Social networks, rice value chain participation and market performance of smallholder farmers in Ghana

Abdul‐Rahaman, Awal; Abdulai, Awudu Article — Published Version Social networks, rice value chain participation and market performance of smallholder farmers in Ghana African Development Review Provided in Cooperation with: John Wiley & Sons Suggested Citation: Abdul‐Rahaman, Awal; Abdulai, Awudu (2020) : Social networks, rice value chain participation and market performance of smallholder farmers in Ghana, African Development Review, ISSN 1467-8268, Wiley, Hoboken, NJ, Vol. 32, Iss. 2, pp. 216-227, https://doi.org/10.1111/1467-8268.12429 This Version is available at: https://hdl.handle.net/10419/230240 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ Afr Dev Rev. 2020;32:216–227.216 | wileyonlinelibrary.com/journal/afdr DOI: 10.1111/1467-8268.12429 ORIGINAL ARTICLE Social networks, rice value chain participation and market performance of smallholder farmers in Ghana Awal Abdul‐Rahaman 1 |Awudu Abdulai 2 1 Department of Agribusiness Management and Finance, Faculty of Agribusiness and Applied Economics, University for Development Studies, Nyankpala Campus, Tamale, Ghana 2 Department of Food Economics and Consumption Studies, University of Kiel, Kiel, Germany Correspondence Awudu Abdulai, Department of Food Economics and Consumption Studies, University of Kiel, Johanna‐Mestorf str. 5, 24118 Kiel, Germany. Email: [email protected] Abstract This paper examines the impact of rice value chain participation and social networks on smallholder farmers' market performance outcomes (paddy price, quantity of paddy traded, and net returns), using data from a recent survey of 458 smallholder rice farmers in northern Ghana. We employed a treatment effects model to account for potential selection bias associated with observable and unobservable factors. The empirical results reveal that smallholder farmers' participation in a rice value chain is associated with increased paddy price, quantity traded, and net returns. We also find that value chain participation decisions and market performance are positively and significantly influenced by social networks. The empirical results also suggest that sex, farm size, mobile phone ownership, and access to credit significantly increase paddy prices, quantity traded, and net returns of smallholder rice farmers in the value chain. 1|INTRODUCTION In the past 2–3 decades, agricultural value chains in developing countries have experienced dramatic structural transformation, driven by several factors such as population growth, rising urbanization, increasing consumer incomes, and varying consumer dietary requirements (Henderson & Isaac, 2017; Mensah, Adu, Amoah, Abrokwa, & Adu, 2016;Swinnen& Kuijpers, 2019). While the value chain transformation is considered important in reducing rural poverty, improving food and nutrition security, and ensuring overall economic growth, smallholder farmers' inclusion in these chains still remains a major challenge in developing countries. This is largely due to lack of institutional and infrastructural support, inadequate resources for effective value chain coordination, high transaction costs associated with accessing inputs and markets, and other challenges related to accessing services such as extension, finance, and transportation, all of which impact farm production and market performance (Trienekens, 2011; Verdier‐Chouchane & Boly, 2017). Developing country governments, non‐governmental organizations (NGOs), and the private sector have increasingly recognized agricultural value chain development as an important area of donor interventions, and a centerpiece of agricultural development policies (Humphrey & Navas‐Aleman, 2010). Motivated by concerns for agribusiness development, value chain development interventions do not only focus on strengthening the capacities of value chain actors, but also the institutions and enabling policy environment that ensure effective and efficient coordination and competitiveness of these chains (Anyanwu & Kponnou, 2017; Humphrey & Navas‐Aleman, 2010). They facilitate vertical linkages, and foster governance of relationships between smallholder farmers and agribusinesses through written or ------------------------------------------------------------------------------------------------------------------------------------------- This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. African Development Review published by John Wiley & Sons Ltd on behalf of African Development Bank (AFD) verbal contracts within value chains for improved welfare gains (Devaux, Torero, Donovan, & Horton, 2018). Smallholder farmers have been recognized as important actors for the diffusion of value chain innovations such as information and technology (Ramirez, Bernal, Clarke, & Hernandez, 2018). With the underlying assumption that the behavior of social network members influences farmers' decision‐making with direct implications on welfare outcomes (Mano, Yamano, Suzuki, & Matsumoto, 2011; Wydick, Hayes, & Kempf, 2011), the important role of farmers' social networks in improving value chain efficiency and rural economic transformation needs to be highlighted in the empirical literature. The concept of social networks emphasizes the connections among individuals (e.g., farmers) through which goods and services, money, and information flow (Mano et al., 2011). Within the context of value chains, social networks focus on horizontal and vertical relationships that exist among value chain actors (Trienekens, 2011). Horizontal social networks reflect cohesive social relationships that promote collective action for successful inclusion in value chains (Ramirez et al., 2018). They can draw upon their social capital to strengthen vertical relationships with buyers and other actors within the value chain (Bijman, Omta, Trienekens, Wijnands, & Wubben, 2006). More importantly, smallholder farmers organized into a strong social network can benefit from improved access to credit and extension services, increased bargaining power, exchange of input and output price information, and buyer quality requirements through participation in such formalized value chains (Ma, Abdulai, & Goetz, 2018). This is in line with the increasing calls for smallholder collective action (e.g., collective marketing) to improve farmers' bargaining power, while enhancing value chain efficiency (Bernard, Spielman, Taffesse, & Gabre‐Madhin, 2010; Dillon & Dambro, 2017). In addition, smallholder farmers often rely on their social connections and goodwill with other farmers in their communities to enjoy cost advantages associated with production and sales, as well as achieve optimum benefit from participating in agricultural value chains (Ramirez et al., 2018). Such social connections are based on agreed‐upon norms, and established trust, as well as facilitate sharing of value chain information, skills, labor, and financial resources. Some recent studies have examined the welfare gains associated with smallholder farmers' participation in value chains in developing countries (e.g., Henderson & Isaac, 2017;Ma&Abdulai,2017;Maertens&VandeVelde,2017;Michelson,2013). Michelson (2013) found that value chain participation is associated with increased productive asset holdings among smallholder farming households in Nicaragua. Other studies also found that smallholder participation in value chains significantly improves prices received, farm profit, and gross income (e.g., Ma & Abdulai, 2017), farm yields, household income, and net farm income (e.g., Maertens & Vande Velde, 2017), and land and labor redistribution (e.g., Henderson & Isaac, 2017). However, there is also increasing interest in the role of social networks on the economic behavior and decision‐making of smallholder farmers in developing countries. For example, the study by Bandiera & Rasul (2006) reveals that social networks influence smallholder farmers' sunflower adoption decisions in northern Mozambique. Moreover, the role of social networks in improving households' access to credit, income diversification, and non‐farm employment decisions has also been revealed by past studies (e.g., Kinyondo & Kagaruki, 2019;Manoetal.,2011;Wydicketal.,2011). This study examines the implications of social networks and value chain participation on smallholder farmers' market performance, using recent survey data collected by the authors from northern Ghana. In particular, this study makes three contributions to the empirical literature. First, we explore the role of social networks and other farm and household characteristics in influencing smallholder farmers' participation in a rice value chain. Second, we examine how social networks and rice value chain participation influence smallholder rice farmers' market performance outcomes: prices received, quantity traded, and net returns. Finally, we examine whether rice value chain participation effects vary with farm size. Insights from this study could be relevant from a development policy perspective. The study focuses on the rice sub‐sector, because value chain development efforts to revamp the cereal staples sub‐sector are already in progress in Ghana (Donkor, Matthews, & Ogundeji, 2018). The government, donor agencies, and the private sector have collaboratively rolled out a number of value chain development interventions with the objective of ensuring smallholder market competitiveness, and rural economic transformation. These interventions promote value chain development by linking smallholder farmers with large agribusinesses and other produce buyers for output transactions. Findings from this study can inform policy on the design and implementation of value chain development programs for the benefit of smallholder farmers in Ghana. The rest of this paper is organized as follows: Section 2describes the conceptual framework that guides the empirical analysis, followed by specification of the empirical model in Section 3. Section 4presents the data and descriptive statistics of the variables used in the analysis. Section 5presents and discusses the empirical results, while conclusions and policy implications are provided in the final section. ABDUL‐RAHAMAN AND ABDULAI | 217 2|CONCEPTUAL FRAMEWORK 2.1 |Rice value chain participation decision and the role of social networks In this section, we explore the effects of social networks on rice value chain participation by smallholder farmers. Social network effects on individual behavior include endogenous effects, exogenous (contextual) effects, and correlated effects. The influence of a network member's behavior on the individual's decision‐making, such as value chain participation, is termed an endogenous network effect, while the effect that a network member's specific characteristics (e.g., age, education, sex etc.) may have on the individual's decision to participate in a value chain is referred to as an exogenous (contextual) effect (Manski, 2000). Correlated network effects stem from controlling for unobservable invariant characteristics between network points in a community or district that may influence a farmer's value chain participation decision (Mekonnen, Gerber, & Matz, 2018). It is accounted for in an empirical analysis by controlling for location dummies in the model. To examine the effects of social networks on rice value chain participation, we assume that farmers make binary decisions whether to participate or not to participate in a rice value chain, depending on his/her characteristics. This decision is determined by comparing the expected utility from the participation, U ( ) iP , and the expected utility from non‐participation, U () iN. Intuitively, a farmer decides to participate in a value chain if the utility difference VC ( ) i⁎is positive, that is, V CUU=−>0 iiPiN ⁎ , which implies that the utility the farmer derives from participating in the value chain outweighs the utility derived from non‐participation. However, V C i ⁎ is a latent variable, and cannot be directly observed. In this context, what is observed is the actual decision by the farmer to participate in a value chain, VC . Therefore, we specify it as a function of observable farm, household, and social network characteristics as follows: V CDϕηVC VC VC Zδ=+ +,= 1if >0 0if 0 , iiiiii i ⁎ ⁎ ⁎ ⎧ ⎨ ⎩≤ (1) where V C i is a binary indicator variable that equals one if a farmer participates in a rice value chain, and zero otherwise; Z is a vector of farm and household characteristics believed to influence rice value chain participation decision. These include age, education, sex, farm size, distance to market, bicycle ownership, road status, mobile phone ownership, access to credit, and market perception; Didenotes variables representing social networks; δ is a vector of parameters capturing the direct effects of exogenous observable characteristics in Z while φ measures the endogenous, contextual, and correlated network effects on farmer's decision to participate in the rice value chain; and η i is an error term, with zero mean and variance σ 2 . The choice of the explanatory variables in this study was based on existing literature, and observations from the field survey. Building on the existing literature on the effects of social networks on technology adoption (e.g., Bandiera & Rasul, 2006), and on modern supply chain participation (e.g., Herforth, Theuvsen, Vasquez, & Wollni, 2015; Ramirez et al., 2018), we include in our analysis a variable representing farmer's horizontal social network relationship, which is assigned a value of one if a farmer belongs to such a network, and zero otherwise. In addition, a variable that captures the number of value chain participants in a farmer's social network is included in the analysis. These social network variables measure the existence of endogenous effects on value chain participation. However, measurement of these effects poses simultaneity issues, resulting from the fact that the behavior of an individual is influenced by the mean behavior of the group, who also in turn influences the group's behavior. Manski (1993) refers to this identification problem as a reflection problem. Moreover, smallholder farmers are normally organized horizontally into farmer groups, and taken through capacity building training to become a strong and cohesive social network group for participation in value chains. Conversely, agribusiness companies and other produce buyers normally prefer to engage with smallholder farmers in the form of strong and cohesive groups to reduce transaction costs associated with having to aggregate paddy from individual farmers. Therefore, farmers may decide to be members of the horizontal social networks to be able to participate in the value chain, making both decisions jointly determined. We also argue that a farmer's network members who are already participants in a value chain can serve as sources of useful information, and potential avenues for sharing valuable experiences about the marketing opportunities in a value chain. The information and experience sharing between farmers and their network members are expected to influence the decisions of this category of farmers to participate in the value chain. On the other hand, produce buyers can also rely on participants for the information and recommendation of their network members for inclusion in the value chains. Similarly, access to credit variable in Z may 218 | ABDUL‐RAHAMAN AND ABDULAI also pose potential endogeneity problems in the value chain participation equation. In the study area, government and NGOs who facilitate smallholder participation in value chains, also facilitate farmers' access to credit through linkages with financial service providers. In that case, some farmers may decide to participate in a value chain to be able to access credit to expand their farming operations, and to benefit from a guaranteed market. This makes the decisions to participate in a value chain and to access credit jointly determined. To address these issues, some approaches have been suggested in the literature. Manski (2000) suggests the introduction of dynamism to the model whereby an individual's behavior is influenced by the lagged behavior of his/her network instead of contemporaneous values of mean behavior of the group. Another approach is to use instruments to address these challenges (Manski, 2000). Due to data limitation, we use the latter approach in the present study, which is a two‐stage approach clearly outlined in Wooldridge (2015). The procedure and first‐stage regression results are not reported due to space limitation but are available upon request. Following Mekonnen et al. (2018), we control for contextual network effects by averaging the values of observable exogenous characteristics of the farmers in the sample, based on the subsamples drawn from each study community. Based on our data, the exogenous characteristics used are the averages of age, education, sex, and farm size. We also control for correlated network effects by controlling for district dummies in the model. 2.2 |Impact of social networks and rice value chain participation As indicated previously, we also examine the impact of social networks and rice value chain participation on smallholder rice farmers' market performance. The market performance outcome measures considered in this study include paddy price received, quantity traded, and net returns. To link value chain participation decision to the market performance outcomes, we assume a linear function between a vector of the outcome measures and a vector of farm, household, and social network characteristics X () i, and a dummy variable representing value chain participation VC () i , specified as: θVC βμYX=+ + , ii i i(2) where Yiis a vector of outcome variables: θand β are parameters to be estimated; and μ i is the error term. Farmer's value chain participation decision involves self‐selection, which is influenced by unobservable factors such as farmers' risk preferences, motivation, and innate skills. These factors may also influence the market performance outcomes leading to potential selection bias ( ημcorr( , ) 0 ii ≠ ), and using ordinary least squares method would generate biased estimates. However, propensity score matching (PSM) method accounts for selection bias from only observable factors. In the present study, we use treatment effects model in the empirical analysis (Cong & Drukker, 2000), which accounts for observable and unobservable factors. 3|EMPIRICAL SPECIFICATION 3.1 |Treatment effects model In this context, the treatment effects model estimates the factors influencing smallholder rice farmers' decisions to participate in a rice value chain, and their impacts on paddy price received, quantity traded, and net returns. Aside accounting for selection bias, the model provides direct marginal effect of value chain participation on market performance. Following Cong and Drukker (2000), we specify the model as: E VC θβEμVC θβρσ ϕZδ Zδ YX X( | =1)= + + ( | =1)= + + () Φ() ii iiημ ημ i i (3) E VC θEμVC θρσ ϕZδ Zδ YX X(| =0)= + (| =0)= −() 1−Φ() , ii iiημ ημ i i (4) ABDUL‐RAHAMAN AND ABDULAI | 219 where ϕ(. ) is the standard normal density function, and Φ(. ) denotes the standard normal cumulative distribution function. The ratio of ϕ(. ) and Φ(. ) is referred to as the inverse Mills ratio. θand β are vectors of parameters to be estimated; σ η μ is the covariance between the two error terms, η μρ,; η μ is the correlation coefficient, and an indicator of selection bias on unobservable factors. The average treatment effects (ATE) of value chain participation on the outcomes for sample Ncan be computed as the difference between the expected outcome from participation (Equation 3), and the expected outcome from non‐participation (Equation 4), specified as: ATE NEVC EVCYY=1(| =1)−(| =0) . i N ii =1 ⎡ ⎣⎤ ⎦ ∑ (5) The model is identified using the variable representing farmer's perception of high market demand for paddy in the previous year as instrument, measured as dummy, where farmer's perception of high market demand for paddy in the previous year is assigned a value of one, and zero otherwise. A simple falsification test (Di Falco, Veronesi, & Yesuf, 2011) reveals that the instrument is valid. The test results are not presented in the interest of brevity, but are available upon request. 4|DATA AND DESCRIPTIVE STATISTICS The data used in this study were collected from a recent farm household survey (June–August, 2016) conducted by the authors in five selected districts of northern Ghana; Tolon, Kumbungu, Sagnarigu districts, Savelugu Nanton municipal, and Tamale metropolis. We employed a multi‐stage sampling approach in drawing our sample for the study. First, we used a purposive sampling method to select the five study districts because of their geographic accessibility, and the intensity of rice production in these districts. About two to three communities from each district were randomly selected, based on the number of communities in each district. Finally, we used random matching within sample, whereby at least 20 households were randomly selected in each community. Each household was then matched with five farmers randomly drawn from the community sample. In total, we sampled 458 smallholder rice farmers, comprising 206 value chain participants and 252 nonparticipants. In the context of this study, value chain participants are smallholder farmers who are beneficiaries of the ongoing rice value chain development project (USAID Feed the Future [FtF]) in northern Ghana, and have established contractual relationships (written or verbal) with produce buying or processing companies under the facilitation and coordination of officials of the project (formalized value chain) (Birthal et al., 2016; Seville, Buxton, & Vorley, 2011). These farmers have received capacity building and input support from the project, and have also successfully supplied paddy to these produce buying and processing companies for at least the past three years. On the other hand, nonparticipants are smallholder farmers who produce and supply paddy in the traditional or open market (informal value chain) (Birthal et al., 2016; Seville et al., 2011). These farmers normally supply to traders/aggregators, who do not usually hold farmers to quality and packaging standards. Both categories of farmers were then engaged in face‐to‐face interviews, using a structured questionnaire. The social network variables captured include horizontal social networks and number of value chain participants in a farmer's network. In addition, information was gathered from farmers on the number of farmers in their social networks who are also participants in the rice value chain. This was based on whether resources such as credit, labor, and/or land, farming and marketing information have ever been exchanged between the farmer and the network members. In addition, we considered whether they are relatives, friends, neighbors (farm‐plots or residential), belong to the same religion, or ever visited each other. Other information gathered include farm and household characteristics, asset ownership, production, and marketing activities related to the 2015 growing season. Table 1presents the variable definition and statistical difference between participants and nonparticipants. The outcome variables include average selling price of paddy per kilogram, quantity of paddy traded in kilograms, and net returns. We measured net returns as the difference between value of rice output and variable input costs per hectare. Table 1shows that value chain participants sold higher quantities of paddy rice, received higher paddy price, and generated higher net returns than nonparticipants. In addition, participants are mostly members of horizontal social networks, and have a higher number of network members who are also value chain participants. Table 1also reveals that value chain participants constitute a higher proportion of farmers who accessed enough credit and/or are not credit constrained than nonparticipants. In this study, the access to credit variable is constructed in the context of whether the farmer is credit constrained or not. It is captured as a dummy variable, whereby one is assigned to a farmer who did not need credit, or the one who needed credit, applied for it and received the required 220 | ABDUL‐RAHAMAN AND ABDULAI amount (not credit constrained). On the other hand, zero is assigned to a farmer who is credit constrained. This group of farmers include those that needed credit, but did not apply for it, or applied for it and did not receive the required amount, or had their credit applications rejected (Jappelli, Pischke, & Souleles, 1998). 5|EMPIRICAL RESULTS AND DISCUSSION 5.1 |Social network and other factors influencing rice value chain participation decisions The results of the factors influencing farmers' decisions to participate in a rice value chain are presented in Table 2. The coefficients of the residuals predicted from the first‐stage regression for the potentially endogenous variables such as TABLE 1 Variable definition and differences in characteristics of farmers by rice value chain participation Variable Definition Participants Nonparticipants Diff. (t‐stat.) Mean SD Mean SD Age Age of respondent (years) 39.29 11.94 35.97 11.22 3.05*** Education Education of respondent (years) 3.09 4.60 2.40 4.20 1.66* Sex 1 if farmer is male, 0 otherwise 0.89 0.30 0.87 0.33 0.83 Farm size Size of farm (hectares) 1.19 1.27 1.10 1.24 0.75 Distance to market Distance to market (km) 7.20 4.37 6.06 3.76 2.99*** Bicycle 1 if a farmer owns bicycle, 0 otherwise 0.69 0.46 0.71 0.45 −0.37 Road status 1 if market road is motorable, 0 otherwise 0.81 0.39 0.66 0.47 3.49*** Mobile phone 1 if farmer owns mobile phone, 0 otherwise 0.56 0.49 0.36 0.48 4.21*** Access to credit 1 if farmer is not credit constrained, 0 otherwise 0.53 0.50 0.30 0.45 5.17*** Market perception Farmer perception of market demand (1 = high, 0 = low) 0.50 0.50 0.22 0.41 6.47*** Horiz. social network 1 if farmer is member of HSN, 0 otherwise 0.76 0.42 0.15 0.35 16.85*** VCP in farmer's network No. of value chain participant farmers in network 7.89 13.91 2.25 5.29 5.93*** Average age Average age of farmers in the sample 37.95 2.92 36.97 3.19 3.40*** Average education Average value of education of farmers in the sample 2.00 0.86 2.04 0.95 −0.48 Average sex Average value of sex of farmers in the sample 0.87 0.09 0.88 0.10 −0.54 Average farm size Average value of farm size of farmers in the sample 1.12 0.49 1.15 0.49 −0.72 Tolon 1 if farmer is located in Tolon district, 0 otherwise 0.24 0.43 0.20 0.40 1.05 Kumbungu 1 if farmer is located in Kumbungu district, 0 otherwise 0.23 0.42 0.24 0.43 −0.32 SaveluguNanton 1 if farmer is located in Savelugu Nanton Municipal, 0 otherwise 0.12 0.32 0.26 0.44 −3.89*** Price Average selling price of paddy rice (GH¢/kg) 1.33 0.35 1.11 0.12 9.22*** Quantity sold Quantity of paddy rice sold (kg/ha) 1,191.27 800.63 820.17 695.69 5.305*** Net returns Gross revenue from paddy production minus input cost (GH¢/ha) 1,057.95 1157.21 453.31 667.36 6.99*** Sample size 206 252 Note: *, *** represent significance at 10% and 1% levels, respectively; GH¢ is Ghanaian currency (US$1 = GH¢ 4.19); SD, standard deviation. ABDUL‐RAHAMAN AND ABDULAI | 221 horizontal social network relationship, number of value chain participants in farmer's network and access to credit are not significantly different from zero, suggesting that these variables have been consistently estimated (Wooldridge, 2015). As shown in Table 2, the coefficients of horizontal social network and number of value chain participants in farmer's network are positive and significantly different from zero, suggesting that farmers who are members of a horizontal social network, and those with a higher number of value chain participants as network members are more likely to participate in a value chain. This finding presents evidence of the role of social network externalities in rice value chain participation. The coefficients of all the average exogenous characteristics of the farmer's peers, with the exception of sex, were not statistically different from zero, suggesting absence of contextual effects. This means that farmer's rice value chain participation decision is not correlated with the exogenous characteristics of his network members in the sample. Similarly, we did not find evidence of correlated network effects in our results, as revealed by the statistically insignificant effects of the district dummies on rice value chain participation decisions. Other factors influencing value chain participation decisions include access to credit, distance to market, and mobile phone ownership. As shown in the results, farmers with access to sufficient credit, and who are not credit constrained are more likely to participate in value chains, as revealed by the positive and statistically significant effect of this variable on value rice chain TABLE 2 Factors influencing smallholder farmers’rice value chain participation decisions Variable Value chain participation Coefficient Standard Error Constant −11.624*** 3.407 Horizontal social network 1.530*** 0.220 VCP in farmer's network 0.043*** 0.012 Average age 0.036 0.043 Average education 0.097 0.133 Average sex 5.310*** 2.040 Average farm size 0.222 0.229 Age 0.010 0.012 Education 0.016 0.020 Sex 0.138 0.314 Farm size 0.049 0.068 Distance to market 0.051** 0.024 Bicycle −0.006 0.205 Road status 0.111 0.354 Mobile phone 0.072** 0.022 Access to credit 0.454*** 0.164 Tolon 0.243 0.415 Kumbungu −0.386 0.333 Savelugu Nanton 0.733 0.575 Market perception 0.712*** 0.170 Residual (HSNR) −3.298 2.995 Residual (VCPN) 0.011 0.036 Residual (Access to credit) −0.736 2.061 Sample size 458 Note: *, **, *** represent significance at 10%, 5%, and 1% levels, respectively. 222 | ABDUL‐RAHAMAN AND ABDULAI participation. Farmers with access to credit from financial institutions participate in value chains to ensure guaranteed market for their paddy and probably timely credit repayment. 5.2 |Impact of social networks and rice value chain participation on market performance The estimation results of the impact of value chain participation and social networks on paddy price, quantity traded, and net returns are presented in Table 3. As can be observed, the correlation coefficient ρ ε μ is found to be negative and significant in all the model specifications, suggesting the presence of selection bias due to unobservable factors, which means that farmers with below average paddy price received, quantity traded, and net returns have a higher probability of participating in the rice value chain. This finding is consistent with other recent studies that participation in value chains tends to benefit smallholder farmers in developing countries (e.g., Ma & Abdulai, 2017; Michelson, 2013). TABLE 3 Impact of value chain participation and social networks on market performance Variable Paddy price Quantity traded Net returns Coefficient SE Coefficient SE Coefficient SE Constant 0.859*** 0.124 5.043*** 0.754 5.222*** 1.537 VC participation 0.120*** 0.021 0.612*** 0.181 0.863** 0.408 Horizontal social network relationship 0.011** 0.005 0.771*** 0.113 0.260** 0.140 VCP in farmer's network 0.007** 0.004 0.018*** 0.006 0.062*** 0.006 Average age 0.003 0.002 0.024* 0.013 0.019 0.026 Average education 0.009 0.006 0.064* 0.037 0.013 0.076 Average sex 0.032 0.087 0.267 0.529 0.302 1.077 Average farm size 0.000 0.011 0.215*** 0.071 0.140 0.147 Age 0.003 0.010 0.003 0.002 0.004** 0.002 Education 0.001 0.001 0.009 0.006 0.010 0.013 Gender 0.004 0.017 0.319*** 0.100 0.592*** 0.201 Farm size 0.008** 0.003 0.228*** 0.023 0.072*** 0.026 Distance to market −0.001 0.001 −0.005 0.007 −0.012 0.015 Bicycle 0.000 0.011 −0.014 0.065 0.079 0.133 Road status −0.014 0.010 0.039 0.064 −0.039 0.131 Mobile phone 0.009** 0.004 0.029** 0.013 0.071** 0.029 Access to credit 0.008*** 0.003 0.013* 0.060 0.354*** 0.122 Tolon 0.028 0.017 0.147** 0.010 0.584*** 0.217 Kumbungu 0.039** 0.015 0.346*** 0.092 1.168*** 0.187 Savelugu Nanton 0.733 0.575 0.151 0.104 0.595*** 0.215 ath( ρ ε μ )−0.222*** 0.012 −0.359*** 0.100 −0.211*** 0.024 ρ ε μ −0.218*** 0.011 −0.344*** 0.106 −0.208*** 0.021 ln( σ )−2.340*** 0.034 −0.576*** 0.022 0.120*** 0.037 Wald test ( ρ = 0 εμ ) 12.26***, p rob = 0.000 15.06***, p rob = 0.000 25.82***, p rob = 0.000 Sample size 458 458 458 Note: *, **, *** represent significance at 10%, 5%, and 1% levels, respectively; SE, standard error. ABDUL‐RAHAMAN AND ABDULAI | 223