social network and norms influencing adoption of organic farming accounting for heterogeneities among vietnamese small household farmers

Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 Social network and norms influencing adoption of organic farming: Accounting for heterogeneities among Vietnamese small household farmers Tran Nam Quoc1, Tiet Tong Tuyen2 University of Management and Technology1, Ho Chi Minh City (Vietnam) bBETA, INRAE, University of Strasbourg, Strasbourg (France)2 ARTICLE INFO Keywords: Personal norm; Organic farming adoption; Social networks; Social norms; Vietnam ABSTRACT This research examines the influence of socio-psychological factors in promoting organic agriculture in Vietnam, including social networks, social norms and personal norms Our findings suggest that social networks, such as the frequency of communication and the existence of organic farming neighbors, are critical components of organic agriculture Moreover, social and personal norms could also play a key role in incentivizing envi- ronmentally concerned farmers to convert to organic farming Therefore, policymakers should encourage neighborhood collaboration, establish a channel for farmers to promote interaction between farmers and promote farmers’ recognition of the importance of or- ganic agriculture to effectively drive them toward the sustainable adoption of organic farming Introduction Food safety has currently been a global issue Organic farming, an alternative farming system to produce healthier food, has expanded significantly with 11.7 million hectares of agricultural land (about 20% of agricultural area) dedicated to organic farming in 2017 (Willer and Lernoud,2019) According to current evidence, organic agriculture is better than conventional agriculture since organic farmers rely less on chemical inputs (e.g., pesticides and chemical fertilizers) Several studies indicated that organic farming could contribute to societal well-being by supplying healthy meals to consumers and also help reduce chemical reliance that results in a detriment to water resources, crop pollution and soil fertility (Bengtsson et al.,2005;Zhengfei et al.,2005) Numerous developing nations, such as Vietnam, are now in a start-up phase, and their agriculture has been restructured by improving crop diversification, crop produc- tion, international trade and application of agricultural inputs (Rapsomanikis,2015) According to the authors’ study, about 40% of land in Vietnam is dedicated to agri- culture, and agricultural products are the primary source of income for the majority of smallholder farmers (nearly 89%) living in rural areas (Rapsomanikis,2015) Nonethe- less, the authors’ study established that Vietnamese farmers heavily rely on agricultural inputs such as pesticides, fertilizers, and crop protection inputs (Rapsomanikis,2015) As a consequence, organic agriculture accounted for less than 0.5% of total agricultural land in Vietnam in 2017 (Willer and Lernoud,2019) The present study indicates that farmers’ decisions to practice organic farming are influenced by a range of economic and political factors, including agricultural policy, market structure and technology design (Schneeberger et al.,2002;Darnhofer et al.,2005; Jaime et al.,2016) For instance, policies subsidizing organic farmers with guaranteed in- comes and access to agricultural credit could encourage them to convert their farmlands to organic agriculture (Schneeberger et al.,2002;Darnhofer et al.,2005) Along with economic and political factors, social factors (e.g., social influence, norms and nudges) could influence 180 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 farmers’ decision-making processes and also help incentivize them to be- have more positively toward organic farming (Lynne et al.,1988;Dessart et al.,2019; Streletskaya et al.,2020) While numerous studies have investigated the impact of social factors, such as norms and nudges, on conversion to organic farming, a handful of evidence on the social network factors (e.g., connections and interaction with other organic farmers) that assess farmers’ decision to pursue organic agriculture This article contributes to the literature by assessing the impact of several socio-psychological factors, including social network, social norms and personal norms, on promoting organic farming in Vietnam In addi- tion, we also account for farmer heterogeneity in environmental concerns to quantify the heterogeneous effects of these factors on farmers’ organic farming conversion decisions Furthermore, our study aims to advance our understanding of (mostly small household) farmers’ organic farming investment decisions in Vietnam Note that small household farmers in developing countries like Vietnam are experiencing severe negative impacts from food safety issues than those in developed countries The remainder of the paper is structured as follows Section is a review of the literature Section details the data collecting procedure, the econometric model, and descriptive statistics Section discusses the estimation results Section discusses and concludes Literature review Organic farming in Vietnam Even though conventional agriculture could provide Vietnamese farmers with high yields and income from farming, this kind of farming technique that heavily relied on chem- ical inputs, such as pesticides and fertilizers, is considered unsustainable (Berg,2002) The impacts of chemical substances are permanent For instance, pesticide misuse re- sults in the development of insect resistance, which has long-term negative consequences for both farmers’ and consumers’ health as well their neighborhood living environment (i.e., environmental degradation) (Berg and Tam,2018) As a result, nearly two million Vietnamese farmers have been affected by pesticide and fertilizer-related health problems (Thai et al.,2017) Moreover, the adverse effects of chemical inputs would also have the reverse effect on yield and thus result in decreased productivity and profit in the long-term (Berg and Tam,2018) As a consequence, in 2008, the Vietnamese government issued a legal framework and standardizing organic agriculture operations, namely “Vietnam Good Agricultural Practices” (VietGap), to develop safe farming methods and resolve food insecurity issues (Mergenthaler et al.,2009;Chau and Anh,2015;Thai et al.,2017;Willer and Lernoud,2019) Since Vietnamese consumers think labeled food is more healthy, safe and willing to pay a premium for it, VietGap has been shown to have several positive impacts on the food security and productivity of farmers (Chau and Anh,2015) According to the World Bank report, organic agriculture accounted for about 58,018 hectares (i.e., 0.5% of total agricultural land) with 10,150 organic farmers in 2017 (Willer and Lernoud,2019) This organic farmland is still smaller compared to that of other countries in South-East Asia (e.g., Thailand and Cambodia) (FAO,2017) The low rate of adoption of organic farming is because of many reasons, such as the lack of awareness about the harmful effect of chemical input and information about organic farming practices (Vu et al.,2020), high cost of production and lack of marketability (i.e., the low prospect of high income) (Bui and Nguyen,2021) However, despite the challenges to the development of organic farming in Vietnam, the demand for organic foods is growing because of the increase in consumers’ awareness about food hygiene and safety related to pesticide use in agriculture (Van Huy et al., 2019) In this perspective, the Vietnamese government, particularly the Ministry of Agriculture and Rural Development (MARD), and foreign agencies (e.g., International Global Change Institute (IGCI), etc.), involved in the development of organic farming in Vietnam (Scott et al.,2009) For instance, the Organic Promotion Program has been developed to train farmers and 181 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 collaborate with local companies to develop an organic production system (Bui and Nguyen,2021) However, it is unfortunate that the increased demand for organic food and supports from government and international agencies has not sufficiently brought good development of organic agriculture in Vietnam Determinants of organic farming adoption Organic farming is known to be more beneficial to the environment, ecosystems, and individual health than conventional farming (Tuomisto et al.,2012;Muneret et al.,2018) Therefore, previous studies have pointed out several factors, such as economic incentives (e.g., subsidies, etc.), psychological factors (e.g., attitude, perception, norms, etc.) and demographic characteristics, influencing farmers’ decisions in adopting organic farming Economic incentives (e.g., subsidizing organic output and input prices) could play a crucial role in encouraging farmers to switch to organic farming (Pietola and Lansink, 2001;Kerselaers et al.,2007;Breustedt et al.,2011) For instance, in their study, the authors found that organic farms with lower returns (i.e., low yields per hectare) are more likely to switch to standard farming (Pietola and Lansink,2001;Breustedt et al., 2011) Thus, in order to make organic farming more attractive, it is substantial to impose agricultural policies that reduce output prices and compensate for income losses to increase farmers’ incentives to convert to organic farming (Pietola and Lansink,2001) Besides the economic incentives, psychological factors, including farmers’ atti- tudes, perceptions and norms, are non-economic factors that could positively impact farmers’ decisions to adopt organic farming For instance, a positive attitude toward organic farming (e.g., farmers believe that organic farming is good for the environment, health, etc.) could facilitate their organic farming investment and management (Lăapple and Van Rensburg,2011;Lăapple and Kelley,2013) Moreover, farmers’ risk perceptions (i.e., perceived risks) associated with the organic farming investment are a barrier to organic farming (Kallas et al.,2010;Sapbamrer and Thammachai,2021) Notably, less risk-averse farmers can tolerate risk situations, such as high input costs, market price fluctuation, and market demand (Sapbamrer and Thammachai,2021) In addition, some studies suggested that the adoption of organic farming is also constrained by social norms and farmers ability (Lăapple and Kelley,2013) For instance, improving social acceptance of organic farming is needed to shift farmers to promote the uptake of organic farming (Lăapple and Kelley,2013) Several studies have suggested that farm size, farming experience and education are factors that could be positively associated with the adoption of organic farming (Lăapple and Van Rensburg,2011;Hoang-Khac et al.,2021;Sapbamrer and Thammachai,2021) For instance, well-educated farmers often have a greater capacity to comprehend and appreciate the benefits of organic agriculture (Hoang-Khac et al.,2021;Sapbamrer and Thammachai,2021) On the other hand, some studies found that farmers’ age results in a negative association with organic farming adoption (Suwanmaneepong et al.,2020; Sapbamrer and Thammachai,2021) For instance, in their study, the authors indicated that older farmers are usually more risk-averse and have less time to invest for the long- term than younger farmers (Sapbamrer and Thammachai,2021) Moreover, younger farmers are often more open minds and more educated about organic farming, and thus they would have more opportunities to assess organic farming technology Data and methodology Data collection Our data were gathered through a survey of household farmers in 31 villages in Northern Vietnam The data collection took place in eight villages in August 2019, 11 villages in November 2019, and 12 villages from December 2019 to January 2020 A total of 586 farmers were interviewed in the survey These villages in the vicinity of Hanoi were chosen because they produced the greatest quantity of agricultural products (vegetables, rice, and fruits) in Northern Vietnam 182 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 Our survey collects comprehensive data on farmers’ production activities (e.g., primary agricultural products, production costs, etc.) Additionally, we collected data on sociodemographic characteristics such as age, gender, farm size, household size, type of residence, individual and household income, health, the highest level of education attained, marital status, number of children in the household, and individual attitudes toward risks, among others Along with socio-demographic characteristics of farmers, we elicited information on several environmental concerns via 15 NEP questionnaires to evaluate farmers’ environmental perceptions (details of the NEP questionnaires are included in TableA1in the Supplementary Materials) (Dunlap et al.,2000) Additionally, several questions about environmental concerns were included to elicit farmers’ opinions, attitudes, and appre-hensions about the environment The following section includes a detailed summary of the statistics in Table1 Descriptive statistics The descriptive statistics of all variables, including dependent variable, explanatory variables and socio-economic characteristics, are reported in Table1 The dependent variable “Adoption of organic farming” is a continuous variable ranging from to This variable is used to capture the percentage of total farming land that farmers devoted to organic agriculture We observe that farmers in the survey areas devoted on average about 17% of their land to organic farming The explanatory variables include three groups of factors, such as social networks, social norms and personal norms Social networks include: “Communication” is a dummy variable taking a value of if a farmer frequently communicated, discussed or talked with other neighborhood farmers and “Neighborhood organic” is dummy variable taking a value of if there is at least one his or her neighborhood farmer was doing organic farming Social norms include: “Injunctive norm” is a dummy variable taking a value of if a farmer believed that doing organic farming is approval by others in the village and “Descriptive norm” is a dummy variable taking a value of if a farmer believed that most farmers in the village are doing organic farming “Personal norm” is a dummy variable taking a value of if a farmer believed that farmers should adopt organic farming to protect the environment and deal with food safety issues The socio-economic control variables include: “Female” is a dummy variable taking a value of if a farmer is a female; “Age” is a log of farmer’s age; “High school” is a dummy variable taking a value of if a farmer graduated from high school, college or university; “Good health” is a dummy variable taking a value of if a farmer claimed that he or she had very good health; “High income” is a dummy variable taking a value of if a household farmer belonged to a high-income group (i.e., average monthly income is higher than 20 million VND); “Farm size” is a log of farmers’ farm size; “Cooperative” is a dummy variable taking a value of if a farmer participating in a farming cooperative; “Rice”, “Vegetables”, “Fruits”, “Coins” and “Others” are dummy variables taking a value of if farmers mainly produced rice, vegetables, fruits, coins or other agricultural products, respectively We observe that farmers’ age was on average 51.3 years old and about 67% of them were female Only 12.6% of farmers graduated from high school or above high school Most farmers in our sample (i.e., about 78%) claimed that they have good or very good health Only 7% of farmers belonged to the high-income group We also observe that 41.6% of farmers participated in farmer’s cooperative, and their farm size was on average 4,221 m2 Farmers in our survey areas mainly produced rice and vegetables rather than Table 1: Summary statistics of survey respondents (N=586) Dependent variable Mean Std.Dev 0.204 0.125 Adoption of organic farming 183 Min Max 0.64 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 0.203 0.402 Neighborhood organic 0.244 0.429 Social norms 0.670 0.470 Descriptive norm 0.542 0.498 Personal norm 0.590 0.492 Control variables 0.667 0.471 Age (yrs) 51.300 11.685 18 74 Age (in log) 3.908 0.256 2.890 4.304 High school 0.126 0.332 Good health 0.78 0.41 High income 0.07 0.25 Farm size (m2) 4,221 7,041 50 70,000 Farm size (in log) 7.797 0.964 3.912 11.156 Cooperative 0.416 0.493 Types of products 0.735 0.441 Vegetables 0.653 0.476 Fruits 0.223 0.416 Coins 0.290 0.454 Others 0.121 0.326 Explanatory variables Social networks Communication Injunctive norm Female Rice Notes: Other agricultural products include coin and other types of products other types of agricultural products, such as fruits, coins or other products We also elicited information on several questions related to environmental concerns via 15 revised New Ecological Paradigm (NEP) questions to help us to identify the individual perceptions toward the environment (details of the NEP questions in TableA1 in the Supplementary Materials) (Dunlap et al.,2000) The total NEP score is the aggregate score of these NEP questions, in which Cronbach’s alpha is equal to 65.45%1 and questions number 2, 4, 6, 8, 10, 12, 14 (even number questions) are reversely coded Cronbach’s alpha is equal to 65.45% in the reliability test, which suggests that 65.45% of the variance in the score is reliable (Cronbach,1951) Econometric model In this section, we discuss the methodologies applied to investigate farmers’ decisions in adopting organic farming Latent Class Cluster Analysis (LCA) is first employed to account for farmers’ heterogeneity based on their attitude toward the environment Then, fractional regression is used to identify the effects of different groups of factors and control variables on farmers’ organic farming adoption 184 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 Regarding the LCA, farmers with similar environmental attitudes could share the same response patterns, and thus they are highly correlated The LCA model assumes each respondent belongs to a latent membership class c The probability that a re- spondent belongs to class c is denoted πc) The probability that a respondent in group g chooses an answer s to question q is defined as πqs|c Since the class membership is unknown, the expectationmaximization algorithm is used to maximize the following log-likelihood function (Lăapple and Kelley,2013): In this study, we adopt the fractional regression model to deal with dependent variable Adopti, which is defined on the closed interval Adopti ∈ [0, 1] The variable Adopti has a fractional nature because it is the percentage investment in organic farm- ing and thus could not take a value less than zero and greater than one (Papke and Wooldridge,1996;Wooldridge,2009;Ramalho et al.,2011) A descriptive introduction of the fractional regression model is generally described in the analysis ofWooldridge (2009) The fractional model with the dependent variable Adopti as a fraction bounded between zero and one has the following structure: E(Adopti|Zi) = H(Ziβ), where Zi represents a set of regressors including explanatory variables, such as “Communication”, “Neighborhood organic”, “Injunctive norm”, “Descriptive norm” and “Personal norm”, and control variables as previous discussed For the logistic link-function H(.) satisfying < H(.) = exp(.) < (Wooldridge,2009), the fractional logistic model can 1+exp(.) be written as follows : eZiβ (3) + eZi β The proposed estimator for β is the Quasi Maximum Likelihood Estimator (QMLE), which maximizes the following Bernoulli log-likelihood function (McCullagh,1983): Σ Σ J J Li (β) = Adopti log[H(Z β)] + (1 − T Ei )log[1 − H(Z β)] (4) i i Since there is the non-linear estimation of the conditional mean, the fractional logit model performs well if there are not many observations at the boundary levels (Ramalho et al.,2011) According to the density plot of our dependent variable Adopti, there is only some small percentage investment in organic farming is left-censored at 0% and no right-censored at 100% (see FigureB1in Supplementary Materials B) Additionally, for robustness check, the estimation results with the Tobit regression model are also reported in Table3(in the Supplementary E(Adopti |Zi ) = 185 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 Materials) The standard errors of the fractional and Tobit regression are estimated with 500 bootstrap replications Results Latent class clusters identified In this study, the revised NEP questions are used to identify the number of clusters The NEP questions are known as the most widely used method to measure individuals’ atti- tudes to pro-environmental behaviors It should be noted that the seven even-numbered items (i.e., NEP questions “2,4,6,8,10,12,14”), if agreed to by a respondent, are meant to represent statements endorsed by the Dominant Social Paradigm (DSP), while the eight odd items (NEP questions “1,3,5,7,9,11,13,15”) if agreed to by a respondent, are meant to reflect the endorsement of the new environmental paradigm (NEP) (Dunlap et al., 2000) Results of the Bayesian Information Criterion (BIC) for LCA with the different number of classes in TableA2(in Supplementary Materials A) suggest that a model with three classes is preferable compared to other models FigureB2(in Supplemen- tary Materials B) represents the conditional item response probabilities for each class In particular, there are three distinct classes identified by LCA: Class (“Concern”) represents farmers who have strong environmental concern since most of them stated agree and strongly agree with the eight odd NEP items and disagree with the seven NEP items, Class (“Moderate”) represents farmers who are a moderate pro-environmental concern since most of the farmers express less concern about the environment comparedto those in Class 2, and Class (“Indifferent”) consists of farmers who are indifferent to environmental concern since most of them gave “unsure” responses to the 15 NEP items In order to ensure the robustness of the results, we also perform LCA with two classes (see FigureB3(in Supplementary Materials B) The estimation results of fractional regression models with two classes are reported in TableA3(in Supplementary Materials A) Looking at the estimation results of two and three classes, we observe that the LCA with three classes helps us better capture the heterogeneity compared to those with two classes Furthermore, according to the LCA in both FigureB2andB3, we observe that a majority of farmers in our survey areas expresses a strong environmental concern The descriptive statistics of survey respondents in three LCA classes are reported in Table2 Table 2: Summary statistics of survey respondents in different LCA classes Dependent variable Concern Mean (N=350) Std.Dev Indifferent Mean (N=110) Std.Dev Moderate Mean (N=126) Std.Dev 0.254 0.437 0.151 0.120 0.222 0.136 0.191 0.393 0.136 0.344 0.293 0.457 Neighborhood organic Social norms 0.228 0.420 0.281 0.451 0.163 0.129 Injunctive norm 0.682 0.466 0.572 0.496 0.722 0.449 Descriptive norm 0.560 0.497 0.454 0.500 0.571 0.496 Personal norm Control variables 0.691 0.492 0.581 0.495 0.595 0.492 Female 0.720 0.449 0.745 0.437 0.452 0.499 Age (yrs) 51.45 11.58 51.28 11.79 50.90 11.95 Adoption of organic farming Explanatory variables Social networks Communication 186 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 High school 0.100 0.300 0.118 0.324 0.142 0.351 Good health 0.157 0.364 0.172 0.379 0.214 0.411 High income 0.120 0.325 0.109 0.313 0.158 0.366 Farm size (m2) 4081 6787.6 3636 5705.2 5120 8593.3 Cooperative Types of products 0.414 0.493 0.390 0.490 0.444 0.498 Rice 0.734 0.442 0.727 0.447 0.746 0.437 Vegetables Fruits Coins 0.642 0.217 0.288 0.479 0.412 0.453 0.654 0.154 0.263 0.477 0.363 0.442 0.682 0.301 0.317 0.467 0.460 0.467 Others 0.494 0.500 0.409 0.493 0.579 0.495 The summary statistics in Table2suggested that farmers in the “Concern” class place more weight on organic farming than the other two classes Farmers in the “Mod- erate” class communicate more frequently with other organic farmers than other classes In contrast, the “Indifferent” class has a higher probability to live close to at least one neighbor doing organic farming Moreover, farmers, who concern about the environment (i.e., “Concern” and “Moderate” classes), strongly believe that most of the villagers is doing organic farming and adoption of organic farming will also be approved by others in the village Furthermore, we also observe that the“Concern” class expresses a strong perception of adopting organic farming to protect the environment and deal with food safety issues Estimation results In this section, we discuss the estimation results of the impacts of different social factors on farmers’ decisions in adopting organic farming The estimation results are reported in Table3 Model (1) in Table3shows results of a pooled sample, while Model (2), (3) and (4) in Table3present the results of the sub-samples divided into three groups, including “Concern”, “Indifferent” and “Moderate” groups Results of Tobit estimation with three LCA classes are also reported in TableA3(in Supplementary Materials A) However, we observe that the estimation results of Tobit models are close to those of the fractional regressions Moreover, a likelihood-ratio Chow test between the full sample results and three LCA classes, χ2(17) = 50.65 with p-value < 0.001, suggests that three LCA classes should not be pooled but rather assessed independently Thus, our interpretation will base on the results of the fractional regression with three LCA classes reported in Table The results in Table3show that “Neighborhood organic” (i.e., a presence of or- ganic farming neighbors) is positive and statistically significant in all three Models (2), (3) and (4) This result suggests that the presence of organic farming neighbors could significantly influence farmers’ decisions to convert their lands to organic agriculture This result is also in line with the existing literature that social network (i.e., connections and interaction with other individuals) helps individuals to share their knowledge and envi- ronmental concern with others and thus could drive them toward more pro-environmental behaviors (Ataei et al.,2019;Vu et al.,2020;Nguyen-Van et al.,2021) For instance, studies showed that farmers who behave following their neighbors’ expectations and with more availability of information in their community network are more inclined to adopt organic agriculture (Wollni and Andersson,2014) Moreover, we observe that “Communication” (i.e., frequency of communication with other organic farming neighbors) is positive and significant only in the “Moderate” group This result is straightforward since our descriptive statistics indicate that farmers in the “Moderate” group communicate more frequently with other organic producers in their village than those in other groups Thus, a high frequency of communication via interpersonal channels (i.e., with other farmers) could help farmers to express their opinions, organic farming method and their 187 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 preferences toward organic agriculture to other farmers (Hall and Rhoades,2010;Nguyen-Van et al.,2021) Therefore, it helps influence their decisions to adopt or not organic farming Table 3: Estimation results of the fractional regressions with three LCA classes Full sample Three LCA classes All farmers (1) Variables Social networks 0.995∗ ∗ ∗ Concern (2) Indifferent (3) 0.675 Moderate (4) 0.927 2.084∗ ∗ Communication Neighborhood organic (0.308) (0.464) (0.837) (0.841) 2.072∗ ∗ ∗ 1.913∗ ∗ ∗ 2.666∗ ∗ 4.198∗ ∗ ∗ (0.304) (0.467) (1.243) (0.864) Social norms Injunctive norm 0.066 0.242 (0.349) Descriptive norm ∗∗∗ 0.841 (0.496) 0.779 (0.325) Personal norm Control variables 0.686∗ ∗ 1.909 ∗ ∗ (0.247) 0.978∗ ∗ -1.003 (1.751) (0.684) -0.145 2.682∗ ∗ ∗ (1.418) (0.855) 2.266∗ -0.968 (0.341) (0.449) (1.289) (1.055) -0.422 -0.234 -1.610∗ -1.442 (0.876) (1.030) Female Age (0.309) ∗∗ -0.912 High school Good health High income (0.451) -1.015 (0.446) (0.916) -0.169 -0.427 -1.764 (0.389) (0.593) (1.325) 0.545 (0.347) -0.639 0.705 0.068 -0.201 ∗∗ (1.230) -2.142 ∗∗ (0.126) (1.056) 0.515∗ ∗ ∗ -0.925∗ (0.197) (0.537) (0.145) 0.336 0.548 (0.528) (0.319) Farmer’s cooperative -1.916 ∗∗ (0.303) ∗∗ Farm size ∗∗ 0.162 0.374 (0.290) (0.392) (0.994) -0.271 -0.286 -0.671 (0.338) (0.828) (0.627) -1.832∗ ∗ (0.779) -0.227 (0.886) 0.366 (0.755) -2.798∗ ∗ (1.222) -0.350 (0.290) 0.766 (0.652) Types of products Rice Vegetables -0.071 0.137 188 -0.025 -1.051 (0.951) -1.246 Kỷ yếu Hội thảo Khoa học ngành Kinh tế năm 2022 (0.296) Fruits 0.433 (0.410) Coins Intercept Nb of observations Class shares 0.259 2.238∗ ∗ 0.092 (0.316) (0.830) (0.415) 0.544 (1.117) 1.223 (0.328) (0.439) (1.344) -0.624 -4.283 -5.487 (2.627) (3.592) (6.923) 586 - 350 110 58.40% 19.57% (0.973) 0.578 (0.795) -0.791 (0.899) 7.821 (5.975) 126 22.03% Notes: Standard errors with 500 bootstrap replications are in parentheses ∗p