 0.873 | ||
| PBC | −0.032 | −0.061** | −0.053* | 0.922 | |
| BI | 0.196** | 0.179** | 0.166** | 0.120** | 0.889 |
- ** The correlation is significant at the 0.01 level (two-tailed).
- * The correlation is significant at the 0.05 level (two-tailed).
4.4 Confirmatory factor analysis (CFA)
After confirming the EFA results, we performed CFA to measure the adequacy of the manifest variables in explaining the constructs and to ascertain whether the number of factors and the factor loadings of the manifest variables were in line with our expectations. Figure 2 illustrates the measurement model of the determinants of consumers' willingness to pay more for sustainable food.
4.5 Model fit measurement
We further confirmed the model fit by calculating the fit indices. Table 7 shows the results, indicating a well-fitting model (CMIN/DF = 2.643; GIF = 0.983; RMSEA = 0.027; NFI = 0.987; CFI = 0.992; IFI = 0.992; RFI = 0.985).
| Index | Recommended value | Observed value | Acceptance |
|---|---|---|---|
| Chi-square likelihood ratio (CMIN) | 375.352 | ||
| Degree of freedom (DF) | 142 | ||
| CMIN/DF | <3 | 2.643 | Good |
| Goodness of fit (GFI) | >0.9 | 0.983 | Good |
| (Adjusted) goodness of fit (AGFI) | >0.9 | 0.977 | Good |
| Root mean square residual (RMR) | <0.05 | 0.036 | Good |
| Root mean square error of approximation (RMSEA) | <0.05 | 0.027 | Good |
| Normed fit index (NFI) | >0.9 | 0.987 | Good |
| Relative fit index (RFI) | >0.9 | 0.985 | Good |
| Incremental fit index (IFI) | >0.9 | 0.992 | Good |
| Comparative fit index (CFI) | >0.9 | 0.992 | Good |
| Parsimony goodness-of-fit index (PGFI) | >0.5 | 0.735 | Good |
| Parsimonious normed fit index (PNFI) | >0.5 | 0.820 | Good |
| Parsimony comparative fit index (PCFI) | >0.5 | 0.824 | Good |
| Non-centrality parameter (NCP) | a) | 233.352 | |
| Expected cross-validation index (ECVI) | a) | 0.206 | |
| Akaike information criterion (AIC) | a) | 471.352 | |
| Consistent Akaike information criterion (CAIC) | a) | 794.611 | |
| Browne-Cudeck Criterion (BCC) | a) | 472.199 | |
| Bayesian Information Criterion (BIC) | a) | 746.611 |
- Note: a)Smaller value implies better model fit.
4.6 Hypotheses testing
Table 8 reports the results of structural equation modeling. The standardized path coefficients from PQ, SNs, and BA to BI were positive, while that of PBC was negative. They were all statistically significant at the 0.001 level.
| Path | Standardized path coefficient | Standard error | CR | p value |
|---|---|---|---|---|
| BI←PQ | 0.115 | 0.031 | 4.927 | *** |
| BI←PBC | −0.155 | 0.02 | −7.209 | *** |
| BI←SN | 0.129 | 0.032 | 5.733 | *** |
| BI←BA | 0.133 | 0.039 | 5.471 | *** |
| Q15_1←PQ | 0.621 | 0.02 | 33.874 | *** |
| Q15_6←PQ | 0.828 | 0.016 | 55.21 | *** |
| Q15_8←PQ | 0.886 | 0.015 | 62.56 | *** |
| Q15_10←PQ | 0.54 | 0.02 | 27.961 | *** |
| Q15_7←PQ | 0.92 | |||
| q_3←PBC | 0.838 | 0.019 | 54.15 | *** |
| q_2←PBC | 0.897 | |||
| q_1←PBC | 0.924 | 0.016 | 62.388 | *** |
| Q28_3←SN | 0.781 | 0.02 | 45.029 | *** |
| Q28_5←SN | 0.871 | |||
| Q28_4←SN | 0.849 | 0.019 | 50.861 | *** |
| Q28_7←SN | 0.863 | 0.018 | 53.316 | *** |
| Q12_6←BA | 0.868 | 0.038 | 33.98 | *** |
| Q12_4←BA | 0.843 | 0.035 | 34.073 | *** |
| Q12_1←BA | 0.689 | |||
| Q24_3←BI | 0.937 | 0.012 | 85.341 | *** |
| Q24_1←BI | 0.937 | |||
| Q24_2←BI | 0.939 | 0.011 | 85.802 | *** |
| Q21↔BI | 0.638 | 0.019 | 36.916 | *** |
| PQ↔PBC | 0.05 | 0.031 | 1.606 | 0.108 |
| PBC↔SN | 0.072 | 0.029 | 2.458 | 0.014 |
| SN↔BA | 0.168 | 0.019 | 8.88 | *** |
| PQ↔SN | 0.209 | 0.022 | 9.545 | *** |
| PBC↔BA | 0.071 | 0.027 | 2.652 | 0.008 |
| PQ↔BA | 0.282 | 0.021 | 13.176 | *** |
- *** p < 0.001.
Among the four constructs, the results showed that BA had the most significant and positive effect on consumers' willingness to pay more for sustainable food, with its standardized path coefficient as 0.133. H1 is therefore supported. The three manifest variables related to BA, namely concerns about deforestation for agricultural land, damage caused by the use of land and water resources in agricultural food production, and drying lakes and rivers caused by agricultural food production, contributed positively to the latent variable.
The next contributing factor to willingness to pay more was SNs, with a standardized path coefficient of 0.129. H2 is therefore supported. The four manifest variables concerned were positive. This finding indicates that government promotions and information provided in the media by experts, academics, and social media opinion leaders influence Chinese consumers' conception of SNs and therefore increases their readiness to pay more for environmentally friendly food.
The third most influential factor was PQ, with a standardized path coefficient of 0.115. H4 is therefore supported. The contributing manifest variables were ingredients, health benefits, nutritional facts or values, and nutritional benefits, with path coefficients of 0.621, 0.828, 0.886, and 0.54, respectively. Of the four manifest variables, ingredients had the least impact on the latent variable.
The final latent variable was PBC. Its standardized path coefficient was −0.155, indicating that the variable was negatively associated with consumers' intention to purchase sustainable food. Referring to its manifest variables that measured the distance between the respondents' homes and markets for various types of food, the result showed that increased difficulty in accessing food decreased their willingness to purchase sustainable food. In other words, increased convenience in assessing food would increase consumers' intention to purchase sustainable food. Therefore, H3 is supported.
The interactions among PQ, SNs, and BA were positively correlated at a statistically significant level of 0.001. The strongest association was between PQ and BA. The standardized path coefficient was 0.282. The relationship could be two-way. On the one hand, consumers may perceive a clean environment as an enabling condition for growing quality food as they become more concerned about the health benefits and nutritional value (i.e., PQ) of their food. On the other hand, consumers may become more observant of the ingredients used in their food as their concern for the environment increases.
In summary, the four hypotheses proposed in the study were all supported, indicating that the four constructs proposed for use in the TPB model were validated.
5 DISCUSSION
The above results confirm that Chinese consumers are prone to purchase sustainable food. Compared with Vermeir and Verbeke's (2008) work with the context of Belgium, our results show that the proportion of people willing to purchase sustainable food in China is higher than Belgium. While Chinese consumers pay more attention to the quality of sustainable food, Belgian consumers focus on the availability of sustainable food. Our findings echo with Vermeir and Verbeke's (2008) study that people's willingness to purchase sustainable food is affected by those around them, and an individual's attitude also significantly affects their intentions.
Prior research on purchasing intention of consumers toward certified food, green food and organic food (e.g., Thøgersen & Zhou, 2012; Wang et al., 2020; Zhou et al., 2013) mainly considers consumers' personal health concerns; however, in this study, the focus on sustainable food considers more about pro-environment values and behaviors of consumers. While the former is more related to individual benefits, the latter inclines to prioritize collective interests such as the well-being of the future generations and the environment over individual interests. The results also showed that the respondents had different levels of willingness to purchase sustainable food depending on the certification agency involved. Unlike previous findings (Wang et al., 2020; Xiong & Yang, 2011), this study suggests that consumers trust government certification the most and are thus willing to pay more for government-certified sustainable food. The level of trust among the respondents was even higher than that in international organizations, shown to enjoy high trust among Chinese consumers in previous studies (Wang et al., 2020; Wu et al., 2014; Xiong & Yang, 2011; Yin et al., 2017).
Among the four constructs in TPB considered in this study, BA contributed the most to the respondents' willingness to purchase sustainable food. In this study, attitude referred to their concerns about the impact of food production on the environment, in the form of deforestation, damage to land and water resources, and drying lakes and rivers. In other words, consumers with high BA are highly concerned about the impact of their food consumption on the environment, rather than simply considering personal benefits such as food safety. They limit their demand for non-sustainable food (Garnett, 2013) and replace their needs with sustainable alternatives. In this sense, this finding is consistent with studies suggesting that these consumers are more likely to be socially concerned rather than individually concerned (Xu et al., 2012). This willingness to spend extra for sustainable food may be a result of “green consumerism” (Sachdeva et al., 2015). This is similar to buying fair trade products from which consumers derive no extra benefits in terms of food quality or which may instead incur additional monetary costs. Despite this asymmetric relationship, consumers are still satisfied with such “expensive” purchases.
In this study, PQ was also found to have a positive impact on food choice. This factor was measured by the respondents' perceptions of the health benefits and nutritive value of food products. This finding aligns with studies showing that when it comes to food choice, food safety is often one of the main considerations of Chinese consumers (Fabinyi, 2016; Xu & Wu, 2010; Tait et al., 2016). However, healthy food means more than just being safe to consume. Other than causing no harm, consuming healthy food may improve health. This concern about healthy food could also translate into BA toward sustainable food, as consumers may consider that the long-term supply of healthy products can be achieved through sustainable food consumption. Our finding on the interaction between the two constructs supports such a correlation. Consumers concerned about food safety and food quality may be more easily convinced to pay more for sustainable food (Fabinyi, 2016; Gao et al., 2016).
As predicted by TPB, PBC influenced the respondents' willingness to pay. This variable measured the convenience with which the respondents could purchase different types of food. Sustainable food in China is new; it is therefore not widely available. This type of food may be more easily found in larger shops and stores that sell a variety of products, such as supermarkets. People who have less access to different types of food have less intention of consuming sustainable food. This finding is similar to that of Thøgersen and Zhou (2012) who found that the low availability of organic food hinders consumers' purchase intention.
6 CONCLUSION AND POLICY IMPLICATIONS
Based on TPB, this study explores the willingness of consumers to purchase sustainable food in China. While previous studies on food purchase intention mainly focus on consumers' individual health concerns; this study, examining sustainable food, is novel in the sense that it has examined the extent to which consumers are willing to spend more for protecting the environment. The results reflect the current spending power of consumers and the monetary sacrifice that they are willing to make for the common good (i.e., sustainable development in this study). In addition, this study offers a countrywide perspective on Chinese consumers' willingness to pay more for sustainable food. Unlike studies focusing on one specific city or region (e.g., Liu et al., 2012; Thøgersen & Zhou, 2012; Wang et al., 2020), this study's sample was drawn from five provinces spanning the north and south of the country with different income levels, covering both rural and urban areas. The sample is also in general more representative than other studies which relied on online surveys for data collection (e.g., Sheng et al., 2019; Zhu et al., 2021, 2022), which is inadequate in capturing the views of the elderly and people living in rural areas with unavailable or limited internet access.
Given its external validity, the findings of the study could be generalized to a wider population in China as well as countries or regions with a similar context in the world. To a certain extent, we predict that countries with higher education and economic levels than that of China would have a similar or even higher intention to shift to sustainable food consumption, in view of the positive linkage between income growth, education level, and the demand for environmental quality and sustainability (Ready et al., 2002).
The findings of this study have certain policy implications for the development of sustainable food. First, our findings show that PQ and SNs have a positive effect on inducing sustainable behavior. If the parties that disseminate information about sustainable food are considered trustworthy, then consumers' exposure to information is more likely to generate social pressure leading to purchase intention. In this sense, information disseminated by the government, media, experts, and social media opinion leaders could create social pressure and ultimately affect the intention and behavior of consumers. As suggested by previous studies (Moragues et al., 2013; Toussaint et al., 2021), the influence and pressure of these parties on consumers may also mean that the government can cooperate with them to ensure that messages on sustainable development are promoted in various ways. A relatively lower level of trust in food certification in China is also well worth exploring because some environmentally conscious consumers might be reluctant to spend more if they are uncertain of the food packaging claims (Chan, 2000). Consumers' trust in sustainable food is a long-term accumulation process. Regulation and its monitoring framework may also affect consumers' trust toward sustainable food.
In the future, studies may further explore the most effective strategies to communicate the concept of sustainable food effectively with consumers and hence promote their intention to consume sustainable food.
ACKNOWLEDGMENTS
This work was supported by funding (MD18518) from the CUHK–Exeter Joint Centre for Environmental Sustainability and Resilience (ENSURE) awarded to the whole team, and RGC Area of Excellence Scheme (AoE/M-403/16) awarded to H.-M.L. We would like to thank the anonymous interviewees, reviewers, and editors for sharing their insights with us. We also thank Mr. Jackey Wai Yu Cheung and Mr Major Pak Wan Pau for their research and editorial efforts. All errors remain our own.
CONFLICT OF INTEREST
The authors have stated explicitly that there are no conflicts of interest in connection with this article.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
