Tom tat luan an: Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.

Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.Phân tích sở thích, thái độ của người sản xuất và người tiêu dùng đối với phát triển nuôi trồng thủy sản tốt trong nuôi tôm tại Việt Nam.

MINISTRY OF EDUCATION AND TRAINING The research was conducted and completed at University of: CITY UNIVERSITY OF ECONOMICS HO CHI MINH CHAPTER 1: OVERVIEW -University of Economics Ho Chi Minh city 1.1 Research background 1.1.1 Global aquaculture contexts Aquaculture has contributed to half of the global fishery production, reducing the pressure to exploit natural fisheries resources (Pradeepkiran, 2019) and becoming an important economic sector in many countries (FAO, 2020) However, the rapid development and lack of planning have caused negative impacts on the environment and society (FAO, NGOC PHONG 2020) To reduce advisors: adverse TRUONG effects, many countries have promoted the adoption of Good Academic Aquaculture Practices (GAqPs) (Sampantamit et al., 2020) However, few farms apply GAqPs, and only about 14.2% of global seafood is certified with GAqPs (Potts et al., 2016) Associate Prof PhD Vo Tat Thang 1.1.2 Vietnamese shrimp aquaculture contexts Vietnam currently has about 740 thousand hectares of shrimp farming land, and the Prof PhD Nguyen Hoai output is 950 thousand tons (VASEP, 2020a).Trong Shrimp farming will become an important economic sector, with export value reaching US$3.9 billion by 2021 (VASEP, 2021) However, environmental pollution and food insecurity are emerging issues in the shrimp industry To achieve economic and environmental goals, the Government of Vietnam has ANALYZING THE PREFERENCES AND ATTITUDES implemented many policies to promote GAqPs However, few farmers apply GAqPs; currently, only about 2,410 hectares are VietGap certified (General OF PRODUCERS AND CONSUMERS FORDepartment GOOD of Fisheries, 2021), and about 9,000 hectares are certified GlobalGAP, ASC, Naturland (VASEP, 2020a) Reviewer 1: Farmers rarely apply GAqPs duePRACTICES to lacking capital,IN technology, and low shrimp prices (GIZ, AQUACULTURE DEVELOPMENT 2020) Reviewer 2: 1.1.3 Vietnamese seafood POLICY consumptionIN market contexts VIETNAM TheReviewer Vietnamese average shrimp consumption is about 1.81 kg/person/year, reaching 3: 2.27 kg/person/year, with total domestic consumption likely reaching nearly 300 thousand tons by 2030 (General Department of Fisheries, 2021) Shrimp is a typical food in Vietnamese meals, but most of it is fresh, uncertified shrimp, so it is difficult for consumers to Major:growing Economic assess the quality and safety Meanwhile, foodDevelopment safety concerns following food safety scandals (Ha et al., 2019) Certification of GAqPs effectively solves this problem (Hinkes and Code: 9310105 Schulze-Ehlers, 2018) However, very few GAqPs labeled shrimp are available in the The market dissertation will be defended at University of Economics Ho Chi Vietnamese 1.1.4 Theoretical background and empirical research contexts To date, studies have analyzed the attitudes and preferences on both sides of supply MinhnoCity and demand for sustainable aquaculture development (Bergleiter and Meisch, 2015; Weitzman Bailey, 2018), there are no studies No research synthesizes a theoretical At and hour day and month year SUMMARY OF DOCTORAL DISSERTATION framework that combines the values and concerns of both producers and consumers to explain the behavior of sustainable production and consumption of agricultural products Current studies only analyze preferences and estimate WTP either of producers (Ortega et al., 2013; Ngoc et al., 2016) or consumers (Cantillo et al., 2020) Therefore, explaining stakeholder acceptance for sustainable production and consumption has not been effective (Mogendi et al., 2016) 1.2 Research problems GAqPs development requires the involvement of both producers and consumers Theand dissertation can be found at library:no research has evaluated the (Bergleiter Meisch, 2015; Mogendi et the al., following 2016) However, potential for sustainable aquaculture development based on the willingness to pay for both supply and demand sides On the supply side, most research focuses on exploring farmers’ preferences for the economic benefits of GAqPs (Ortega et al., 2013; Ngoc et al., 2016), but Ho Chi Min City - 2022 farmer preferences for environmental protection and food safety assurance have not yet been implemented On the demand side, research into sustainable seafood preferences has yet to be conducted in developing countries (Tsantiris et al., 2018), and very few studies measure WTP for specific GAqPs certifications (Cantillo et al., 2020) Moreover, previous studies mainly apply one of the models of Conditional Logit (CLM), Multinomial Logit (MNL), Mixed Logit (MXL), and Latent Class (LCM) The most common of which is the MXL (Olum et al., 2019; Cantillo et al., 2020), but the MXL is also estimated only in the preference space, which is thought to bias the estimation results (Train and Weeks, 2005; Hole and Kolstad, 2012) There have not been any studies to estimate the above models simultaneously, calculate the MXL model in the preference space and the WTP space, and test to choose the Model that best fits the data and better support the goal of preference analysis Finally, assessing the attitudes of producers and consumers towards the same issue is the negative impacts of traditional aquaculture, and sustainable aquaculture development has not yet been realized (Hynes et al., 2017) Furthermore, the influence of attitude on preference has not yet been thoroughly investigated on both the supply side (Liu et al., 2018; Olum et al., 2019) and the demand side (Carlucci et al., 2015; Cantillo et al., 2015; Cantillo et al., 2020) The study “Analyzing the preferences and attitudes of producers and consumers for Good Aquaculture Practices in development policy in Vietnam” is necessary for the potential research gaps above 1.3 Research Objectives The overall objective of this study is to evaluate the potential development of GAqPs in shrimp farming in Vietnam based on the attitudes and preferences of producers and consumers To achieve this overarching goal, the following seven specific objectives need to be fulfilled: (1) Develop an analytical framework that incorporates the attitudes and preferences of producers and consumers to assess the potential development of GAqPs; (2) Analysis of producer preferences for compliance with environmental protection and food safety regulations; (3) Consumer preference analysis and WTP estimates for GAqPs labeled shrimp; (4) Analysis of the influence of attitudes towards negative impacts from traditional shrimp farming on the preferences of producers and consumers towards the development of GAqPs in shrimp farming; (5) Estimating the utility models according to the CLM, MXL, LCM models, testing and choose the most fitting model with the research data; (6) Estimating the WTP of producers and consumers in both the preference space and the WTP space, testing and selecting the estimation space in fitting with the data; and (7) Evaluation of attitudes towards negative environmental and social impacts from traditional shrimp farming, and attitudes towards the development of GAqPs in shrimp farming among producers and consumers in Vietnam 1.4 Research methodology and data 1.4.1 Research methodology The research applied Choice Experiment (CE) to analyze preferences The utility functions are estimated using CLM, MXL, and LCM models MXL is estimated in both the preference space and the WTP space Likelihood ratio test and Model fit statistics are applied to determine the Model and estimation space that fits the data The influence of attitude on preferences is analyzed using the interactive variable method Producer and consumer attitudes are examined using the Multiple Indicators Multiple Causes (MIMIC) model 1.4.2 data Data were collected from two surveys, one with 450 farmers in Khanh Hoa, Ninh Thuan, Soc Trang, Bac Lieu, and Ca Mau; the other with 459 consumers in Ho Chi Minh City, Nha Trang, Da Nang, and Hanoi 1.5 Research subjects and scopes 1.5.1 Research subjects The research objects are producer and consumer preferences and attitudes towards GAqPs development in shrimp farming Survey subjects are small-scale shrimp farmers and consumers 1.5.2 Research scopes 1.5.2.1 Scope of research contents This study focuses on four main contents: (1) literature review and analytical framework development, (2) developing attitudes and preferences measurement tools, (3) assessment of attitudes; and (4) preference analysis 1.5.2.2 Spatial scope Producers’ data were collected in the five largest shrimp farming provinces: Khanh Hoa, Ninh Thuan, Bac Lieu, Soc Trang, and Ca Mau Consumer data were collected in four cities: Ho Chi Minh City, Nha Trang, Da Nang, and Hanoi 1.5.2.3 Research times This study was carried out from February 2018 to February 2021 1.6 The thesis significance 1.6.1 Theoretical significance 1.6.2 Practical significance 1.7 Thesis layout This thesis consists of chapters: Chapter 1: Overview, Chapter 2: The literature review of theory, Chapter 3: Research methodology, Chapter 4: Research results and discussions, and Chapter 5: Conclusions CHAPTER LITERATURE REVIEW OF THEORY 2.1 The theory of measuring preferences and willingness to pay 2.1.1 Multi-Attribute Utility Theory Lancaster (1966) proposed the multi-attribute utility theory, assuming that consumers choose products with a combination of product attributes to maximize utility The utility (U) that consumer n derives from the attributes (x) of the goods is expressed as equation (2.1) Unlk = Unlk(x1; x2; ; xm) (2.1) 2.1.2 Random Utility Theory – RUT RUT hypothesizes that every individual acts rationally and chooses to maximize utility (McFadden, 1974, 1986) Assuming that decision maker n can choose between two alternatives i and j, the probability that alternative i is chosen is given by equation (2.2) Pnit = Pr[Unit > Unjt] ∀ i ≠ j; i,j ∈ T (2.2) The utility function of decision-maker n with individual characteristics Sn when choosing option i is written as equation (2.3) Unit = Vnit + εnit = Vnit(Xnit, Sn) + εnit(Xnit, Sn) ∀ i ∈ T (2.3) When the error terms εn are assumed following a Gumbel distribution, then the probability that a decision-maker n with characteristics Sn chooses option i is given by equation (2.11) (2.11) For the convenience of estimation, the function Vnit is assumed to be a linear function with parameters βk of the attributes Xnki and is written as equation (2.12) (Catestta, 2009) (2.12) Xnkit is the attribute k of the option i chosen by person n in the choice set t, and βk is the marginal utility of the attribute k Estimation results of coefficients βk show the probability of choosing option i of person n when k attributes change This study analyzes farmers’ choice of GAqPs development policy options and consumers’ choice of GAqP labeled shrimp The GAqPs policy options and GAqPs labeled shrimp are multi-attribute options For example, a GAqPs policy option combines incentives (soft interest rates, aquaculture insurance, technical assistance) and binding regulations (environmental protection, ensuring food safety) When choosing a GAqPs policy option, the farmer’s utility is the combining of the benefits of these attributes Thus, multi-attribute utility theory and RUT are most suitable for analyzing producers’ and consumers’ choice behavior 2.2 The relationship between environmental attitudes and ecological behavior 2.2.1 The concept of environmental attitudes and ecological behavior 2.2.2 The relationship between environmental attitudes and ecological behavior Environmental attitudes (EA) are considered to be a direct predictor of ecological behavioral (EB) intentions, thereby influencing ecological behavior (Milfont and Duckitt, 2004; Milfont, 2009; Best, 2010; Singh and Gupta, 2004) If a person is concerned about the environment, then he/she is more likely to act or adjust actions in a way that benefits the environment (Best, 2010; Singh and Gupta, 2013) In this study, attitudes to adverse impacts of traditional shrimp farming were defined as environmental attitudes The willingness to pay for farmed shrimp’s production and consumption under GAqPs is ecological behavior Therefore, the relationship between EA and EB was applied to analyze the influence of attitudes towards environmental and social impacts on farmed shrimp’s production and consumption preferences following GAqPs 2.3 Empirical literature review 2.3.1 Summary of farmers’ preferences for sustainable agriculture 2.3.1.1 Farmer preferences for GAqPs and sustainable agriculture 2.3.1.2 Farmer preferences for GAqPs development policy attributes 2.3.1.3 Factors affecting farmer preferences 2.3.2 Summary of consumer preferences for sustainable seafood 2.3.2.1 Consumer preferences for sustainable seafood 2.3.2.2 Consumer preference for sustainable seafood attributes 2.3.2.3 Factors affecting sustainable seafood consumption preferences 2.3.3 The measuring preferences methods 2.3.4 Research combines analysis of producer and consumer preferences 2.3.5 The research of assessing attitudes among stakeholders 2.3.5.1 The research of assessing public attitudes toward aquaculture 2.3.5.2 The research of assessing farmer attitudes toward aquaculture 2.3.5.3 The research of assessing the stakeholder attitudes toward aquaculture 2.4 Summary of research gaps From the literature review, the following potential gaps are identified 2.4.1 Theoretical gaps The theoretical review shows that previous studies have only focused on analyzing producer preferences (Bukchin and Kerret, 2018; Olum et al., 2019) or consumers (Cantillo et al., 2020) To date, no studies have synthesized a theoretical framework that simultaneously analyzes the preferences and attitudes of producers and consumers towards sustainable aquaculture development and sustainable agriculture Meanwhile, to develop sustainable aquaculture, it is necessary to consider the value of both the supply and demand sides (Fezzardi et al., 2013; Bergleiter and Meisch, 2015; Weitzman and Bailey, 2018) 2.4.2 Methodological gaps First, most previous research assessing preferences on both the supply and demand sides applied only one of the four common econometric models: MXL, LCM, MNL, and CLM (Olum et al., 2019) ; Cantillo et al., 2020) There have been no studies that simultaneously estimate these types of functions and test the fit with the data to choose the best analysis model and better support the objective of preferences of producers and consumers analysis Second, all previous studies evaluating preferences estimated the MXL model and calculated WTP in the preference space (Olum et al., 2019; Cantillo et al., 2020) Meanwhile, the disadvantage of the MXL estimates in the preference space is that the researcher subjectively specifies the distributions of the parameters As a result, the estimation results can be biased (Lancsar et al et al., 2017) and overestimate the value of WTP (Hensher et al., 2005; Hole and Kolstad, 2012) To the author’s knowledge, no studies have estimated MXL in the WTP space or combined to calculate the MXL in both preference and WTP spaces and test to choose the fittest estimation space with the data 2.4.3 Empirical gaps Firstly, research to explore farmer preferences for regulations on environmental protection and food safety assurance in aquaculture has not yet been carried out; previous reports mainly evaluate the farmer preferences for the economic benefit of sustainable aquaculture (Ortega et al., 2013; Ngoc et al., 2016; Xuan and Sandorf, 2020) Meanwhile, protecting the environment and ensuring food safety are mandatory requirements of GAqPs Second, research on sustainable seafood consumption preferences has not been conducted in developing countries (Tsantiris et al., 2018) The limited of these studies leads to a lack of understanding of sustainable seafood consumption behavior and limits the potential to improve the quality of locally consumed seafood and the well-being of domestic consumers, depriving them of opportunities to change the lives of local small-scale farmers and the sustainability of the local seafood industry In addition, previous research looking at certifications as an attribute of seafood (Carlucci et al., 2015; Cantillo et al., 2020) has not evaluated preferences for specific GAqPs certifications to assist producers in making decisions on the selection of production processes and markets Third, studies evaluating the producer and consumer attitudes towards the same issue of negative impacts of aquaculture and the GAqPs development have not been done (Hynes et al., 2017; Weitzman and Bailey, 2018) Previous studies have focused on assessing stakeholder attitudes toward different concerns (Weitzman and Bailey, 2018; Krøvel et al., 2019) Meanwhile, stakeholder attitudes and interactions play an essential role in determining the social acceptability of aquaculture (Mazur and Curtis, 2006; Freeman et al., 2012) Fourth, the relationship between environmental attitudes and preferences for sustainable seafood production and consumption has been poorly analyzed on both the supply and demand sides (Liu et al., 2018; Olum et al., 2019) Some previous studies only investigated the influence of attitudes on consumer preferences but only analyzed attitudes towards general environmental issues without assessing attitudes towards problems directly related to them with aquaculture (Hinkes and Schulze-Ehlers, 2018; Yi, 2019) Acceptance of a new policy or a new product is a psychological process for decision-makers (Olum et al., 2019), and integrating attitudes into evaluation models can improve the predictive model (Luzar and Cossé, 1998; Greiner, 2015) 2.5 Analytical framework of the thesis From the theoretical basis, potential gaps, and the context of shrimp farming, the analytical framework of the thesis is proposed in Figure 2.1 Compared with previous studies, which only analyzed the preferences and attitudes of consumers or consumers (Olum et al., 2019; Cantillo et al., 2020) The contribution of this study is to add to the theory in the field of research on production preferences and sustainable consumption in agriculture with an analytical framework that combines the assessment of the preferences and attitudes of producers and consumers towards sustainable aquaculture development Figure 2.1: Analytical framework Source: The author proposes based on the literature review CHAPTER RESEARCH METHODOLOGY 3.1 Research process This study was conducted with a combination of qualitative and quantitative research Qualitative research is applied in the first stage to overview research theory, identify gaps, propose an analytical framework, select attributes, design choice cards, and design attitude scales Quantitative research is applied to analyze the preferences and attitudes of producers and consumers This study applies choice experiments and CLM, MXL, and LCM models to investigate preferences Descriptive statistics, EFA, and the MIMIC model were used to assess the attitudes of producers and consumers 3.2 Methods of assessing the attitudes of producers and consumers 3.2.1 Attitude measurement method Attitudes are a latent structure; it is not easy to measure directly This work applied the indirect measurement with a unidimensional scale to create attitude scales (Bard and Barry, 2000) 3.2.2 Measuring environmental attitudes of producers and consumers Attitude scales were selected from relevant previous studies and adjusted through group discussions and expert consultation (Bard and Barry, 2000) Finally, the scale was tested with 30 shrimp farmers and 30 consumers through reliability analysis using Cronbach’s Alpha and exploratory factor analysis (EFA) (Bard and Barry, 2000) These items were evaluated on a 5point Likert scale with the endpoints of (strongly disagree) and (strongly agree) Table 3.1 Scales of environmental attitudes towards negative impacts of traditional shrimp farming Source: The author proposes based on the literature review 3.2.3 Producer and consumer knowledge measuring 3.2.3.1 Measuring knowledge methods 3.2.3.2 The measuring of producer and consumer knowledge about shrimp farming 3.2.3.3 Measuring consumer knowledge of GAqPs certifications 3.2.4 The methods of assessing the attitudes of producers and consumers This study applied the Multiple Indicators Multiple Causes (MIMIC) model to assess the farmer and consumer attitudes The MIMIC Model allows contextualization of the latent variable (attitudes) by simultaneously regressing it with cause factors in the CFA analysis of latent variables Therefore, MIMIC results reflect that all variables co-exist in practice and explain the attitude heterogeneity (Chang et al., 2020) 3.2.5 Multiple Indicators Multiple Causes (MIMIC) The MIMIC Model is an extension of CFA based on the Generalized Structural Equation Model (GSEM) with two parts: a measurement model and a structural model (Rabe-Hesket et al., 2004; Chang et al., 2020) 3.2.6 Model for assessing the attitudes of producers and consumers Figure 3.3: The MIMIC Model assesses attitudes toward traditional aquaculture Figure 3.4: The MIMIC Model assesses attitudes toward GAqPs development 3.3 Methods of analyzing producer and consumer preferences 3.3.1 The method selection of preference analysis and WTP estimation In this study, shrimp production and consumption according to GAqPs are multiattribute options For example, the policy for developing GAqPs is a combination of incentives attributes and binding regulations that farmers must comply with when adopting GAqPs CE is found to be suitable for analyzing preferences for these multi-attribute alternatives, effective in determining the importance of attributes and attribute levels in choice decisions, and allows for testing the trade-off between the attributes of the decision maker (Hanley et al., 2001) CE also allows WTP assessment of new products or attributes that not exist in the real market and for which secondary data is unavailable (Cantillo et al., 2020) Therefore, this study applied CE to analyze preferences for producing and consuming GAqPs labeled shrimp The CE method was developed by Louviere and Hensher (1982) and Louviere and Woodworth (1983) An application study of CE usually has four steps: (1) Attribute selection and attribute level determination; (2) Choice experiment design; (3) Choice card design and survey design; and (4) Measurement of preference 3.3.2 Choice experiment design to investigate farmer preferences 3.3.2.1 Defining of attributes and attribute levels The attributes of two choice experiments that investigate farmer preferences were developed in steps, including (1) synthesis of potential attributes, (2) potential attribute evaluation, and (3) attribute and attribute level selection Finally, four attributes of the experiment to assess farmer preferences for GAqPs (denoted as GAqPs DCE) include: Disease control, Price, Yield change; and Investment costs; four attributes of the experiment to explore farmers’ preferences with GAqPs development policy (denoted as GAqPs Policy DCE) include: Aquaculture insurance, Food safety ensuring, Environmental protection, and Soft interest rates 3.3.2.2 Choice Experiment and choice card design The D-efficiency experimental design was applied to the choice card design because it can optimize the amount of information collected from an experiment (Hensher et al., 2005) There are 24 choice cards generated and divided into three blocks; each block has four choice cards not to overload the respondent Each farmer made eight selection cards, with four scenarios for the GAqPs DCE experiment and four cards for the GAqPs Policy DCE experiment Examples of choice cards are shown in Tables 3.3 and Table 3.4 Table 3.3 An example of the choice card in GAqPs DCE Current Systems (no investment) Plan A Plan B Yield change +15% -15% Price premium 10% Outbreak disease Three times Twice times Once time Investment cost 800 400 □ □ □ Scenario I choose Table 3.4 An example of the choice card in GAqPs Policy DCE Scenario Water treatment Food safety Insurance Interest rate I choose Status quo Plan A Plan B Settling ponds Chemical method Biological processes No control No control Using antibiotics under Decision No.2625/QĐ-BNN-TY No insurance Schedule B No insurance 8.5% 4.5% 5.5% □ □ □ 3.3.3 Choice experiment design to investigate consumer preferences 3.3.3.1 Defining of attributes and attribute levels The labeled experiment was used in this study to determine the WTP for each specific certificate Alternatives are GAqPs labels applied in Vietnam’s shrimp farming, such as VietGap, GlobalGap, ASC, and Naturland 3.3.3.2 Choice Experiment and choice card design The choice cards were designed using Ngene software version 1.2.1 and applied Defficiency experimental design As a result, 32 choice cards were generated and divided into four blocks, which would not overwhelm respondents Only eight choice cards were presented to each participant Table 3.6 shows an example choice set 10 Table 3.6: An example of the choice card to investigate consumer preferences 3.3.4 Survey Design 3.3.5 Models and estimation methods 4.3.5.1 Models to analyze farmer preferences Farmer preferences for GAqPs are written by equations (3.8), (3.9), (3.10), and (3.11) Vi = αCosti + β1Asc_changei + β2Outbreak1i + β3Outbreak2i + β4Pricei + β5IncreaY15i + β6DecreaY15i (3.8) Vi = αCosti + β1Asc_changei + β2Outbreak1i + β3Outbreak2i + β4Pricei + β5IncreaY15i + β6DecreaY15i + β7 DecreaY15*Outb1i (3.9) Vi = αCosti + β1Asc_changei + β2Outbreak1i + β3Outbreak2i + β4Pricei + β5IncreaY15i + β6DecreaY15i + β8DecreaY15*Outb2i (3.10) Vi = αCosti + β1Asc_changei + β2Outbreak1i + β3Outbreak2i + β4Pricei + β5IncreaY15i + β6DecreaY15i + β9 DecreaY15*Pricei (3.11) Farmer preferences for GAqPs development policy are presented by equation (3.12) Vi = αInteratei +β1Asc_borrowi +β2InsuAi +β3InsuBi +β4Fosai +β5Chemisi +β6 Bioi (3.12) The effects of attitude towards negative environmental impact (EA) and attitudes toward food insecurity (FA) of traditional aquaculture on farmer preferences towards GAqPs and GAqPs development policy are presented in equations (3.13) and (3.14) Vi = αCosti +β1Asc_changei +β2Outbreak1i +β3Outbreak2i +β4Pricei +β5IncreaY15i +β6DecreaY15i +β7Change*EAi +β8Change*FAi +β9Deyield*EAi +β10Outbreak1*FAi +β11Outbreak2*FAi (3.13) Vi = αInteratei +β1Asc_borrowi +β2InsuAi +β3InsuBi +β4Fosai +β5Chemisi +β6Bioi +β7Borrow*EAi +β8Borrow*FAi +β9Chemis*EAi +β10Bio*EAi +β11Fosa*FAi (3.14) The effects of individual characteristics on farmer preferences for GAqPs and GAqPs development policy are written as equations (3.15) and (3.16) Vi = αCosti + β1Asc_changei + β2Outbreak1i + β3Outbreak2i + β4Pricei + β5IncreaY15i + β6DecreaY15i + β7Change*Agei + β8Change*Edui + β9Change*Inci + β10Change*Meki + β11Change*Sizei (3.15) Vi = αInteratei + β1Asc_borrowi + β2InsuAi + β3InsuBi + β4Fosai + β5Chemisi + β6Bioi + β7Borrow*Agei + β8Borrow*Edui + β9Borrow_Inci + β10Borrow*Meki + β11Borrow*Sizei (3.16) In functions (3.8) to (3.16), Vi is the observed utility when farmers choose the investment option or policy option i (i = 1; 3) Asc_change and Asc_borrow are dummy 13 three workers The average yield is 4.38 tons/crop, with an average yield of about 3.89 tons/ha The farms raise two crops yearly, earning nearly 17 million VND/month 4.1.2 Consumer data overview Out of 459 consumers, there are 375 women (81.7%) The average age of respondents was 37 years old More than half (57.08%) of consumers have a college or university degree, and the majority (77.78%) of respondents have full-time employment Each family has an average of members, and the average household income is 23.82 million VND/month The average frequency of buying shrimp is 2.81 times/month, each time buying 0.7 kg and spending about 153 thousand VND/time 4.2 Attitudes and knowledge of producers and consumers 4.2.1 The assessment of the reliability of the attitude scales 4.2.1.1 The assessment of the reliability of the environmental attitude scales Generally, farmers and consumers share the same attitude towards negative environmental impacts, the rating points of the two groups are 3.62 and 3.39, respectively (5point Likert scale) In contrast, consumers expressed high negative attitudes towards food safety issues with an average rating of 3.72 points Meanwhile, most farmers disagreed that traditional shrimp farming is at risk of food safety, with an average score of 2.51 points Cronbach’s alpha and EFA analysis for the farmer data, consumer data, and pooled data showed that seven observations were satisfactory, and two factors were extracted The first group consists of items, EA1, EA2, EA3, and EA4 measuring attitude towards environmental impact, denoted EA The second group consists of items, FA1, FA2, and FA3 measuring attitudes towards food safety problems, denoted FA 4.2.1.2 The assessment of the reliability of the scales of the attitudes toward GAqP development The mean rating points for attitudes toward GAqPs development of farmers and consumers are 3.9 and 3.29, respectively Cronbach’s alpha and EFA analysis with the farmer, consumer, and pooled data showed that three items were satisfactory; only one factor was extracted and denoted Gap_At 4.2.2 The assessment of consumer and producer knowledge 4.2.2.1 The assessment of consumer knowledge of traditional shrimp farming Most consumers did not understand conventional shrimp farming (39.22% of the total sample) Only 5.01% and 15.25% of the respondents were classified as having high and medium knowledge, respectively 4.2.2.2 The assessment of knowledge of farmers and consumers about GAqPs Nearly 90% of farmers reported they knew GAqPs In contrast, only about 43% of consumers self-reported knowing about GAqPs However, the test results showed that only 16.44% of farmers showed a high understanding of GAqPs, this rate in the consumer group was only 9.59% Most farmers have moderate knowledge (37.33%), and most consumers have no knowledge of GAqPs (56.64%) 4.2.2.3The assessment of consumer knowledge of GAqPs certifications Most of the respondents did not know the certifications and what they meant VietGap certification is the most well-known, with 31.59% of respondents correctly answering the meaning of this certification Consumers know fewer about GlobalGap, ASC, and NaturLand certifications Only 18.08% of respondents knew of GlobalGap labels, followed by Naturland and ASC, with 12.64% and 9.59% correctly recognizing these labels 14 4.3 Small-scale shrimp farmer preferences for GAqPs development 4.3.1 Small-scale farmer preferences for GAqPs A CLM model and five MXL models were estimated to investigate farmer preferences for properties of GAqPs Table 4.11 presents the estimated results Table 4.11 Analysis of shrimp farmer preferences for GAqPs Variables CLM Basic models MXL Preference space MXL with interactive variables Mơ hình Mơ hình Mơ hình Parameters Cost -0,002*** -0,003*** -0,003*** -0,003*** Asc_change 0,054 0,730** 0,751** 0,781** IncreaY15 0,421*** 0,648*** 0,644*** 0,574*** DecreaY15 -0,173** -0,192** -0,581*** -0,369** Price 0,270*** 0,195* 0,381*** 0,216* Outbreak1 1,143*** 1,434*** 1,159*** 1,416*** Outbreak2 0,702*** 0,889*** 0,973*** 0,657*** DecreaY15*Outbreak_1 0,968*** DecreaY15*Outbreak_2 0,418 DecreaY15*Price Standard deviation Cost Asc_change 3,059*** 3,103*** 3,037*** IncreaY15 0,215 0,249 0,212 DecreaY15 -0,340 -0,429* -0,327 Price 1,116*** 1,175*** 1,108*** Outbreak1 0,596*** 0,628*** 0,593*** Outbreak2 0,111 0,129 0,124 DecreaY15*Outbreak1 -0,171 DecreaY15*Outbreak2 -0,043 DecreaY15*Price Model fit statistics Respondents 450 450 450 450 Log-likelihood 1.818,45 -1.554,02 -1.548,48 -1.553,31 Wald chi2 260,76*** 260,77*** 311,10*** 328,46*** AIC 3.650,90 3.134,04 3.126,96 3.136,62 BIC 3.689,37 3.219,76 3.225,87 3.235,50 Likelihood ratio test 528,86*** 11,08*** 1,42 ***; **; * coefficients were significant at 1%; 5%; 10%, respectively -0,003*** 0,672** 0,616*** 0,314** 0,690*** 1,140*** 0,717*** MXL in WTP space -0,008*** 261,810*** 220,377*** -57,255** 69,485*** 441,365*** 266,807*** -1,134*** 3,071*** 0,226 -0,386 1,154*** 0,630*** 0,114 0,014* 895,432*** -138,057*** -97,207*** -345,235*** -59,163 152,702*** -0,186* 450 -1.546,56 269,28*** 3.123,11 3.222,03 14,92*** 450 -1.545,89 1.719,40*** 3.119,77 3.212,09 16,26*** The results showed that the MXL model fitted the data better than the CLM model, and the MXL model in the WTP space fitted the data better than the MXL model in the preference space WTP estimates in the WTP space were found to be more consistent with the claim that these values are less exaggerated and avoid subjectively assigning distributions of estimated parameters (Train and Weeks, 2005) (see Likelihood ratio test) Generally, the economic benefits (disease control, yield increase, and price) increase farmer preference for GAqPs However, suppose farmers must trade off economic benefits (reduced yields) for increased sustainability when adopting GAqPs In that case, they are not compensated by other financial benefits (e.g., increased prices, reduced disease) it is challenging to attract farmers Farmers were willing to pay the highest for the disease control of GAqPs, with a willingness to invest at VND 441 million/ha and VND 267 million/ha, respectively, with disease frequency reduced to only and times per crop If the yield of GAqPs increases by 15%, farmers are willing to pay 220 million VND/ha for the investment in GAqPs In contrast, 15 farmers need a 57 million VND/ha subsidy if the yield decreases by 15% compared to traditional shrimp farming Table 4.12 Farmer willingness to pay for shrimp farming following GAqPs WTP to invest GAqPs (million VND) Variables Preference space WTP space CLM MXL MXL Mean WTP 247 262 The yield increased 15% per crop 194 219 220 The yield decreased 15% per crop -80 -65 -57 Price increased by 10% 125 66 69 Disease outbreak time/crop 528 486 441 Disease outbreak time/crop 324 301 267 WTPs were only reported for significant parameters 4.3.2 Small-scale farmer preference for GAqPs development policy Small-scale farmer preferences for GAqPs development policy were analyzed using CLM, MXL, and LCM models Table 4.13 show that the MXL model fitted the data better than the CLM model The MXL model estimated in the WTP space improves over similar estimates in the preference space (see the Likelihood ratio test and the Model fit statistics) The WTP simulation values in the WTP space are lower than in the preferences space (see Table 4.15) This result is consistent with the recommendations that WTP in the willingness to pay space is more reliable, limiting the researcher’s subjective assignment of parameter distributions (Train and Weeks, 2005; Hole and Kolstad, 2012; Lancsar et al., 2017) Calculating WTP in both the preference space and the WTP space is a relatively new point compared to previous studies that only estimated WTP by simply scaling the non-monetary attribute parameter to monetary attributes (Olum et al., 2019) Generally, farmers were not willing to participate in the GAqPs development policy, nor were they willing to pay to protect the environment and ensure food safety, as this may reduce their economic benefits While preserving the environment and ensuring food safety are mandatory GAqPs regulations (Schwarz et al., 2017) However, farmers are willing to pay an average soft interest rate of about 3.66% and 2.92% on a one-year loan to invest in GAqPs if the policy covers Insu_A and Insu_B, respectively Investigating farmer preferences for environmental protection and food safety ensuring in GAqPs development is a departure from previous studies that focused on the financial benefits of GAqPs (Ortega et al., 2013; Ngoc et al., 2016; Xuan and Sandorf, 2020) Table 4.13: Farmer preferences for GAqPs development policy Variables CLM Parameters -13,343*** -0,734*** 0,534*** 0,390*** -0,080 0,089 -0,131 MXL in preference space Parameters Std.Dev -16,986*** 0,315 5,738*** 0,662*** -0,079 0,507*** -0,073 -0,101 -0,654*** 0,125 -0,086 -0,242* -0,679*** Interest Asc_borrow Insu_A Insu_B Fosa Chemis Bio Model fit statistics Respondents: 450 450 Log-likelihood: -1.931,22 - 1.478,042 Wald chi2 397,44*** 260,77*** AIC: 3.876,44 2.982,08 BIC: 3.914, 91 3.067,81 Likelihood ratio test: 906,36*** ***; **; * coefficients were significant at 1%; 5%; 10%, respectively MXL in WTP space Parameters Std.Dev -29,113*** 38,514 0,048 0,399*** 0,037*** 0,015** 0,029*** -0,004 -0,004 0,036*** 0,009 0,007 -0,013* 0,037** 450 -1.475,33 1.719,40*** 2.978,66 3.070,96 14,94*** 16 Many of the standard deviation parameters in the MXL model in Table 4.13 are significant, indicating the possibility of different classes of farmers Table 4.14 presents the LCM results with latent classes Table 4.14 Analyzing farmer preferences for GAqPs development policy using LCM Variables Class (Sustainable farmers) Parameters LCM model Class (Conservative farmers) Parameters Class (Innovative farmers) Parameters Choice model parameters Interest -29,442*** 3,361 -10,707*** Ascborrow 3,740 -6,846 0,811*** Insu_A 1,055*** 3,068 0,409*** Insu_B 0,175 3,101 0,522*** Fosa -0,505** -0,670 0,111 Chemis 1,313*** 1,175 -0,433*** Bio 0,859*** -0,135 -0,789*** Class membership model parameters (Class3 = Reference class) Farm size 0,489*** -0,32** Disease outbreak frequency -0,281 -0,202 Knowledge of GAqPs -0,311* -0,011** Environmental attitudes (EA) -0,090** 0,045 Food safety insecurity attitudes (FA) 0,098* -0,151* Attitudes toward GAqPs development (Gap_Att) 0,164 -0,131* Constans -0,701 0,931 Class probability (%) 25,9 30,9 43,2 Respondents 117 139 194 Class Characteristics Age (years) 44,04 49,47 46,57 Farms of intensive farming (%) 66,41 70,71 70,33 Farms of semi-intensive farming (%) 33,59 29,29 29,67 Farms size (ha) 2,25a 2,14a 2,13a Wastewater treatment size (ha ) 0,13a 0,12a 0,12a Percentage of the wastewater treatment area (%) 5,78 5,61 5,63 a a Yield per crop (tons) 4,40 4,28 4,43a Average yield (ton/ha) 3,71 3,88 4,03 a a Environmental attitudes (EA) 3,52 3,69 3,64a a a Food safety insecurity attitudes (FA) 2,58 2,43 2,51a a a Knowledge of GAqPs (5 points) 2,09 2,51 2,61a Attitudes toward GAqPs development (GAP_Att) 3,99c 3,71c 3,98c Model fit statistics Respondents: 450 Observations: 5.400 Log-likelihood: -1.455,20 AIC: 2.952,22 BIC: 3.136,85 a, b, c ***; **; * coefficients were significant at 1%; 5%; 10%; significant differences at 1%, 5%, 10% Class explains 25.9% of the observations (117 people) Farmers in Class were unwilling to participate in the policy of developing GAqPs Still, they were willing to pay an average interest rate of about 4.46% and 2.92% for a year loan to comply with wastewater treatment requirements by chemical and biological methods, respectively (see Table 4.15) The farms in this class have the highest percentage of land used for wastewater treatment of the three subclasses (5.78%), and 33.59% of farms are semi-intensive farms They also showed the highest pro-GAqPs attitude compared to the other two groups Therefore, farmers 17 in Class are referred to as the “Sustainable Farmers.” Class explains 30.9% of the observations (139 farmers) representing a group of farmers unwilling to participate in policy development GAqPs All estimated parameters are not statistically significant, indicating that these farmers are not interested in the proposed policy alternatives Most of the farms in Class are intensive shrimp farms (70.71%), with less support to develop GAqPs in the three classes (gap_Att mean score is 3.71 points) This group of farmers was named “Conservative Farmers.” Class explained 43.2% of the observations (194 people) Farmers in Class are willing to pay an average preferential interest rate of about 7.58% for a year loan to participate in the GAqPs development policy However, they request an interest subsidy of 4.04% and 7.37% for a year loan to comply with the chemical and biological wastewater treatment requirements, respectively (see Table 4.15) This farmer group was also the most knowledgeable about GAqPs, with an average knowledge score of 2.61 on a 5-point scale, and they also strongly support the development GAqPs (see Table 4.14) These results showed that Class is the group of farmers with the highest innovation ability, so Layer is called the “Innovative Farmer.” The combination of CLM, MXL, and LCM models to analyze farmer preferences is a new contribution compared to previous studies that only applied probit models (Ngoc et al., 2016) or MXL (Ortega et al., 2013) The simultaneous application of MXL and LCM models allows for better analysis of farmer preferences In this study, the combination of MXL and LCM models showed that farmer preferences were heterogeneous, and there were three different classes, including sustainable farmers, conservative farmers, and innovative farmers This result is beneficial for designing policies to promote the development of GAqPs in shrimp farming and can also be explained by the failure of previous policies, which have seen farmers are a homogeneous group Table 4.15 Farmer’s willingness to pay interest for a year loan WTP for interest rate (%) Preference space Variables CLM MXL LCM Mean Mean Class Class Asc_borrow 7,58 Insu_A 4,00 3,89 3,58 3,82 Insu_B 2,92 2,99 4,88 Fosa -1,72 Chemis 4,46 - 4,04 Bio -1,42 2,92 - 7,37 Note: WTPs were only reported as significant parameters WTP space MXL Mean 3,66 2,92 -1,26 4.3.4 The effects of socioeconomic characteristics on farmer preferences 4.4 Analyzing consumer preferences for GAqPs farmed shrimp 4.4.1 Consumer preferences for GAqPs labeled farmed shrimp The Likelihood ratio test and the Model fit statistics showed that the MXL results improved significantly compared to the CLM results Moreover, all standard deviation parameters in the MXL results were significant, indicating heterogeneity preferences Therefore, the MXL model fits the data better than the CLM model The parameters of MXL in the WTP space are similar to those in the preference space However, the Likelihood ratio test and the Model fit statistics showed that the MXL results in the WTP space are better than those in the preference space (see Table 4.17) Therefore, the MXL model in the WTP space is more suitable for the data and overcomes the subjective parameter distribution assignment by the researcher (Train and Weeks, 2005) Generally, consumers have a higher benefit when they buy GAqPs labeled shrimp than conventional shrimp Contrary to the perception that consumers in developing countries are less interested in sustainable seafood (Gardiner and Viswanathan, 2004; Ward and Phillips, 2008; Tsantiris et al., 2018); this study found that Vietnamese consumers preferred and were 18 willing to pay for GAqPs labeled farmed shrimp These results are similar to those found in studies in developed countries (Chen et al., 2015; Bronnmann and Asche, 2017; Hinkes and Schulze-Ehlers, 2018) These new findings broadened the understanding of sustainable consumption behavior in a typical developing market like Vietnam and showed the potential for sustainable aquaculture development based on domestic market incentives for certified farmed seafood Table 4.17 Results of analysis of consumer preferences for GAqPs labeled shrimp Choice CLM Parameters -0,019*** 6,667*** 5,798*** 5,916*** 6,213*** 4,418*** MXL in preference space Parameters Std.Dev -0,033*** 10,384*** 2,345*** 9,125*** 1,961*** 9,175*** 2,270*** 9,334*** 2,554*** 6,458*** 2,132*** Price VietGap GobalGAP ASC Naturland Nolabel Model fit statistics Respondents 459 459 Observations 22,032 22,032 Log-likelihood: -4.663,53 -3.778,59 Wald chi2 2.735,85*** 1.769,88*** AIC: 9.339,07 7.579,19 BIC: 9.376,32 7.667,19 Likelihood ratio test: 1.769,88*** ***; **; * coefficients were significant at 1%; 5%; 10%, respectively MXL in WTP space Parameters Sta.Dev -0,059*** 0,049*** 323,089*** 65,335*** 285,143*** 57,498*** 285,235*** 64,511*** 298,389*** 76,915*** 205,013*** -57,388*** 459 22,032 -3.643,99 9.568,62*** 7.311,98 7.407,98 269,21*** 4.4.2 The effects of consumer characteristics on preferences 4.5 Analyzing the effects of attitude on farmer preferences and consumer preferences 4.5.1 The effects of the attitudes on farmer preferences Table 4.20 presented the influence of attitude towards environmental impacts (EA) and attitude toward food safety insecurity (FA) in traditional shrimp farming on farmer preferences Table 4.20 The effects of attitudes on farmer preferences The effects of attitudes on preference for GAqPs The effects of attitudes on preferences for GAqPs development policy Variables Parameters Std Dev Variables Parameters Std Dev Cost -0,003*** Interest -16,986*** Asc_change 0,750** 2,966*** Asc_borrow 0,309 4,261*** IncreaY15 0,649*** 0,274 Insu_A 0,665*** 0,043 DecreaY15 -0,216** -0,367 Insu_B 0,508*** -0,264 Price 0,225** 1,136*** Fosa -0,097 0,559*** Outbreak1 1,462*** 0,683*** Chemis 0,132 0,124 Outbreak2 0,897*** -0,042 Bio -0,241* 0,713*** The effects of attitudes toward the environmental impacts of conventional shrimp farming (EA) Change*EA 0,325** 0,546*** Borrow*EA 0,707** 5,354*** Deyield*EA -0,101* 0,154 Chemis*EA 0,152* 0,073 Bio*EA 0,099 0,291 The effects of attitudes toward food safety insecurity of conventional shrimp farming (FA) Change*FA 0,172 0,699** Borrow*FA -0,534** 0,150 Outbreak1*FA -0,157 -0,102 Fosa*FA -0,055 -0,222* Outbreak2*FA 0,030 -0,213 Model fit statistics N: 450 Wald chi2 (9): 342,89*** N: 450 Wald chi2(9): 100,05*** AIC: 3.143,05 Observations: 5.400 AIC: 2.984,02 Observations: 5.400 BIC: 3.295,17 Log-likelihood: -1.548.75 BIC: 3.135,68 Log-likelihood: -1.469,01 Likelihood ratio test: 10,53*** Likelihood ratio test: 18,07*** 19 ***; **; * coefficients were significant at 1%; 5%; 10%, respectively The results showed that negative attitudes toward environmental impacts increase farmer preferences for GAqPs and GAqPs development policy The findings of this study are similar to those reported in previous studies that farmers with a positive environmental attitude will be willing to pay to adopt sustainable agriculture (Ruto and Garrod, 2009; Buckley et al., 2012; Greiner, 2015) The different findings of this work are that although farmers were aware of the negative environmental impacts of conventional shrimp farming However, in cases where GAqPs reduce economic benefits, incentives are still needed to encourage farmers to adopt more sustainable GAqPs or comply with environmental protection regulations It suggested that financial benefits are still the more important factor; environmental benefits can be achieved indirectly through applying GAqPs, where GAqPs must ensure that economic benefits are maintained This work is the first study to investigate the effects of attitudes toward environmental impacts on farmer preferences for GAqPs development Previous studies have overlooked this issue (Ortega et al., 2013; Ngoc et al., 2016), while aquaculture uses a lot of land and water resources and has severe environmental impacts (Nguyen Van Cong, 2017) Contrarily, it was found that attitudes towards food safety insecurity in traditional aquaculture did not increase farmer preferences for GAqPs Combining these results, it can be concluded that farmers are not aware of the problem of food safety insecurity in conventional shrimp farming However, this problem has been and is a big challenge for the shrimp farming industry in Vietnam (Chi et al., 2017) This work is the first study on farmer attitudes about food safety insecurity in aquaculture These findings showed that raising farmer awareness of food safety is very important 4.5.2 The effects of attitudes and knowledge on consumer preferences Overall, consumer attitudes towards negative environmental impacts from traditional shrimp farming (EA) did not increase the benefits for GAqPs labeled shrimp, except for VietGap labeled shrimp In contrast, negative attitudes towards food safety insecurity (FA) and knowledge of certifications (Know) increased consumer utility Table 4.21 The effects of attitudes and knowledge on consumer preferences Variables CL MXL in preference space MXL in WTP space Parameters Parameters Std.Dev Parameters Std.Dev Price -0,020*** -0,034*** -0,083*** 0,104*** VietGap 6,699*** 10,550*** 2,230*** 309,860*** -56,323*** GlobalGAP 6,032*** 9,533*** 1,539*** 280,930*** -35,174*** ASC 6,095*** 9,284*** 1,599*** 289,851*** 50,477*** Naturland 6,279*** 9,673*** 2,186*** 294,718*** -44,581*** Nolabel 4,594*** 6,877*** 2,089*** 207,390*** 56,391*** Interaction terms with the attitudes towards negative environmental impacts (EA) VietGap*EA 0,197*** 0,451* -0,708*** 13,128*** -18,314*** GlobalGap*EA -0,018 0,039 -0,339 1,808 -4,793 ASC*EA 0,123 0,208 -0,680*** 11,036*** -3,377 Naturland*EA 0,089 0,150 0,439 -4,912 19,542*** Interaction terms with the attitudes towards food safety insecurity (FA) VietGap*FA 0,430*** 0,507** 0,780*** 8,688*** 32,354*** GlobalGap*FA 0,551*** 0,829*** 1,384*** 35,472*** 37,033*** ASC*FA 0,557*** 0,977*** 1,269*** 25,386*** 12,595*** Naturland*FA 0,536*** 0,581*** 0,862** 21,570*** -38,649*** Interaction terms with the knowledge of certifications (Know) VietGAP*Know 0,895*** 1,269*** -0,223 35,136*** 39,379*** GlobalGap*Kno 0,444** 0,775** 0,604 28,199*** -49,372*** w ASC*Know 1,222*** 1,881*** 0,003 44,593*** -63,308*** NaturLand*Know 1,636*** 1,850*** 2,620*** 59,718*** 19,410*** Model fit statistics 20 Variables CL MXL in preference space Parameters Parameters Std.Dev Respondents: 459 459 Log-likelihood: -4.438,05 -3.621,93 AIC: 8.912,10 7.313,86 BIC: 9.023,85 7.593,87 Likelihood ratio test: 1.632,24*** ***; **; * coefficients were significant at 1%; 5%; 10%, respectively MXL in WTP space Parameters Std.Dev 459 -3.475,26 7.022,52 7.310,53 293.34*** Many of the standard deviation parameters reported in Table 4.21 were significant, implying that consumers had heterogeneity preferences One LCM model with three latent classes is presented in Table 4.22 Table 4.22: The latent class results Variables Class (Safety Consumers) Parameters Class (Traditional Consumers) Parameters Choice model parameter Price -0,002 -0,012*** VietGap 3,658*** -41,127 GlobalGAP 2,340*** -0,572 ASC 2,834*** -0,199 Naturland 2,588*** -0,973 Nolabel -0,778 1,045* Interaction terms with the attitudes towards negative environmental impacts (EA) VietGap*EA 0,038 -3,938*** GlobalGap*EA -0,472 -2,281*** ASC*EA 0,332 -1,511*** Naturland*EA -0,763 -1,135*** Interaction terms with the attitudes towards food safety insecurity (FA) VietGap*FA 1,067*** 7,725*** GlobalGap*FA 2,563*** 1,407** ASC*FA 1,813*** 1,441*** Naturland*FA 1,151*** 3,946*** Interaction terms with the knowledge of certifications (Know) VietGAP*Know 1,253*** 39,294 GlobalGap*Know -1,189** 1,479 ASC*Know -0,657 5,151*** NaturLand*Know 12,866 -2,685*** Class membership model parameters: Class3 = Reference class Buying frequency -0,191** -0,267*** Aquaculture knowledge 0,161 -0,310** GAqPs knowledge -0,055 0,123 Attitudes toward GAqPs 0,371** 0,082 Income 0,032*** 0,019 Constant -3,281*** -1,787*** Class probability (%) 16,2 10,3 Respondents 74 47 Class Characteristics Monthly household income (VND million) 30,81a 17,42a Monthly shrimp buying frequency 2,57a 2,08a (time/month) Amount per shrimp purchasing (1000 VND) 203,54a 133,75a a Environmental Attitudes (EA) 3,18 3,22a a Food safety insecurity Attitudes (FA) 3,86 3,63a Class (Sustainable Consumers) Parameters -0,032*** 10,955*** 10,449*** 10,444*** 10,682*** 8,118*** 0,378*** 0,228** 0,227** 0,216** 0,499*** 0,532*** 0,463*** 0,756*** 1,250*** 1,568*** 0,962*** 2,286*** 73,5 338 23,25a 2,96a 145,68a 3,46a 3,70a 21 Variables Class (Safety Consumers) Parameters 26,39a 25,78a 15,27a 12,03a 3,53a Class (Traditional Consumers) Parameters 33,33a 12,08a 7,44a 9,05a 3,33a Class (Sustainable Consumers) Parameters 32,44a 17,30a 8,69a 13,27a 3,23a VietGap knowledge (%) GlobalGAP knowledge (%) ASC knowledge (%) Naturland knowledge (%) Attitudes toward GAqPs development Model fit statistics Respondents: 459 AIC: 6,943,48 Log-likelihood: -3,439,53 BIC: 7,471,50 ***; **; * coefficients were significant at 1%; 5%; 10%; a, b, c significant differences at 1%, 5%, 10% Class accounted for 16.2% of the observations (74 people), representing consumers willing to pay any price to buy GAqPs certified farmed shrimp and choosing only GAqPs labeled shrimp Consumers in this class were very concerned about food safety, which may be the main reason for their willingness to pay premium prices for certified shrimp Therefore, Class is named “Safety consumers.” Class explained that 10.3% of observations (47 people) represented a group of consumers who were willing to pay only for conventional shrimp (Nolabel) Therefore, Class is labeled “Traditional Consumers.” Class accounted for 73.5% of observations (338 people) and was a group of consumers willing to pay for both GAqPs and conventional shrimp Class includes those who have a negative attitude towards environmental impacts, and this attitude increases the consumer utility for certified shrimp Therefore, Class is called the “Sustainable Consumer.” Combining CLM, MXL, and LCM models to analyze preferences for sustainable seafood in this study is a new contribution compared to previous studies that only applied the MXL model (Cantillo et al., 2020) The simultaneous application of these models allows for a better analysis of consumer preferences The findings showed that consumers were heterogeneous preferences, and there were three different classes, including “Safety consumers,” “Traditional consumers,” and “Sustainable consumers.” Unlike the findings in previous studies stating that consumers in developing countries ignore environmental issues when purchasing food (Faltmann, 2019; Tsantiris et al., 2018) From the LCM results, this study found that a class of consumers (more than 73% of survey respondents) put environmental concerns into their decisions Combining MXL and LCM provides a deeper understanding of consumer behaviors and concerns when choosing seafood Table 4.23 presented the estimated WTPs under the effects of attitudes and knowledge Table 4.23: WTP for GAqPs labeled shrimp under the effects of attitude and knowledge Shrimp products WTP (1000 VND/kg) Preference space CLM MXL LCM Mean Mean Class Class VietGap 331 314 337 GlobalGap 298 284 322 ASC 301 276 321 Naturland 311 288 329 Nolabel 227 205 90 250 The WTP for sustainability labeled shrimp under the effect of environmental attitudes VietGap 341 328 349 GlobalGap 329 ASC 328 Naturland 335 The WTP for sustainability labeled shrimp under the effect of food safety concerns VietGap 353 329 352 WTP space MXL Mean 310 281 290 295 207 323 301 319 22 Shrimp products WTP (1000 VND/kg) Preference space CLM MXL LCM GlobalGap 326 309 338 ASC 329 305 336 Naturland 337 305 352 The WTP for sustainability labeled shrimp under the effect of consumer knowledge VietGap 376 352 376 GlobalGap 320 307 370 ASC 362 332 351 Naturland 391 343 399 Note: WTPs were only presented for positive and significant coefficients WTP space MXL 316 315 316 345 309 334 354 Calculating WTP in both the preference space and the WTP space is a new contribution compared to previous studies that only estimated WTP by taking the ratio between the nonmonetary to the monetary attribute parameters (Cantillo et al events, 2020) These findings confirm that the WTP values in the WTP space are more reliable than those in the preference space, limiting the researcher’s subjective designation of the parameter distribution (Train and Week, 2005; Hole and Kolstad, 2012) Generally, Vietnamese consumers preferred and were willing to pay premium prices for GAqPs labeled shrimp VietGap has the highest willingness to pay (VND 310 thousand/kg) However, if consumers know certifications, Naturland shrimp will have the highest WTP (399 thousand VND/kg) In conclusion, consumers preferred and were willing to pay for GAqPs labeled shrimp, driven by food safety concerns and knowledge of GAqPs labels Notably, the LCM results show that more than 70% of consumers put environmental concerns into their decisions 4.6 The assessment of the attitudes of producers and consumers 4.6.1 Assessing attitudes toward negative impacts of traditional shrimp farming Table 4.24 presents the results of farmer and consumer attitudes assessment toward the adverse effects of conventional shrimp farming Table 4.24 Assessing attitudes towards negative impacts of traditional shrimp farming Variables Farmers Consumers Pooled data Structural equation Latent variable EA – The Attitude toward the environmental impacts of traditional shrimp farming Producer 0,379*** Gender -0,176 0,392*** 0,198*** Age 0,013 0,008* 0,008*** Marital status -0,045 -0,203* -0,098* Secondary school 0,219** 0,043 0,092 High school diploma 0,009 0,125 0,032 Undergraduate degree 0,354* 0,359*** Graduate degree 0,179 0,375*** Household size -0,018 0,001 -0,002 Monthly households income 0,014** 0,004 0,006*** Shrimp farming experiences -0,006 Farm size 0,105** Monthly shrimp buying frequency 0,061*** Low knowledge 0,247*** Medium knowledge 0,779*** High knowledge 1,237*** Latent variable FA – The Attitudes toward food insecurity in traditional shrimp farming Producer -1,177*** Gender -0,111 0,243* 0,201** Age -0,006 0,005 0,003 Marital status 0,280 0,222* 0,278*** Secondary school 0,018 0,124 -0,007 23 Variables Farmers Consumers Pooled data High school diploma 0,055 0,634** 0,238** Undergraduate degree 0,532** 0,365*** Graduate degree 0,558** 0,482*** Household size 0,018 0,014 0,007 Monthly households income -0,002 0,005 0,004 Shrimp farming experiences 0,005 Farm size -0,022 Monthly shrimp buying frequency 0,027 Low knowledge 0,070 Medium knowledge 0,425*** High knowledge 0,539** Measurement equation Latent variable EA – The Attitude toward the environmental impacts of traditional shrimp farming EA1 (constrained) (constrained) (constrained) EA2 1,031*** 0,984*** 1,061*** EA3 1,007*** 0,885*** 1,557*** EA4 0,804*** 0,913*** 1,654*** Latent variable FA – The Attitudes toward food insecurity in traditional shrimp farming FA1 (constrained) (constrained) (constrained) FA2 1,293*** 0,949*** 1,064*** FA3 1,095*** 0,869*** 0,925*** ***; **; * coefficients were significant at 1%; 5%; 10%, respectively Generally, this study is the first to assess the attitudes of both supply and demand sides to the negative impacts of aquaculture, unlike previous studies that only evaluated the attitudes of farmers, the public, and stakeholders but excluded consumers The results showed the heterogeneity in the attitudes of producers and consumers towards the same concerns in traditional shrimp farming Producers have a negative attitude towards environmental impacts, about 0.379 units higher than consumers on average In contrast, consumers have a higher negative attitude about food insecurity than the producer, about 1,117 units These results differ from what has been shown in previous studies reporting that have shown that people with little direct involvement in aquaculture have negative attitudes higher than those directly involved (Chu et al., 2010; Bacher et al., 2014; Weitzman and Bailey, 2018; Krøvel et al., 2019) 4.6.2 Assessing attitudes toward GAqPs development in shrimp farming Table 4.25 presents the results of assessing the attitudes of producers and consumers toward GAqPs development in shrimp farming The results showed that, on average, farmers have a higher pro-GAqPs attitude than consumers by 0.718 units Contrary to previous studies reporting that only those directly involved in aquaculture (farmers, managers, researchers) believe sustainable aquaculture has many benefits and is addressed given the stakeholders’ environmental and social concerns In contrast, the local public and environmentalists are skeptical about the potential benefits of sustainable aquaculture (Bacher et al., 2014; Weitzman and Bailey, 2018) This study shows that both farmers and consumers believe that shrimp farming under GAqPs can address the adverse effects of shrimp farming Table 4.25 Assessing producer and consumer attitudes toward GAqPs development Variables Structural equation Latent variable Gap_Att Producer Gender Age Marital status Secondary school Farmers Consumers Pooled data 0,023 -0,005 -0,002 0,122 0,068 0,003 -0,293*** 0,094 0,718*** 0,055 0,001 -0,232*** 0,110 24 Variables Farmers Consumers High school diploma 0,222** 0,230 Undergraduate degree 0,378*** Graduate degree 0,294 Household size 0,008 Monthly households income 0,098*** Shrimp farming experiences 0,006 0,010 Farm size 0,006 0,009*** Monthly shrimp buying frequency 0,085*** Low knowledge 0,154 -0,064 Medium knowledge 0,535*** 0,718*** High knowledge 0,496*** 0,874*** Measurement equation Gap_Att Gap_Att1 (constrained) (constrained) Gap_Att2 1,122*** 1,126*** Gap_Att3 1,011*** 1,087*** ***; **; * coefficients were significant at 1%; 5%; 10%, respectively Pooled data 0,213** 0,437*** 0,398*** 0,014 0,009*** 0,066 0,599*** 0,649*** (constrained) 1,118*** 1,094*** CHAPTER CONCLUSIONS AND RECOMMENDATION 5.1 Conclusions This study analyzed and showed that producers and consumers have positive preferences and attitudes towards the development of GAqPs, showing the potential to develop shrimp farming under GAqPs in Vietnam The thesis has achieved all the research objectives set out Specifically, the key findings of this study are summarized below First, the study has developed a theoretical framework to combine the analysis of the attitudes and preferences of producers and consumers towards the development of GAqPs in shrimp farming The analytical framework is presented in Figure 2.1 Second, the analysis of preferences of small-scale shrimp farmers shows that they were unwilling to pay to protect the environment and ensure food safety, even if there was a preferential loan policy Third, consumer preference analysis shows that they preferred and were willing to pay a premium price for GAqPs labeled shrimp than conventional shrimp Consumers prefer organically farmed shrimp when they understand the certifications Fourth, environmental attitudes increased farmer preferences for GAqPs and GAqPs development policy However, if the economic benefits of GAqPs are reduced (for example, yield reduction), farmers still need incentives to comply with environmental protection and food safety regulations From the demand side, the LCM results showed that over 73% of consumers were willing to pay more for GAqPs labeled shrimp under the influence of environmental attitudes Fifth, the MXL and LCM models are more consistent with the research data than the CLM models The combination of MXL and LCM models better supports the analysis of heterogeneity preference and shows that farmers and consumers are not a homogeneous group Sixth, the WTP estimates in the willingness to pay space are less exaggerated than those in preference space Seventh, producers and consumers share a common negative attitude towards the environmental impacts of traditional shrimp farming, but producers show a higher negative attitude than consumers In contrast, consumers have higher negative attitudes than producers about food insecurity in conventional shrimp farming Notably, producers and consumers support the development of GAqPs in shrimp farming 5.2 Policy Implications 5.2.1 Policy implications for promoting the development of GAqPs in shrimp farming First, financial support policy for farmers to set up wastewater treatment systems in shrimp farming is necessary Second, compliance with regulations on antibiotics and germicidal chemicals should be closely monitored and considered a binding condition for farmers to receive preferential loans Third, aquaculture insurance should be used to 25 incentivize farmers to apply GAqPs to shrimp farming Fourth, raise farmers’ awareness of the disease prevention benefits of GAqPs Fifth, increase farmers’ awareness of the negative environmental impacts of traditional shrimp farming and the environmental benefits of GAqPs Finally, ensure output markets and price stability to attract farmers to switch and maintain shrimp farming under GAqPs 5.3.2 Policy implications for market development for sustainably farmed shrimp First, communication strategies should focus on raising consumer awareness of the food safety benefits of GAqPs labeled shrimp Second, public communication programs must highlight the contrast between the environmental benefits of GAqPs and the negative environmental and social impacts of conventional shrimp farming Third, increase the presence and understanding of GAqPs standards in the community through supermarkets and convenience stores Fourth, there should be product strategies for different consumer segments 5.2.3 Policy implications for promoting simultaneous production and consumption of GAqPs labeled shrimp First, reducing the asymmetric information in the production and consumption of farmed shrimp Second, the supply chain from farm to supermarket and dining table should be applied in shrimp farming to ensure high prices, encourage farmers to invest in production, and be willing to pay consumers Third, in the long term, organic certification should be the most developed priority 5.3 New contributions of thesis The results of this thesis are an attempt to fill the identified research gap From the analysis results, this study has the following contributions (1) Theoretical contributions The thesis has built a framework for analyzing the preferences and attitudes of producers and consumers toward sustainable aquaculture, which can be considered a necessary addition to the research theory of production and consumption preferences for sustainable agriculture (2) Methodology contributions First, combining estimation of CLM, MXL, and LCM models, testing, and selecting the suitable model fitted with data is a new contribution to the estimation procedure compared with previous studies which mainly applied the MXL model (Olum et al., 2019; Cantillo et al., 2020) This combination improves efficiency in preference analysis Second, estimating WTP in both preference space and WTP space is an improvement in the estimation method compared to previous studies, which mainly calculated WTP by taking the ratio between nonmonetary attribute and monetary attribute parameters (Olum et al., 2019; Cantillo et al., 2020) (3) Empirical contributions First, analyzing farmer preferences for externality benefits of GAqPs such as environmental protection and ensuring food safety has filled the gap left by previous studies by focusing only on farmer preferences for the economic benefits of sustainable aquaculture practices (Ortega et al., 2013; Ngoc et al., 2016; Xuan and Sandorf, 2020) Second, analyzing Vietnamese consumer preferences for GAqPs labeled shrimp is the first empirical study in developing countries This work is an attempt to fill the gap left by previous studies that only focus on analyzing the potential for sustainable seafood markets (Kehoe et al., 2016), the willingness to pay for information on food safety (Danso et al., 2017; Zou, 2017), and behavior of seafood producers to promote exports (Tsantiris et al., 2018) Contrary to previous studies that consumers in developing countries are not interested in sustainable seafood, this study shows that Vietnamese consumers are willing to pay for sustainably farmed shrimp Third, assessing the producer and consumer attitudes toward the adverse effects on the environmental and social from traditional shrimp farming is a new contribution compared to previous studies that only assessed public attitudes in different countries towards 26 aquaculture (Freeman et al., 2012; Hynes et al., 2017), or stakeholder attitudes toward various concerns (Bacher et al., 2014; Weitzman and Bailey, 2018; Krøvel et al., 2019) Fourth, analyzing the effects of attitudes toward environmental and social impacts on preferences for sustainably farmed shrimp production and consumption is an attempt to fill the gap left by previous studies that focus on individual and socioeconomic factors (Kumar et al., 2018; Olum et al., 2019; Cantillo et al., 2020), and measures only attitude toward general environmental issues (Hinkes and Schulze-Ehlers, 2018; Yi, 2019) not attitudes towards direct problems related to aquaculture 5.4 Thesis limitations and further research Although the study achieved its objectives, it also has some limitations that further studies can exploit and expand on Firstly, this study all constrained the parameters of the attributes that were not correlated with each other Further studies should estimate the benefit models with different assumptions and parameter constraints, including the assumption of random and correlated parameters, thereby comparing the calculated results to get the best results Second, this study could not recommend aquaculture insurance programs in detail but only focused on compensation schedules It will be more interesting if there is a new study analyzing the effect of aquaculture insurance on the willingness to invest in sustainable aquaculture Third, this study did not include risk attributes in the choice experiments, so the impact of risk on farmer preferences and WTP was not analyzed Further studies should consider risk as an attribute of investment options and policy options to analyze farmer preferences LIST OF RESEARCH WORKS OF THE AUTHOR Articles (1) What motivates farmers to accept good aquaculture practices in development policy? Results from choice experiment surveys with small-scale shrimp farmers in Vietnam Economic Analysis and Policy, 72(2021), 454–469 (2) Willingness to adopt improved shrimp aquaculture practices in Vietnam Aquaculture Economics and Management, 25(2021: 4), 430-449 Accepted paper (3) The Effect of Sustainability Labels on Farmed Shrimp Preferences: Insights from a Discrete Choice Experiment in Vietnam Aquaculture Economics and Management Conference Proceedings (1) Consumer Preferences for Sustainable Certified Farmed Shrimp Products in Vietnam: The Role of Sustainable Certification, Consumer Attitudes, and Knowledge The Handbook of The 2nd Asia Conference on Business and Economic Studies, Abstract paper 42 ISBN: 978-604-922-768-4 (2) Phân tích đặc điểm cầu tiêu dùng thị trường nội địa cho mặt hàng tôm Việt Nam Kỷ yếu Hội thảo Phương pháp Thống kê Kinh tế lượng - Ứng dụng Kinh 27 tế Kinh doanh, Trường Đại học Kinh tế TP Hồ chí Minh, Trang 173-192 ISBN: 978-604-80-3675-1 (3) Những mối quan tâm định đầu tư sản xuất bền vững nông dân: Trường hợp nghề nuôi tôm Việt nam Kỷ yếu Hội thảo khoa học Quốc gia: Tăng trưởng xanh: Quản trị Phát triển Doanh nghiệp, Trang 927-940 ISBN: 978-604-9963-46-9 (4) Do Vietnamese Small-scale Shrimp Farmers Prefer Good Aquaculture Practices? A Result of Choice Experiments Proceedings of The Second International Conference in Business, Economics & Finance, Pages 123-132 ISBN: 978-604-965-469-5 ... Những mối quan tâm định đầu tư sản xuất bền vững nông dân: Trường hợp nghề nuôi tôm Việt nam Kỷ yếu Hội thảo khoa học Quốc gia: Tăng trưởng xanh: Quản trị Phát triển Doanh nghiệp, Trang 927-940... Economic Studies, Abstract paper 42 ISBN: 978-604-922-768-4 (2) Phân tích đặc điểm cầu tiêu dùng thị trường nội địa cho mặt hàng tôm Việt Nam Kỷ yếu Hội thảo Phương pháp Thống kê Kinh tế lượng -... METHODOLOGY 3.1 Research process This study was conducted with a combination of qualitative and quantitative research Qualitative research is applied in the first stage to overview research theory, identify