Building Community-Based Social Capital by Enhancing Individual Social Capital: The Case of Farmers in Turkey’s Konya Region

Abstract

Social capital refers to the formation of relationships, or working collectively, to achieve a common outcome. The objective of the current study is to determine whether community-based agricultural policy initiatives can succeed in the absence of consistently strong levels of individual farmer social capital behaviors. Targeting farmers from Turkey’s Konya region who also took part in a household survey, we present findings from a hypothetical field experiment of how farmers would allocate wheat endowment during a drought. We found that farmers allocated 48% of wheat endowment to social capital choices. Our results indicate that the more a farmer relies on himself or herself and resources available outside of the community, the less likely it is that the farmer will invest in community efforts.

1. Introduction

Social interaction among farmers has been widely incorporated into analysis to predict farmer behavior [1,2,3,4]. Most prior studies investigated social interaction as a means of exchanging information and learning from others, but those studies did not examine the allocation of endowments to improve relationships and community-based social capital [1,2,5,6,7,8]. Modeling social interaction through the allocation of endowments sheds light on farmers’ willingness to build and sustain social relationships, which can influence their own decisions [3,4,9,10,11]. For example, a higher presence of social capital can mitigate common societal problems, such as the free rider issue in voluntary provision of public goods and the tragedy of the commons [9,12,13,14,15]. A gap in the literature is understanding the role of the individual in contributing to community-based needs and initiatives. For example, Cascioli et al. [16] investigated how social, human, and institutional capitals influence the level of waste sorting. They did not investigate how changes in exogenous factors influence levels of capitals. While they find that capitals influence community efforts, they are unable to indicate strategies that drive higher levels of the capitals themselves. The objective of the current study is to determine whether community-based agricultural policy initiatives can succeed in the absence of consistently strong levels of individual farmers’ social capital, and we model factors contributing to individual farmer social capital levels that impact the individual’s choice to enhance community social capital.

The current study contributes to the literature by measuring the existence of farm-level social capital. Pairing a farmer social capital experiment with a household survey also allows the study to identify socio-economic factors that influence each motive at the farm level. To our knowledge, this is the first study using the Robison et al. [10] approach of measuring farm-level social capital and analyzing socio-economic factors that affect the existence and strength of social capital motives. Government program staff and policy makers can refer to this study while developing policies meant to incentivize sustainability-minded farmer behavior through community involvement and farmer collaboration. We believe we are the first to tie together opportunities to influence individual farmer social capitals and spill-over into community needs and initiatives.

Previous studies identified the need for regional collaboration among farmers to address common environmental and natural resource problems, such as water pollution and loss of biodiversity [17,18,19]. Golfinopoulos and Kouparou [20] describe the role of government policies in creating an environment incentivizing individuals to converse and ultimately work together to achieve community infrastructure needs. Examples of programs that fund voluntary involvement and collaboration among multiple farmers at the regional scale include the Regional Conservation Partnership Program in the U.S. [21], the European Union’s Agri-Environmental Scheme [18], and the United Kingdom’s Countryside Stewardship Facilitation Fund [22]. Social capital has been identified as a prerequisite for succeeding with programs that require farmer collaboration, and creating social capital can take many years [18,19]. Wang and Li [23] identified that for government to fully recognize the intended impact of community-based programs requires not only strong levels of community capital but also increased efforts to expand the dimensions of social capital. Hence, identifying the factors that influence the existence and strength of social capital is needed to achieve natural resource conservation collaboration among a region’s farmers [14,15,18,19,24,25].

Incorporating social capital measurements into farmer behavior analysis is challenging. The literature lacks consensus in defining and measuring social capital, such as membership to a cooperative or trust in the community [10,14,15,26], which will be further discussed in the next section. The current study adopts the social capital definition introduced by Robison et al. [10]. Based on the “Theory of Human Motivation” described by Maslow [27], Robison et al. [10] define social capital simply as “sympathy towards other people” and note four human motives for investing in social capital: ethics/promise-keeping (internal validation/moral norms), goodwill (external validation), altruism (caring), and belonging. We conducted a hypothetical field experiment among farmers in Turkey’s Konya region to measure various social capital motives following the work of Robinson et al. [10].

2. Definition and Measurement of Social Capital

According to Putnam [28], social capital refers to “the features of social organizations, such as trust, networks, and norms that facilitate coordination and cooperation for mutual benefit”. Social capital is seen as a tool to achieve socially beneficial outcomes through cooperation among individuals. Pretty and Ward [24] identify four attributes of social capital: trust, reciprocity, norms, and networks. Robison et al. [29] criticized social capital definitions presented in studies such as Putnam [28], Coleman [30], Portes [31], and Pretty and Ward [24] for not being precise and lacking “capital-like” properties. Robison et al. [29] pointed out that researchers’ social capital definitions tend to combine where social capital resides (i.e., networks), what it produces (i.e., cooperation), and what it is. Robison et al. [29] refined Putnam’s definition to include what gives rise to social capital (trust and norms), where it resides (networks), and what it produces (coordinated actions in society).

In an extension of Maslow’s “Theory of Human Motivation”, Robison et al. [10] provided an alternative social capital definition. According to Maslow [27], once a human’s physiological needs, such as hunger and safety, are satisfied, emotional needs emerge. Emotional needs involve love and belonging, esteem, and self-actualization. Therefore, Robison et al. [10] define social capital as “… sympathy or sense of obligation that one person or group of persons has toward another person or group of persons …”. Defined in terms of sympathy, social capital produces relational goods, which satisfy emotional needs [10,11], and bears capital-like properties [11]. Robison et al. [29] compare physical capital and social capital and describe how social capital has capital-like properties when it is defined in terms of sympathy. Studies by Robison and Oliver [32] and Siles et al. [33] found that farmers tend to discount the minimum price of farmland if selling land to a friend or family member (i.e., show sympathy), but ask for a premium if selling land to unfriendly neighbors. Hence, framing social capital in terms of sympathy is expected to affect farmers’ economic decisions.

Robison et al. [10] define five motives related to Maslow’s human motivation theory, and four—called social capital motives—relate to emotional needs. The first social capital motive links to internal validation and self-actualization. The relationship between a person and his or her ideal self can be viewed as doing the “right thing” and feeling good about it—for example, keeping a promise or recycling [10,25,34,35]. The second motive is to satisfy a need for external validation or gain esteem from other people. This motive means behaving to win “goodwill” or approval of important people [10]. Investing in this social capital motive will create relational goods, such as approval of others, which would satisfy our need for external validation or esteem [11]. An example for this motive could be a farmer giving a gift to an important person in the town to obtain his or her approval or goodwill. The third—belonging—involves increasing an individual’s understanding of other people’s feelings and in turn growing their understanding of others [10,36]. This motive is related to why people become members of groups or provide voluntary public goods [10,36]. The last motive relates to internalizing others’ well-being and sharing resources with them [25]. Hence, it is related to “altruism” and can help to satisfy love needs. This motive can explain why parents pay expenses of their children or why a person takes care of a friend or a relative when needed [11,25].

Because the current study intends to measure farm-level social capital to achieve sustainable agriculture outcomes, identifying the strength of farmers’ relationships with other farmers is important. Hence, defining social capital as sympathy—as done by Robison et al. [29]—allows for quantitatively measuring the magnitude of different social capital aspects using a metric similar to ones developed by Schmid [13] and Robison et al. [10]. To measure the relative importance of social capital motives and self-use, Robison et al. [10] conducted a hypothetical experiment among college students in the United States. The students were asked to split their endowment among alternative uses in a hypothetical war situation. The study found that, on average, students were not selfish and used part of their endowment based on social capital motives. In another case, Schmid [13] analyzed the relative importance of social capital motives and self-rewards using a hypothetical survey for returning a lost wallet among residents in the state of Michigan. The results of the study indicate that sympathy and norms can sometime be more influential on human behavior self-rewards. A line of previous studies that analyzed farmers’ behavior used membership in a farmer organization, such as an agricultural cooperative, as an indicator of farm-level social capital [3]. Previous studies that analyzed farmers’ behavior mostly used membership in a farmer organization, such as an agricultural cooperative, as an indicator of farm-level social capital [3]. However, organizational membership does not assure investment in social relationships and existence of social capital [29,37]. Producer organization membership could lead to knowing more people and receiving appropriate agricultural information but not really influence strategic interaction or relationships among members, especially for natural resource conservation. Another line of former studies assessed farmers’ opinions about the existence of trust in the community and the strength of their interaction with the community members [38]. However, existence of general trust in the community may not be directly linked to farmers’ use of natural resources.

3. Empirical Model

In the present study, some farmers made no contribution based on a social capital motive, as Figure 1 shows, and some made positive contributions. Hence, there are zero and positive values. Similar situations exist in the literature: number of hours worked, amount of charitable donations, and number of cigarettes smoked [39,40]. Each of these cases has some zero values and positive amounts. For example, the number of hours is zero for those who are not employed and positive for those who are employed [40]. Similarly, the amount of charitable donations and number of cigarettes smoked have zero and positive values [40]. This situation can be modeled as two-step decision based on Cragg’s hurdle model [39]. Cragg’s hurdle model gives the flexibility to separately identify the reasons for having zero values versus positive values and having different levels of positive values, as there could be different factors influencing each step [39,40]. In the current study, for each social capital motive, the first step indicates whether or not to make an investment in that social capital motive ($y_k$ = 0 versus $y_k$ > 0), and the second step indicates the amount invested in that social capital by those who decided to make an investment (amount of $y_k$ when it is positive) [39]. For example, for the belonging motive, the first step of the model represents the determinants of the decision to make an investment in the belonging motive, and the second step represents the determinants of the amount of investment made in the belonging motive for those who decided to make an investment in the first stage. The first part of the model (i.e., participation equation) is based on a univariate probit model, and the second part is based on a truncated regression model (i.e., for positive outcomes). The first part can be represented as follows, where $\mathsf{\Phi }(.)$ is the standard normal cumulative distribution function [40]:

$$P\left(s_k=1|{\mathit{z}}_{\mathit{k}}\right)=\mathsf{\Phi }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right)$$

(1)

$$P\left(s_k=0|{\mathit{z}}_{\mathit{k}}\right)=1-\mathsf{\Phi }\left({\mathit{z}\prime }_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right)$$

(2)

where $s_k$ is a binary dependent variable and is equal to one, if the farmer makes a positive investment into a social capital motive (i.e., $y_k$ > 0) and is equal to zero otherwise [40]. ${{\mathit{z}}^{\prime }}_{\mathit{k}}$ is a vector of independent variables and ${\mathit{\gamma }}_{\mathit{k}}$ is a vector of regression parameters to be estimated. The second part, the intensity equation, is given as [39,40]:

$$y_k={y_k}^{*}={\mathit{x}\prime }_k\mathit{\beta }+{\epsilon }_k \mathrm{if} s_k=1$$

(3)

$$y_k=0 \mathrm{if} s_k=0$$

(4)

where $y_k$ is the dependent variable representing the amount of investment in a social capital motive, ${\mathit{x}\prime }_k$ is the vector of independent variables, $\mathit{\beta }$ is the vector of regression paramenters to be estimated, and ${\epsilon }_k$ is the error term of the regression. Cragg’s hurdle model relies on the conditional independence assumption for the distribution of ${y_k}^{*}$, which can be represented as:

$$D\left({y_k}^{*}|{s_k,\mathit{x}}_{\mathit{k}}\right)=D\left({y_k}^{*}|{\mathit{x}}_{\mathit{k}}\right)$$

(5)

Hence, conditional on the explanatory variables, ${y_k}^{*}$ and $s_k$ are independent. In Cragg’s model, the latent variable ${y_k}^{*}$ is assumed to have a truncated normal distribution. Hence, the likelihood function for $y_k$ can be written as follows [40,41]:

$$\begin{array}{c}f\left(y_k|{\mathit{z}}_{\mathit{k}},{\mathit{x}}_{\mathit{k}}\right)\\ ={\left[1-\mathsf{\Phi }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right)\right]}^{1[y_k=0]}{\left\{\mathsf{\Phi }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right){\left[\mathsf{\Phi }\left({{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma \right)\right]}^{-1}\mathit{\varnothing }\left({{(y_k-\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k)/\sigma \right)/\sigma \right\}}^{1[y_k>0]}\end{array}$$

(6)

Estimating the ${\mathit{\gamma }}_{\mathit{k}}$, ${\mathit{\beta }}_{\mathit{k}}$, and $\sigma$ parameters is done using the maximum likelihood method [37,38]. When the model’s homoscedasticity assumption for the variance of error terms ${\epsilon }_k$ failed, we estimated the heteroskedastic version by replacing $\sigma$ in the likelihood function with $\sigma \left({\mathit{w}}_k\right)={exp({\mathit{w}}^{\prime }}_k{\mathit{\theta }}_k)$, where ${\mathit{w}}_k$ are the variables that influence the variance and ${\mathit{\theta }}_k$ are parameters to be estimated [40,42].

Based on the conditional independence assumption mentioned earlier and the assumption that $y_k$ is generated as ${y_k=s}_k{y_k}^{*}$, we can obtain the following expressions for expected value of $y_k$ [40]:

$$E\left(y_k|{s_k,\mathit{x}}_{\mathit{k}}\right)=s_{k}E\left({y_k}^{*}|{\mathit{x}}_{\mathit{k}}\right)$$

(7)

The conditional expectation of $y_k$ (i.e., when $s_k=1$) can be obtained as follows where any c, $\mathsf{\lambda }\left(c\right)\equiv \mathit{\varnothing }\left(c\right)/\mathsf{\Phi }\left(c\right)$ is the inverse Mills ratio, $\mathit{\varnothing }$(.) is the standard normal density function, and $\mathsf{\Phi }(.)$ is the standard normal cumulative distribution function [40]:

$$E\left(y_k|{\mathit{x}}_{\mathit{k}},y_k>0\right)=E\left({y_k}^{*}|{\mathit{x}}_{\mathit{k}}\right)={{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k+\sigma \mathsf{\lambda }({{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma )$$

(8)

The unconditional expectation of $y_k$ (i.e., with respect to the value of $s_k$) can be obtained as follows [40]:

$$E\left(y_k|{\mathit{z}}_{\mathit{k}},{\mathit{x}}_{\mathit{k}}\right)=E\left(s_k|{\mathit{z}}_{\mathit{k}}\right)E\left({y_k}^{*}|{\mathit{x}}_{\mathit{k}}\right)=\mathsf{\Phi }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right){[{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k+\sigma \mathsf{\lambda }({{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma )]$$

(9)

Marginal Effects

In Cragg’s model, three types of partial effects of an independent variable $x_j$ on the dependent variable can be calculated. For a continuous variable $x_j,$ assuming it is included in both ${\mathit{z}}_{\mathit{k}}$ and ${\mathit{x}}_{\mathit{k}}$, the first one is the partial effect of $x_j$ on the probability that $y_k>0$, which is given as follows [40], where ${\gamma }_{kj}$ is the coefficient of $x_j$ in ${\gamma }_k$ to be estimated:

$${\displaystyle \frac{\partial P\left(s_k=1|{\mathit{z}}_{\mathit{k}}\right)}{\partial x_j}}={\displaystyle \frac{\partial \mathsf{\Phi }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right)}{\partial x_j}}={\gamma }_{kj}\mathit{\varnothing }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right)$$

(10)

This partial effect can be interpreted as the influence of a variable on the probability of making an investment based on social capital motives. Hence, the results from this partial effect can be used by policy makers to increase the likelihood of farmers to generate social capital. The second partial effect of $x_j$ is on the expected value of $y_k$ given that $y_k>0.$ Hence, this can be interpreted as the partial effect of $x_j$ on the expected value of the positive investment made. This partial effect can be calculated as follows [40,41]:

$$\begin{array}{cc}\hfill {\displaystyle \frac{\partial E\left(y_k|{\mathit{x}}_{\mathit{k}},y_k>0\right)}{\partial x_j}}& ={\beta }_{kj}+{\beta }_{kj}{\displaystyle \frac{\partial \mathsf{\lambda }\left(\frac{{{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k}{\sigma }\right)}{\partial x_j}}\hfill \\ & ={\beta }_{kj}\left\{1-\mathsf{\lambda }\left({\displaystyle \frac{{{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k}{\sigma }}\right)\left[{\displaystyle \frac{{{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k}{\sigma }}+\mathsf{\lambda }\left({\displaystyle \frac{{{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k}{\sigma }}\right)\right]\right\}\hfill \end{array}$$

(11)

The term in brackets {.} is strictly between zero and one [40]. Hence, the partial effect of $x_j$ has the same sign as ${\beta }_{kj}$. This partial effect can be used by policy makers to increase the level of social capital among farmers who already invest in social capital. Lastly, the partial effect of $x_j$ on the unconditional expected value of $y_k$ can be calculated as follows [40,41]:

$$\begin{array}{l}{\displaystyle \frac{\partial E\left(y_k|{\mathit{z}}_{\mathit{k}},{\mathit{x}}_{\mathit{k}}\right)}{\partial x_j}}={\displaystyle \frac{\partial P\left(s_k=1|{\mathit{z}}_{\mathit{k}}\right)}{\partial x_j}}E\left(y_k|{\mathit{x}}_{\mathit{k}},y_k>0\right)+P\left(s_k=1|{\mathit{z}}_{\mathit{k}}\right){\displaystyle \frac{\partial E\left(y_k|{\mathit{x}}_{\mathit{k}},y_k>0\right)}{\partial x_j}}\\ ={\gamma }_{kj}\mathit{\varnothing }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right){[{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k+\sigma \mathsf{\lambda }({{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma )]+\mathsf{\Phi }\left({{\mathit{z}}^{\prime }}_{\mathit{k}}{\mathit{\gamma }}_{\mathit{k}}\right){\beta }_{kj}\left\{1-\mathsf{\lambda }({{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma )\left[{{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma +\mathsf{\lambda }({{\mathit{x}}^{\prime }}_k{\mathit{\beta }}_k/\sigma )\right]\right\}\end{array}$$

(12)

This partial effect considers that a farmer who originally had $y_k=0$ may switch to $y_k>0$ when $x_j$ changes [40]. This partial effect can be used to generate and increase the level of social capital, when farmers do not have pre-existing investment in social capital. The marginal effects can be evaluated at different values of the sample, or an average of the marginal effect evaluated at each observation can be reported (i.e., average marginal effect). In the current study, following Burke [41] and Wooldridge [40], we calculate and report average marginal effects. Standard errors for the average marginal effects are calculated using bootstrapping [40,41].

4. Data

The current study is part of a research project funded by the Scientific Research Council of Turkey to promote sustainable agriculture in the Konya region of Turkey. Recognized as a top producer of wheat, barley, corn, and feed crops, the Konya region is a major agricultural production region in Turkey [43]. In 2022, the Konya region contributed 9% of wheat production, 15% of barley production, 16% of corn production, 14% of sunflower seed production, and 31% of sugar beet production to the total Turkey production of these commodities. The Konya region’s average precipitation is 330 mm/year, which is lower than Turkey’s average of 573 mm/year [44]. Farmers rely heavily on groundwater for irrigation, and the region faces significant water quantity and quality problems. It is estimated that there are around 75,000 unregistered irrigation wells and 1000 sinkholes in the Konya region due to overuse of groundwater resources [44]. A changing climate is expected to contribute to water scarcity. Hence, maintaining social capital and reaching a cooperative outcome on water use are important.

The region has small-scale farms. Each farm averages 13 hectares and reports having 18 cattle and 152 sheep and goats on average. The region has an estimated 212,000 registered farmers [43]. The current study’s data were collected during 2018 through a face-to-face survey of 255 farmers in the Konya region. Before finalizing the survey instrument, semi-structured interviews and a survey pre-test were conducted with area farmers. To measure the relative importance of social capital attributes, following Robison et al. [10] and Schmid [13], we conducted a hypothetical field experiment. Farmers were provided with the following scenario:

“Assume, due to a drought happening this year, there is a lack of enough food for people in your town. You are given 100 kg of wheat from an aid organization and you are told that you can use this 100 kg of wheat any way you prefer. You can decide if you would use the wheat fully for your own consumption or share it with others. Please indicate below how you would allocate the 100 kg of wheat among alternative uses. There is no right or wrong usage endowment. The total endowment must add up to 100 kg. You do not have to share wheat.”

Farmers could keep wheat stocks for themselves (selfishness), give wheat to another farmer with whom a mutual promise existed to help each other during food shortages (internal validation/ethics), share with a relative who lived in the same community (altruism), supply wheat to a town official to show goodwill (external validation/goodwill), and transparently supply wheat stocks to a central pool for community efforts (belonging). The questionnaire used for the scenario analysis can be found in the Appendix A.

Table 1. Summary of statistics.

Variable	Description	Mean	Range
Social Capital Hypothetical Field Experiment
Own Consumption	Amount of wheat spared for own consumption in the scenario.	52	10–100
Promise-Keeping	Amount of wheat promised to someone in the scenario.	12	0–50
Altruism	Amount of wheat provided to a relative in the scenario.	11	0–50
Goodwill	Amount of wheat directed to the town official in the scenario.	8	0–50
Belonging	Amount of wheat directed to community efforts in the scenario.	17	0–90
Independent Variables
Farm Size	Decares of land owned (1 decare $\cong$ 0.25 acre)	201	0–2500
Land Rented	1 if the farmer rented land from other farmers, 0 otherwise.	0.56	0–1
Age	Age of the farmer in years	50	18–78
Education of the Farm Operator		-	0–7
	0 if illiterate
	1 if literate but has no formal education
	2 if primary school graduate
	3 if secondary school graduate
	4 if high school graduate
	5 if has associate degree
	6 if has bachelor’s degree
	7 if has graduate degree
Off-Farm Income	1 if has off-farm income, 0 otherwise	0.58	0–1
Livestock Production
Cattle	1 if has cattle, 0 otherwise	0.56	0–1
Sheep and Goat	1 if has sheep and/or goats, 0 otherwise	0.27	0–1
Crop Production
Barley	1 if grows barley, 0 otherwise	0.77	0–1
Corn	1 if grows corn, 0 otherwise	0.42	0–1
Alfalfa	1 if grows alfalfa, 0 otherwise	0.14	0–1
Cooperative Membership	1 if is current member of a sugar beet producer cooperative, 0 otherwise	0.73	0–1
Home Internet	1 if has broadband internet access at home, 0 otherwise	0.36	0–1
Cell Phone Internet	1 if has internet access from cell phone, 0 otherwise	0.70	0–1
Impact on Agricultural Production Decisions
Other Farmers	Amount of influence on farmer’s agricultural production decisions (Likert scale: 1 = none, 2, 3 = some, 4, 5 = very much) 1	3.58	1–5
Non-farming Neighbors		1.88	1–5
Extension Programs		2.44	1–5
Banking Institutions		1.19	1–5
Own Experience		4.82	1–5
Hired Labor	1 if hired labor, 0 otherwise	0.52	0–1
Asset Ownership	1 if owns a tractor, 0 otherwise	0.82	0–1

¹ The same Likert scale is used for non-farming neighbors, extension programs, banking institutions, and own experience variables below.

shows the results of the hypothetical field experiment. On average, farmers kept 52 kg of the wheat for themselves and directed 48 kg to social capital motives. Hence, farmers were not found to be exclusively selfish. The belonging motive had the highest stock allocation (17 kg), followed by the ethics/promise-keeping motive (12 kg), altruism motive (11 kg), and goodwill motive (8 kg). The belonging motive can be considered a social norm because the amount farmers provide is declared to everyone in the town. A farmer could behave in accordance with town norms or otherwise face sanctions if they exist [45]. A social norm in Turkish towns, “imece”, suggests a “do-it-together culture”, so farmers may have viewed the belonging motive as a way to demonstrate their participation in local culture.

The results are consistent with prior research. Ward et al. [15] analyzed cooperation and management of local common resources in remote rural communities, and they found that contributions to others ranged between 40% and 52% of a total endowment. Robison et al. [10] found that respondents allocated 33% to oneself and 9% to the goodwill motive. The latter is consistent with our finding for goodwill. Robinson et al. [10] found that altruism had the highest endowment among social capital motives—whereas belonging is found to be the most important social capital motive in this study. Recall also that Robison et al. [10] sampled U.S. college students, whereas the current study surveyed farmers in rural Turkey.

Using the dot plot feature of Stata^®, Figure 1 presents the endowments for one’s own consumption and social capital motives. Thicker dots represent more observations. The belonging motive is more heavily represented than the other three social capital motives, and the goodwill motive has the most zeros. Of the 252 surveyed farmers, 52 kept all of the endowment for themselves.

Figure 1. Dot plot of one’s own consumption and social capital motives.

Figure 1. Dot plot of one’s own consumption and social capital motives.

5. Factors Affecting the Allocation of Endowment Findings

Next, we seek to understand the farm business attributes, demographics, and socio-economic factors influencing respondents’ wheat stock allocation decisions. This part reports regression results for each social capital motive separately. We report the regression coefficients and marginal effects from the Cragg’s hurdle model with exponential specification, as the model with linear specification had lower log-pseudolikelihood values. We report the results of the Cragg’s hurdle model with linear specification in the Appendix A

Table A1. Results of the hurdle models for social capital motive of promise-keeping ¹.

Variable	Cragg’s Hurdle Model (Linear)				Heckman Selection Model (Two-Step)				Heckman Selection Model 2 (MLE)
	Intensity Equation		Participation Equation		Intensity Equation		Participation Equation		Intensity Equation		Participation Equation
	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.
Farm Size	−0.001	0.002	0.000	0.000	0.000	0.005	0.000	0.000	0.000	0.002	0.000	0.000
Land Rented	−2.015	1.734	0.158	0.204	−0.626	3.476	0.158	0.204	−0.496	1.666	0.148	0.182
Age	0.068	0.056	−0.018 **	0.008	−0.077	0.203	−0.018 **	0.009	−0.036	0.058	−0.008	0.009
Education	−0.552	0.653	0.090	0.096	0.380	1.708	0.090	0.091	0.513	0.668	0.038	0.071
Off-Farm Income	−0.862	1.574	0.055	0.214	−0.123	3.382	0.055	0.219	−0.513	1.804	−0.006	0.195
Owns Cattle	−1.963	1.800	0.331	0.210	1.099	4.763	0.331	0.212	0.032	1.677	0.390 **	0.179
Owns Sheep and Goats	1.621	1.791	−0.569 **	0.224	−3.424	6.641	−0.569 **	0.230	−1.640	1.931	−0.433 **	0.200
Grows Barley	−2.302	2.048	0.333	0.248	0.636	5.034	0.333	0.260	0.904	2.140	−0.033	0.216
Grows Corn	−1.764	1.643	0.716 ***	0.230	4.804	8.039	0.716 ***	0.234	3.134 *	1.733	0.355 *	0.215
Grows Alfalfa	2.414	1.979	0.728 **	0.323	7.723	7.495	0.728 **	0.345	4.978 **	2.066	0.746 ***	0.260
Cooperative Membership	−0.575	1.502	−0.143	0.212	−2.018	3.720	−0.143	0.220	−0.410	1.626	−0.337 *	0.183
Broadband Internet	3.103 *	1.653	−0.269	0.214	0.581	4.213	−0.269	0.221	0.581	1.712	−0.121	0.167
Cell Phone Internet	2.496	1.900	−0.008	0.241	2.151	3.825	−0.008	0.252	1.810	2.014	−0.029	0.268
Other Farmers	−0.561	0.681	−0.053	0.078	−0.954	1.295	−0.053	0.079	−0.442	0.711	−0.078	0.071
Non-farming Neighbors	0.733	0.640	−0.097	0.077	−0.072	1.552	−0.097	0.087	−0.225	0.682	−0.015	0.074
Extension Programs	0.560	0.590	−0.024	0.064	0.381	1.108	−0.024	0.077	0.198	0.571	−0.021	0.059
Banks	0.262	0.633	0.094	0.071	1.094	1.503	0.094	0.079	0.756	0.615	0.074	0.072
Own Experience	−1.768	1.223	−0.140	0.189	−2.464	2.646	−0.140	0.209	−2.054 **	0.988	−0.188	0.277
Hired Labor	3.492 **	1.660	−0.428 **	0.208	−0.307	5.063	−0.428 **	0.209	−0.253	1.551	−0.166	0.187
Asset Ownership	−0.444	1.872	0.185	0.285	1.880	5.232	0.185	0.274	−0.135	2.318	0.334	0.252
Constant	24.450 ***	8.282	1.386	1.180	17.887	17.546	1.386	1.272	19.839	.	1.401 *	0.796
N	218				218				218
R2	0.053
Σ	8.094 ***				18.77				10.41
$\rho$					1				1
Λ					18.77				10.41
Chi2(20)	22.89				5.04
p-Value for Chi2	0.2941				0.9997
log-pseudolikelihood	−603.565								−585.686

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the promise-keeping motive of social capital in continuous format for the outcome model and binary format for the selection model. ² Convergence for the maximum likelihood method was not achieved; hence, the Chi-squared test statistic is not reported. *, **, *** indicates statistical significance at the 10%, 5%, and 1% levels, respectively.

,

Table A2. Results of the hurdle models for social capital motive of altruism ¹.

Variable	Cragg’s Hurdle Model (Linear)				Heckman Selection Model (Two-Step)				Heckman Selection Model (MLE)
	Intensity Equation		Participation Equation		Intensity Equation		Participation Equation		Intensity Equation		Participation Equation
	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.
Farm Size	0.001	0.002	0.001 *	0.000	0.002	0.003	0.001 *	0.000	0.003	0.002	0.001 ***	0.000
Land Rented	−0.808	1.509	−0.131	0.197	−1.065	1.620	−0.131	0.204	−0.411	1.500	−0.285	0.227
Age	−0.050	0.053	−0.025 ***	0.009	−0.106	0.114	−0.025 ***	0.009	−0.170 ***	0.061	−0.018 **	0.008
Education	−0.057	0.636	0.014	0.099	−0.009	0.700	0.014	0.086	0.412	0.666	−0.029	0.080
Off-Farm Income	1.282	1.661	0.576 ***	0.210	2.607	2.770	0.576 ***	0.221	3.615 **	1.798	0.497 ***	0.187
Owns Cattle	1.468	1.612	0.311	0.208	1.992	1.982	0.311	0.211	1.892	1.661	0.350 **	0.166
Owns Sheep and Goats	1.511	2.025	−0.252	0.220	0.752	1.953	−0.252	0.226	0.213	2.026	−0.087	0.230
Grows Barley	−2.462	1.904	0.206	0.240	−1.677	2.080	0.206	0.258	1.034	1.948	−0.120	0.230
Grows Corn	−1.493	1.877	0.669 ***	0.220	0.529	3.357	0.669 ***	0.232	3.049 *	1.599	0.452 ***	0.171
Grows Alfalfa	2.289	1.440	0.428	0.283	3.191	2.647	0.428	0.312	4.107 ***	1.634	0.319	0.274
Cooperative Membership	−1.276	1.426	−0.216	0.215	−1.742	1.904	−0.216	0.219	−2.009	1.514	−0.297	0.187
Broadband Internet	0.201	1.894	−0.379 *	0.208	−0.677	2.089	−0.379 *	0.218	−2.127	1.794	−0.145	0.198
Cell Phone Internet	−0.770	2.106	−0.208	0.241	−1.296	2.046	−0.208	0.247	−2.250	1.970	−0.120	0.199
Other Farmers	0.982 **	0.516	0.073	0.078	1.054	0.676	0.073	0.077	1.309 **	0.579	0.080	0.076
Non-farming Neighbors	1.701 **	0.771	−0.073	0.083	1.389 **	0.665	−0.073	0.083	0.895	0.756	−0.013	0.084
Extension Programs	−0.874	0.568	−0.009	0.068	−0.786	0.534	−0.009	0.073	−1.244 **	0.592	0.023	0.064
Banks	−0.396	0.633	0.011	0.072	−0.249	0.584	0.011	0.076	0.211	0.578	−0.046	0.064
Own Experience	−2.005 **	0.931	−0.470 **	0.216	−2.693	1.821	−0.470 **	0.231	−3.460 ***	0.812	−0.232 *	0.129
Hired Labor	−0.924	1.451	−0.419 **	0.207	−1.858	2.225	−0.419 **	0.206	−3.131 **	1.520	−0.306	0.193
Asset Ownership	−6.265 **	2.902	−0.193	0.275	−6.167 ***	2.179	−0.193	0.281	−5.074 *	2.927	−0.367	0.286
Constant	33.012 ***	6.936	3.496 ***	1.307	34.635 ***	8.879	3.496 ***	1.359	34.744		2.328 ***	0.868
N	218				218				218
R2	0.056
σ	7.853 ***				8.381				10.014
$\rho$					0.680				1
λ					5.701				10.014
Chi2(20)	26.26				29.35
p-Value for Chi2	0.158				0.081
Log pseudolikelihood	−592.514								−570.751

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the altruism motive of social capital in continuous format for the outcome model and binary format for the selection model. *, **, *** indicates statistical significance at the 10%, 5%, and 1% levels, respectively.

,

Table A3. Results of the hurdle models for social capital motive of goodwill ¹.

Variable	Cragg’s Hurdle Model (Linear)				Heckman Selection Model (Two-Step)				Heckman Selection Model (MLE)
	Intensity Equation		Participation Equation		Intensity Equation		Participation Equation		Intensity Equation		Participation Equation
	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.
Farm Size	0.001	0.001	0.000	0.000	−0.002	0.005	0.000	0.000	0.000	0.001	0.000	0.000
Land Rented	0.434	1.225	0.163	0.198	−1.018	3.443	0.163	0.199	0.145	1.375	0.157	0.197
Age	−0.015	0.037	−0.010	0.009	0.056	0.152	−0.010	0.009	−0.005	0.040	−0.010	0.008
Education	0.598	0.411	0.085	0.089	−0.214	1.641	0.085	0.087	0.484	0.406	0.089	0.088
Off-Farm Income	−0.975	1.170	0.514 **	0.211	−4.801	6.366	0.514 **	0.216	−1.461	1.383	0.516 **	0.211
Owns Cattle	0.033	1.689	0.461 **	0.207	−3.731	6.173	0.461 **	0.208	−0.394	1.536	0.489 **	0.224
Owns Sheep and Goats	0.121	1.236	−0.182	0.219	1.634	3.722	−0.182	0.220	0.306	1.256	−0.194	0.222
Grows Barley	−1.626	1.792	0.188	0.250	−3.195	4.028	0.188	0.247	−1.857	1.889	0.187	0.253
Grows Corn	−1.003	1.134	0.660 ***	0.223	−6.292	8.087	0.660 ***	0.225	−1.705	1.503	0.699 ***	0.240
Grows Alfalfa	4.452 ***	1.282	0.165	0.281	2.978	4.326	0.165	0.285	4.192 ***	1.346	0.148	0.285
Cooperative Membership	0.210	1.073	−0.395 *	0.208	3.411	5.340	−0.395 *	0.217	0.620	1.167	−0.404 **	0.208
Broadband Internet	−1.026	1.051	−0.454 **	0.211	2.480	5.800	−0.454 **	0.217	−0.532	1.138	−0.468 **	0.211
Cell Phone Internet	3.160 **	1.542	0.197	0.237	1.733	3.873	0.197	0.247	2.895 **	1.442	0.204	0.239
Other Farmers	0.770 *	0.431	0.022	0.078	0.587	1.126	0.022	0.076	0.705	0.435	0.018	0.078
Non-farming Neighbors	1.472 **	0.756	−0.206 ***	0.083	3.185	2.819	−0.206 **	0.086	1.688 **	0.840	−0.201 **	0.084
Extension Programs	−0.507	0.435	−0.015	0.068	−0.445	0.984	−0.015	0.072	−0.487	0.436	−0.010	0.072
Banks	−1.281 ***	0.466	0.036	0.070	−1.567	1.123	0.036	0.076	−1.298 ***	0.458	0.028	0.074
Own Experience	−0.791	0.858	0.065	0.148	−1.404	2.388	0.065	0.175	−0.835	0.821	0.073	0.138
Hired Labor	−3.093 ***	1.075	−0.403 **	0.198	0.046	5.195	−0.403 **	0.203	−2.636 **	1.083	−0.422 **	0.201
Asset Ownership	−2.361	2.117	−0.122	0.287	−2.167	3.949	−0.122	0.277	−2.383	2.138	−0.171	0.316
Constant	20.382 ***	5.573	−0.344	0.983	37.161	27.425	−0.344	1.107	22.714 ***	5.764	−0.331	0.931
N	218				218				218
R2	0.089
σ	5.221 ***				15.088				5.366 ***
$\rho$					−1				0.384
λ					−15.088				2.060
Chi2(20)	67.91				9.12				65.64
p-Value for Chi2	0.000				0.981				0.000
log-pseudolikelihood	−462.692								−463.042

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the goodwill motive of social capital in continuous format for the outcome model and binary format for the selection model. *, **, *** indicates statistical significance at the 10%, 5%, and 1% levels, respectively.

and

Table A4. Results of the hurdle models for social capital motive of belonging ¹.

Variable	Cragg’s Hurdle Model (Linear)				Heckman Selection Model (Two-Step)				Heckman Selection Model (MLE)
	Intensity Equation		Participation Equation		Intensity Equation		Participation Equation		Intensity Equation		Participation Equation
	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.
Farm Size	−0.010 **	0.005	0.000	0.000	−0.010	0.015	0.000	0.000	−0.006 **	0.003	0.000	0.000
Land Rented	−2.617	3.115	−0.023	0.203	−0.977	7.786	−0.023	0.205	−1.725	2.386	−0.023	0.203
Age	0.276 **	0.141	−0.007	0.009	0.325	0.478	−0.007	0.009	0.187 **	0.094	−0.007	0.009
Education	0.951	1.836	−0.049	0.089	1.643	4.180	−0.049	0.089	0.668	1.389	−0.049	0.089
Off-Farm Income	0.078	2.878	0.317	0.217	−6.423	19.120	0.317	0.214	0.172	2.032	0.317	0.217
Owns Cattle	−2.782	3.260	0.042	0.209	−2.201	8.215	0.042	0.211	−1.676	2.334	0.041	0.210
Owns Sheep and Goats	2.485	3.599	−0.379 *	0.215	9.903	23.329	−0.379 *	0.221	1.602	2.571	−0.380 *	0.215
Grows Barley	−6.848	4.286	−0.148	0.269	−2.558	11.364	−0.148	0.260	−5.139 *	3.165	−0.151	0.272
Grows Corn	−2.758	3.423	0.619 ***	0.220	−15.773	35.892	0.619 ***	0.231	−2.339	2.503	0.621 ***	0.221
Grows Alfalfa	0.779	4.052	0.145	0.308	−2.897	13.764	0.145	0.307	0.486	2.896	0.144	0.308
Cooperative Membership	−1.008	3.655	−0.027	0.200	0.196	8.413	−0.027	0.220	−0.615	2.544	−0.029	0.201
Broadband Internet	4.649	3.504	−0.061	0.204	4.332	8.577	−0.061	0.220	3.115	2.430	−0.060	0.204
Cell Phone Internet	4.101	4.288	−0.050	0.243	4.170	9.625	−0.050	0.252	3.107	3.005	−0.051	0.243
Other Farmers	−1.569	1.437	−0.036	0.081	−0.185	3.735	−0.036	0.079	−1.091	1.010	−0.036	0.081
Non-farming Neighbors	3.961 ***	1.464	−0.105	0.080	5.162	6.633	−0.105	0.084	2.929 ***	1.041	−0.104	0.080
Extension Programs	0.828	1.106	0.008	0.066	0.450	2.655	0.008	0.073	0.596	0.796	0.007	0.067
Banks	−0.044	1.133	0.092	0.075	−2.155	5.961	0.092	0.076	−0.149	0.826	0.092	0.075
Own Experience	7.366 **	3.684	0.085	0.161	2.413	8.021	0.085	0.170	4.219 **	1.955	0.086	0.160
Hired Labor	7.772 **	3.386	−0.161	0.197	8.822	11.522	−0.161	0.205	5.435 **	2.246	−0.160	0.198
Asset Ownership	0.646	4.925	−0.238	0.279	5.497	17.054	−0.238	0.287	0.392	3.633	−0.240	0.280
Constant	−32.589	23.715	0.847	1.047	9.249	61.020	0.847	1.098	−7.964	13.187	0.849	1.043
N	218				218				218
R2	0.036
σ	15.820 ***				46.906				13.262 ***
$\rho$					−1				0.056
λ					−46.906				0.740
Chi2(20)	20.59				3.39				29.89
p-Value for Chi2	0.422				0.999				0.072
log-pseudolikelihood	−730.364								−743.814

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the belonging motive of social capital in continuous format for the outcome model and binary format for the selection model. *, **, *** indicates statistical significance at the 10%, 5%, and 1% levels, respectively.

. A discussion of alternative model specifications and model results is provided in the robustness subsection.

A variable likely to create the endogeneity problem is “cooperative membership” [46]. Unobserved factors that influence investment in the ethics/promise-keeping motive could also influence a farmer’s decision to join a cooperative [46]. To test for the endogeneity problem, we ran an instrumental variable regression using the two-stage least squares estimation (2SLS) method, where agreement with the statement “Other farmers consult with me on issues related to agriculture” is used as an instrument for cooperative membership, given that a farmer seeing himself or herself as an opinion leader could influence the decision to join a cooperative [47]. To check the relatedness of the instrumental variable with cooperative membership, we ran OLS and probit regressions of cooperative membership with the instrumental variable and the full set of exogenous variables that were included in the ethics/promise-keeping motive regression. In both the OLS and probit regressions, agreeing with “Other farmers consult with me on issues related to agriculture” was statistically significant at the 1% significance level. Hence, the instrumental variable is related to cooperative membership. To be a valid instrument, agreeing with “Other farmers consult with me on issues related to agriculture” should not be directly linked with the amount of wheat spared by the farmer for the ethics/promise-keeping motive [39]. We cannot run a statistical test for that, given the one instrument, but we do not expect this variable to be directly linked to the ethics/promise-keeping motive, as it could be more related to agricultural experience and knowledge. We conducted the statistical test for endogeneity of the cooperative membership variable using the Durbin–Wu–Hausman (DWH) test statistic with heteroskedasticity-robust errors [48]. We report the results for the endogeneity check of covariates for each social capital motive in the Appendix B.

For the belonging motive, in addition to the cooperative membership variable, another variable that could create an endogeneity problem is “having hired labor”. Unobserved factors that influence investment in the belonging motive could also affect a farmer’s decision to hire labor. To test for the possible endogeneity problem, we ran an instrumental variable regression using the 2SLS method with instruments used for cooperative membership and hired labor variables. We used farmers’ agreement with the statement “I prefer not to take risks in agricultural production” as an instrument for having hired labor. We preferred this variable as an instrument because farmers who hire labor might be less likely to take risks [49]. We did not expect this variable to directly link to the belonging motive. We continued to use agreement with “Other farmers consult with me on issues related to agriculture” as an instrument for the cooperative membership variable.

For each social capital motive, we report the average marginal effects in columns A, B, and C of the results tables. Column C represents the average marginal effect on the probability to spare wheat, column B represents the average marginal effect on the actual amount of wheat spared, and column A represents the average aggregate marginal effect on the amount of wheat spared for the corresponding social capital motive.

5.1. Ethics/Promise-Keeping Motive

For the ethics/promise-keeping motive, which is represented by the amount of wheat given to another farmer with whom a mutual promise existed to help each other during food shortages, regression results from the heteroskedastic–Cragg’s hurdle model are reported in

Table 2. Results of the heteroskedastic–Cragg’s hurdle (exponential) model for social capital motive of promise-keeping ¹.

Variables	Intensity Equation		Participation Equation		Variance Equation 3		Marginal Effects
							(A)	(B)	(C)
	Coeff.	Std. Err. 2	Coeff.	Std. Err.	Coeff.	Std. Err.	$\frac{\mathit{d}\mathit{E}\left(\mathit{y}\right)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{E}(\left.\mathit{y}\right|\mathit{y}>0)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{P}\mathit{r}(\mathit{y}>0)}{\mathit{d}\mathit{X}}$
Farm Size	0.000 *	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Land Rented	−0.011	0.131	0.158	0.204	−0.451 **	0.192	−0.078	0.080	0.158
Age	0.001	0.002	−0.018 *	0.009	0.003	0.010	0.001	0.003	−0.018 *
Education	−0.052	0.035	0.090	0.091	0.140 **	0.059	−0.038	0.033	0.090
Off-Farm Income	0.157 *	0.090	0.055	0.219	−0.643 ***	0.214	−0.013	0.082	0.055
Owns Cattle	0.006	0.089	0.331	0.212	−0.268	0.197	−0.110	0.086	0.331
Owns Sheep and Goats	−0.053	0.087	−0.569 **	0.230	0.122	0.222	0.068	0.090	−0.569 **
Grows Barley	−0.129 *	0.071	0.333	0.260	0.013	0.222	−0.101	0.098	0.333
Grows Corn	−0.129	0.090	0.716 ***	0.234	−0.001	0.174	−0.082	0.084	0.716 ***
Grows Alfalfa	0.162 **	0.077	0.728 **	0.345	−0.051	0.218	0.143	0.099	0.728 **
Cooperative Membership	0.014	0.062	−0.143	0.220	−0.134	0.248	−0.044	0.082	−0.143
Broadband Internet	0.165 **	0.072	−0.269	0.221	0.155	0.239	0.134 *	0.081	−0.269
Cell Phone Internet	−0.028	0.080	−0.008	0.252	−0.042	0.219	0.125	0.094	−0.008
Other Farmers	0.088	0.056	−0.053	0.079	−0.281 **	0.110	−0.002	0.030	−0.053
Non-farming Neighbors	0.028	0.046	−0.097	0.087	−0.006	0.091	0.037	0.031	−0.097
Extension Programs	−0.023	0.033	−0.024	0.077	0.097	0.063	0.004	0.026	−0.024
Banks	−0.048 *	0.026	0.094	0.079	0.143 **	0.071	−0.004	0.028	0.094
Own Experience	−0.124	0.078	−0.140	0.209	−0.063	0.156	−0.085	0.057	−0.140
Hired Labor	−0.022	0.060	−0.428 **	0.209	0.407 **	0.180	0.143 *	0.077	−0.428 **
Asset Ownership	0.122	0.106	0.185	0.274	−0.022	0.270	−0.073	0.120	0.185
Constant	3.109 ***	0.434	1.386	1.272
N	218
Pseudo R2	0.231
Chi2(20)	227.45
p-Value for Chi2	0.000
Log-pseudolikelihood	−378.404

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the promise-keeping motive of social capital in continuous format for the outcome model and binary format for the selection model. ² Conditional heteroskedastic standard errors are reported. ³ Dependent variable is ln(σ) for the variance equation of the intensity equation. *, **, *** indicates statistical significance at 10%, 5%, and 1% levels, respectively.

along with estimation results for the variance equation for the error term to detect heteroskedasticity. The results for the model with homogeneity assumption are available upon request. The results show that the variables of land rented, off-farm income, and the influence of other farmers on the agricultural production decisions of the farmer had negative and statistically significant effects on variance of the error term. On the other hand, the variables of education, banks, and having hired labor had positive effects. Hence, the data show evidence of heteroskedasticity, and using the model’s heteroskedastic version is preferred. The heteroskedastic model has a higher log-pseudolikelihood value than the homoskedastic model. The heteroskedastic model’s R-squared value was 0.231, and the model was statistically significant at the 1% level.

For the participation equation, the variables of age, owning sheep and goats, and hiring labor had negative and statistically significant effects. The result on age is similar to the results in Robison et al. [10] and Bellamare and Kröger [50]. As age increases, the span of time over which an investment’s benefits are realized decreases. Hence, older farmers may decide not to invest in the ethics/promise-keeping aspect of social capital [51]. Older farmers may also believe that others would not keep their promises and not feel morally obligated to keep theirs [25,52]. Sheep and goats are traditionally raised in Turkey, and unlike cattle farmers, sheep and goat farmers rely less on outside resources, such as those to obtain feed [53]. The Konya region relies on an informal labor market consisting of seasonal migrant workers paid daily without social security [54,55]. Hence, farmers who own sheep and goats or hire labor may not value investment in the ethics/promise-keeping motive due to relying less on recurring relationships.

Growing corn and alfalfa had positive and statistically significant effects on the probability that a farmer devotes wheat toward the ethics/promise-keeping motive. Both corn and alfalfa are grown as cash crops and require more intensive management than wheat [56]. Hence, corn and alfalfa farmers could rely more on other people through market transactions than wheat growers [57].

For the intensity equation, farm size, off-farm income, growing alfalfa, and broadband internet access had positive and statistically significant effects on the amount spared for the ethics/promise-keeping motive. The result for farm size mirrors results from Sulemana and James Jr. [58], which found slight positive evidence between farm sales and farmers’ ethical judgement on rinsing before disposing of pesticide containers. The off-farm income result is similar to results from Cardoso and James [59], who found that farmers relying less on farm income (i.e., who have off-farm income) are less likely to use controversial agricultural practices. Broadband internet access’s positive influence on the amount of wheat spared for the ethics/promise-keeping motive is similar to the results of Bauernschuster et al. [60]. Broadband internet access could allow farmers to access markets and information and lead to more concern about ethical values [61,62].

Lastly, the barley production and banks variables had negative and statistically significant effects on the amount of wheat spared for the ethics/promise-keeping motive. In the study region, barley is grown using traditional seed varieties on unirrigated land and primarily used as on-farm animal feed [63]. Hence, barley farmers might rely less on outside sources and value relationships less. Farmers whose agricultural decisions are influenced by banks rely on formal capital markets and legal procedures to secure credit [62]. Hence, for these farmers, ethical values could be less relevant than for farmers who rely on informal markets and family, relatives, or friends for financing. Some support of this claim was demonstrated by the positive coefficient for the variable of influence of other farmers, which had a p-value on the border of being significant.

Overall, market-oriented farmers could be more concerned about transactions conducted and care more for ethics/moral norms [57]. Farmers in Turkey still rely significantly on informal markets and personal transactions based on trust and reputation, especially in marketing and finance [62,64]. This highlights the importance of relationships and social capital. A similar conclusion was reached by Yin and Zhou [65], who used path analysis to find a positive link between the level of an individual farmer’s social capital and the farmer’s entrepreneurial interest. Not finding the cooperative membership variable statistically significant might reflect that cooperative membership may not be a good proxy for ethical values.

In column C, growing alfalfa had the highest marginal effect on probability to invest in the ethics/promise-keeping motive. Farmers who grow alfalfa were 22% more likely to invest in the ethics/promise-keeping motive than farmers who do not. Growing corn had a similar effect. On the other hand, farmers who own sheep and goats were 17% less likely to invest in ethics/promise-keeping than farmers who do not. In column B, a unit increase in the influence of one’s own experience on a farmer’s agricultural production decisions caused a 2.86 kg decrease in the amount of wheat spared for the ethics/promise-keeping motive. For the aggregate average marginal effect in column A, growing alfalfa had the largest and most statistically significant effect.

5.2. Altruism Motive

Table 3. Results of the heteroskedastic–Cragg’s hurdle (exponential) model for social capital motive of altruism ¹.

Variables	Intensity Equation		Participation Equation		Variance Equation 3		Marginal Effects
							(A)	(B)	(C)
	Coeff.	Std. Err. 2	Coeff.	Std. Err.	Coeff.	Std. Err.	$\frac{\mathit{d}\mathit{E}\left(\mathit{y}\right)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{E}(\left.\mathit{y}\right|\mathit{y}>0)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{P}\mathit{r}(\mathit{y}>0)}{\mathit{d}\mathit{X}}$
Farm Size	0.000	0.000	0.001 *	0.000	0.000	0.000	0.004	0.001	0.000 *
Land Rented	0.007	0.098	−0.131	0.204	−0.497 **	0.212	−1.859	−2.088	−0.042
Age	0.000	0.003	−0.025 ***	0.009	−0.001	0.006	−0.140 **	0.001	−0.008
Education	0.011	0.026	0.014	0.086	−0.143	0.098	−0.137	−0.435	0.004
Off-Farm Income	0.034	0.121	0.576 ***	0.221	−0.104	0.275	3.431 *	0.143	0.184 ***
Owns Cattle	0.035	0.085	0.311	0.211	0.056	0.196	2.311	0.878	0.100
Owns Sheep and Goats	−0.009	0.101	−0.252	0.226	0.109	0.207	−1.285	0.323	−0.080
Grows Barley	−0.218 **	0.091	0.206	0.258	−0.217	0.226	−1.751	−4.849 **	0.066
Grows Corn	−0.094	0.084	0.669 ***	0.232	−0.289	0.191	2.106	−2.962	0.214 ***
Grows Alfalfa	0.114	0.091	0.428	0.312	−0.076	0.251	3.542 *	1.697	0.137
Cooperative Membership	−0.124	0.079	−0.216	0.219	−0.104	0.203	−2.864 *	−2.677	−0.069
Broadband Internet	−0.010	0.074	−0.379 *	0.218	0.202	0.195	−1.806	0.721	−0.121 *
Cell Phone Internet	−0.004	0.085	−0.208	0.247	0.100	0.212	−0.999	0.374	−0.067
Other Farmers	0.047	0.031	0.073	0.077	−0.096	0.083	0.718	0.419	0.023
Non-farming Neighbors	0.074 **	0.034	−0.073	0.083	0.010	0.067	0.433	1.372 **	−0.023
Extension Programs	−0.087 ***	0.027	−0.009	0.073	0.167 **	0.072	−0.618	−0.807	−0.003
Banks	0.009	0.029	0.011	0.076	−0.092	0.082	−0.052	−0.245	0.004
Own Experience	−0.092 **	0.041	−0.470 **	0.231	0.048	0.110	−3.605 ***	−1.433 *	−0.150 **
Hired Labor	−0.012	0.076	−0.419 **	0.206	0.093	0.165	−2.316	0.195	−0.134 **
Asset Ownership	0.000	0.145	−0.193	0.281	−0.300	0.252	−1.823	−1.340	−0.062
Constant	3.281 ***	0.315	3.496 ***	1.359
N	218
Pseudo R2	0.238
Chi2(20)	226.94
p-Value for Chi2	0.000
log-pseudolikelihood	−363.714

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the altruism motive of social capital in continuous format for the outcome model and binary format for the selection model. ² Conditional heteroskedastic standard errors are reported. ³ Dependent variable is ln(σ) for the variance equation of the intensity equation. *, **, *** indicates statistical significance at 10%, 5%, and 1% levels, respectively.

reports regression results for the altruism aspect of social capital, which is the amount of wheat shared with a relative who lives in the same community, from the heteroskedastic–Cragg’s hurdle model. Results for the model with the homogeneity assumption are available upon request. For the intensity equation, estimation results for the variance of the error term show that the variables of having rented land and being influenced by extension programs had statistically significant effects on variance. Hence, evidence exists to suggest heteroskedasticity in the data. The heteroskedastic model was overall statistically significant at the 1% level and had an R-squared value of 0.238.

For the participation/selection equation, farm size, off-farm income, and growing corn had positive and statistically significant effects. Hence, farmers with larger farms, off-farm income, and corn area were more likely to share wheat with their relatives in the same town. On the other hand, older farmers were less likely to provide wheat for this motive. Farmers with access to broadband internet, those who hire labor, and those whose agricultural decisions are more influenced by their own experiences were also less likely to invest in altruism. Farmers who hire labor rely less on their relatives, and farmers who have broadband internet can access information and markets. Hence, these farmers might view investing in relationships within their immediate surroundings as less beneficial [66]. Being more influenced by their own experiences also signifies this.

For the intensity equation, the more a farmer’s agricultural production decisions are influenced by non-farming neighbors, the more he or she contributes to the altruism aspect of the social capital. The opposite is true for extension programs and one’s own experience. By relying more on one’s own experiences and on extension programs, farmers are less willing to invest and share with other people. Growing barley, which farmers in the region may use to feed their own animals [67], had similar influence. Hence, this study’s results indicate that independent farmers, with respect to resources and information about their surroundings, might invest less in the altruism motive. Our results are similar to the results from Ellerbrock and Groover [68], who found that farmer independence does not lead to occurrence of a socially optimal outcome.

In column C, growing corn had the largest average marginal effect, followed by the off-farm income variable. In column B, growing barley had the largest statistically significant average marginal effect. Farmers who grow barley spared approximately 4.85 kg less wheat to the altruism motive than farmers who do not grow barley. On the other hand, one’s own experience was the only statistically significant variable in all three columns.

5.3. Goodwill Motive

For the goodwill motive of social capital, that is the amount of wheat given to the town official to show goodwill,

Table 4. Results of the IV–Cragg’s hurdle model (exponential) for social capital motive of goodwill ¹.

Variables	Intensity Equation		Participation Equation		Marginal Effects
	Coeff.	Std. Err.	Coeff.	Std. Err.	(A)	(B)	(C)
	(Bootstrap 2)		(Bootstrap)		$\frac{\mathit{d}\mathit{E}\left(\mathit{y}\right)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{E}(\left.\mathit{y}\right|\mathit{y}>0)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{P}\mathit{r}(\mathit{y}>0)}{\mathit{d}\mathit{X}}$
Farm Size	0.000	0.000	0.000	0.000	0.003	0.001	0.000
Land Rented	0.047	0.087	0.131	0.236	1.077	0.755	0.043
Age	−0.003	0.003	−0.010	0.012	−0.077	−0.042	−0.003
Education	0.030	0.036	0.042	0.107	0.468	0.488	0.014
Off-Farm Income	−0.038	0.091	0.527 *	0.299	2.495	−0.617	0.173 *
Owns Cattle	−0.085	0.119	0.432 *	0.245	1.603	−1.383	0.142 *
Owns Sheep and Goats	0.022	0.094	−0.195	0.247	−0.861	0.350	−0.064
Grows Barley	−0.032	0.132	0.158	0.298	0.580	−0.520	0.052
Grows Corn	−0.031	0.104	0.792 ***	0.306	3.972 **	−0.495	0.260 ***
Grows Alfalfa	0.349 ***	0.103	0.228	0.364	4.068 **	5.661 ***	0.075
Cooperative Membership	−0.044	0.152	−0.633 *	0.378	−3.734 *	−0.719	−0.208 *
Broadband Internet	−0.030	0.092	−0.459	0.293	−2.687 *	−0.479	−0.151 *
Cell Phone Internet	0.176	0.126	0.239	0.249	2.713 *	2.861	0.078
Other Farmers	0.049	0.033	−0.005	0.095	0.377	0.799	−0.002
Non-farming Neighbors	0.080 *	0.046	−0.163 *	0.095	−0.213	1.303 *	−0.054 *
Extension Programs	−0.055 *	0.033	−0.012	0.089	−0.513	−0.892 *	−0.004
Banks	−0.066 **	0.033	0.020	0.077	−0.436	−1.077 **	0.007
Own Experience	−0.062	0.082	0.037	0.204	−0.308	−1.000	0.012
Hired Labor	−0.198 **	0.082	−0.339	0.223	−3.421 **	−3.208 **	−0.111
Asset Ownership	−0.089	0.168	0.004	0.358	−0.705	−1.436	0.001
Constant	3.104 ***	0.460	0.050	1.394
N	218
Pseudo R2	0.110
σ	0.339 ***
Chi2(20)	83.18
p-Value for Chi2	0.000
Log pseudolikelihood	−360.025

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the goodwill motive of social capital in continuous format for the outcome model and binary format for the selection model. ² Bootstrapped standard errors are based on 100 replications. *, **, *** indicates statistical significance at 10%, 5%, and 1% levels, respectively.

reports the results for the Cragg’s hurdle model with an instrumental variable. The model was statistically significant at the 1% level with an R-squared value of 0.11. The selection/participation coefficient σ was statistically significant at the 1% level. Hence, ignoring the selection/participation procedure would result in biased results. For the participation equation, farmers with off-farm income were more likely to spare wheat to spur goodwill. Similarly, farmers who own cattle and grow corn were more likely to invest in goodwill. Because town officials provide a link between government agencies and local residents, farmers contact these officials to acquire the legal documents needed to apply for various government programs, use pasture land, and conduct legal transactions [69]. Those with earned or unearned off-farm income get legal documents from the town official. Therefore, farmers who rely on town officials are more likely to invest in creating goodwill with those officials than farmers who do not need to interact with local officials. In the study region, the Turkish government has established support programs for cattle producers to fund infrastructure investment and increase the number of cattle per farmer [70]. Corn is irrigated using pressurized irrigation systems, and government programs exist to fund irrigation system establishment costs [71]. In the survey conducted for the current study, 95 percent of the corn growers indicated that they use pressurized irrigation systems. Because farmers who own cattle and grow corn may rely on a town official to apply for government support programs, they may be more likely to invest in building goodwill.

Cooperative membership had a negative and statistically significant effect on the likelihood of sparing wheat for goodwill. Previous studies found that social capital among cooperative members could decrease during a co-op’s lifecycle [72,73,74]. Hence, this study’s result supports the previous studies. Also, cooperative membership could provide farmers with technical and financial help and make them less reliant on town officials and government programs [75]. The more a farmer’s agricultural decisions are influenced by non-farming neighbors, the less likely he or she is to spare wheat for a town official. Being influenced by non-farming neighbors could mean that these farmers connect with information sources outside of their social groups or town, defined as “cosmopolite information channels” by Rogers [47], and could make them less reliant on town officials.

For the intensity equation, growing alfalfa had a statistically significant positive effect on the amount spared for the goodwill motive. Government programs have provided greater financial support for alfalfa than grains due to increased animal feed demand and Turkey’s status as an animal feed importer [70,76]. Because alfalfa is irrigated in the study region, farmers are also expected to receive funding for irrigation systems [71,77]. Hence, alfalfa growers are more likely to rely on town officials and invest more in building goodwill with those officials. The more a farmer’s agricultural decisions are influenced by non-farming neighbors, the more he or she spares wheat for the goodwill of the town official. We need further research to specifically identify the influence of non-farming neighbors.

The variables of extension programs and banks had negative and statistically significant effects. The more a farmer’s agricultural decisions are influenced by extension, the less he or she spares wheat for the goodwill motive. In general, farmers who are extension beneficiaries are more likely to apply for government programs and rely on town officials. However, education could also make them more productive and less reliant on government support programs and, ultimately, town officials [78,79,80]. This point requires future research. The more a farmer’s agricultural decisions are influenced by banks, the less he or she spares wheat for the goodwill motive. Farmers whose agricultural decisions are influenced by banks might rely less on government programs and, therefore, town officials. Lastly, having hired labor had a negative and statistically significant effect on the wheat spared for goodwill. Hired labor is expected to make the farmer less dependent on the town official, as agricultural hired laborers generally come to the Konya region from southeast Turkey as seasonal migrant workers [54,55]. Typically, seasonal migrant workers receive daily pay without social security, so farmers who hire labor are less likely to process official documents through town officials than farmers who do not hire labor. Farmers who do not hire labor are likely to receive government support as a family farm and rely more on the town official. Lastly, as shown in the average marginal effects section of Table 4, growing corn had the largest average marginal effect in column C, followed by the cooperative membership variable. On the other hand, growing alfalfa had the largest average marginal effect both in columns A and B.

5.4. Belonging Motive

Table 5. Results of the IV–Cragg’s hurdle model (exponential) for social capital motive of belonging ¹.

Variables	Intensity Equation	Participation Equation	Intensity Equation	Participation Equation	Marginal Effects
	Coeff.	Std. Err.	Coeff.	Std. Err.	(A)	(B)	(C)
	(Bootstrap 2)		(Bootstrap)		$\frac{\mathit{d}\mathit{E}\left(\mathit{y}\right)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{E}(\left.\mathit{y}\right|\mathit{y}>0)}{\mathit{d}\mathit{X}}$	$\frac{\mathit{d}\mathit{P}\mathit{r}(\mathit{y}>0)}{\mathit{d}\mathit{X}}$
Farm Size	−0.001	0.000	0.000	0.000	−0.003	−0.006	0.000
Land Rented	−0.083	0.084	−0.115	0.288	−2.281	−2.006	−0.035
Age	0.005	0.004	−0.007	0.011	0.033	0.127	−0.002
Education	0.030	0.042	−0.103	0.125	−0.265	0.734	−0.032
Off-Farm Income	0.007	0.113	0.387 *	0.215	3.062	0.169	0.119 *
Owns Cattle	−0.073	0.125	−0.027	0.255	−1.454	−1.773	−0.008
Owns Sheep and Goats	0.094	0.102	−0.447 *	0.264	−1.793	2.287	−0.138 *
Grows Barley	−0.110	0.120	−0.086	0.376	−2.533	−2.672	−0.027
Grows Corn	−0.080	0.116	0.697 **	0.282	3.942	−1.934	0.215 ***
Grows Alfalfa	−0.004	0.143	0.131	0.425	0.934	−0.092	0.041
Cooperative Membership	0.080	0.150	−0.311	0.351	−0.997	1.942	−0.096
Broadband Internet	0.147	0.099	−0.013	0.233	2.399	3.555	−0.004
Cell Phone Internet	0.057	0.138	−0.119	0.286	0.076	1.390	−0.037
Other Farmers	−0.010	0.039	−0.057	0.113	−0.606	−0.241	−0.018
Non-farming Neighbors	0.096 **	0.049	−0.066	0.095	1.128	2.324 *	−0.020
Extension Programs	0.002	0.033	−0.011	0.096	−0.042	0.061	−0.003
Banks	0.006	0.034	0.086	0.097	0.748	0.133	0.027
Own Experience	0.145	0.093	0.054	0.205	2.887	3.522	0.017
Hired Labor	0.082	0.111	0.270	0.354	3.450	1.988	0.083
Asset Ownership	0.046	0.163	−0.219	0.353	−0.881	1.120	−0.068
Constant	1.810 ***	0.516	1.224	1.162
N	218
Pseudo R2	0.041
σ	0.503 ***
Chi2(20)	34.21
p-Value for Chi2	0.025
Log pseudolikelihood	−561.051

¹ Dependent variable: amount of wheat (kg) out of 100 kg spared for the belonging motive of social capital in continuous format for the outcome model and binary format for the selection model. ² Bootstrapped standard errors are based on 100 replications. *, **, *** indicates statistical significance at 10%, 5%, and 1% levels, respectively.

reports the results from the Cragg’s hurdle model with instrumental variables used, which are the transparently supplied wheat stocks to a central pool for community efforts. The results for the model without an instrumental variable are available upon request. The model was overall statistically significant at the 5% significance level with an R-squared value of 0.041. For the participation equation, the variables of growing corn and off-farm income had positive and statistically significant effects on a farmer’s likelihood to spare wheat for the belonging motive. Because farmers grow corn to market it and contributing wheat would be declared to everyone in the town, corn growers might value villagers’ opinions and want to gain their recognition to feel a sense of belonging [81]. Moreover, corn growers may want to signal that they are trustworthy and pose less risk in market transactions by investing in social capital [82]. In the study region, farmers with off-farm income most frequently have unearned income, such as retirement income [62].

Farmers with off-farm income might want to have a good reputation in town through social recognition, so that they can obtain farm help from others through “imece” (i.e., do-it-together culture) [81]. However, this point needs further research. Lastly, owning sheep and goats had a negative and statistically significant effect on the probability that a farmer invests in the belonging motive. As mentioned previously, sheep and goats are traditionally raised in Turkey with less reliance on outside resources [53]. Hence, sheep and goat farmers might place less value on social recognition.

For the intensity model, the only statistically significant variable was non-farming neighbors with a positive coefficient. Strength of connection with non-farming neighbors could stem from showing sympathy for townspeople and earning recognition for that [81,83,84]. Investment in the belonging motive could also relate to being part of a religious sect, which may influence members’ economic and social decisions [85]. However, this point requires future research. For the marginal effects shown in Table 5, growing corn had the largest average marginal effect on a farmer’s probability of sparing wheat for the belonging motive in column C. On the other hand, the only statistically significant variable in column B is that of non-farming neighbors.

For all social capital motives, alternative parametric and semi-parametric models check the results’ robustness. The results for the parametric models are presented in the Appendix A Table A1, Table A2, Table A3 and Table A4, and the results for the semi-parametric quantile regression are available upon request. Parametric alternative models include Cragg’s hurdle model with linear specification, Heckman sample selection model with two-step, and maximum likelihood estimation procedures. Cragg’s hurdle model with exponential specification had better overall fit than the other models. For the semi-parametric models, we used simultaneous-equation quantile regression models, where different quantiles are estimated simultaneously. This model has the advantage of not making the normality assumption of the error, but it does not account for the two-step decision-making process existing in this study’s data. Also, because we had piled data on certain values, such as zero, some quantiles did not have enough observations to be estimated. However, our results are robust overall across parametric and semi-parametric alternative models.

5.5. Comparison across Social Capital Motives and Policy Implications

To compare the relative importance of variables’ influence across social capital motives,

Table 6. Comparison of the average marginal effects across social capital motives.

Variable	Probability to Invest 1				Actual Amount Invested 2
	Ethics	Altruism	Goodwill	Belonging	Ethics	Altruism	Goodwill	Belonging
Farm Size	0.000	0.000 *	0.000	0.000	−0.003	0.001	0.001	−0.006
Land Rented	0.048	−0.042	0.043	−0.035	−4.218	−2.088	0.755	−2.006
Age	−0.005 *	−0.008	−0.003	−0.002	0.039	0.001	−0.042	0.127
Education	0.028	0.004	0.014	−0.032	0.287	−0.435	0.488	0.734
Off-Farm Income	0.017	0.184 ***	0.173 *	0.119 *	−2.806	0.143	−0.617	0.169
Owns Cattle	0.101	0.100	0.142 *	−0.008	−2.270	0.878	−1.383	−1.773
Owns Sheep and Goats	−0.174 ***	−0.080	−0.064	−0.138 *	0.096	0.323	0.350	2.287
Grows Barley	0.102	0.066	0.052	−0.027	−2.283	−4.849 **	−0.520	−2.672
Grows Corn	0.219 ***	0.214 ***	0.260 ***	0.215 ***	−2.406	−2.962	−0.495	−1.934
Grows Alfalfa	0.223 **	0.137	0.075	0.041	2.549	1.697	5.661 ***	−0.092
Cooperative Membership	−0.044	−0.069	−0.208 *	−0.096	−0.942	−2.677	−0.719	1.942
Broadband Internet	−0.082	−0.121 *	−0.151 *	−0.004	4.437	0.721	−0.479	3.555
Cell Phone Internet	−0.002	−0.067	0.078	−0.037	−0.887	0.374	2.861	1.390
Other Farmers	−0.016	0.023	−0.002	−0.018	−0.857	0.419	0.799	−0.241
Non-farming Neighbors	−0.030	−0.023	−0.054 *	−0.020	0.467	1.372 **	1.303 *	2.324 *
Extension Programs	−0.007	−0.003	−0.004	−0.003	0.424	−0.807	−0.892 *	0.061
Banks	0.029	0.004	0.007	0.027	0.376	−0.245	−1.077 **	0.133
Own Experience	−0.043	−0.150 **	0.012	0.017	−2.856 **	−1.433 *	−1.000	3.522
Hired Labor	−0.131 **	−0.134 **	−0.111	0.083	3.222	0.195	−3.208 **	1.988
Asset Ownership	0.057	−0.062	0.001	−0.068	2.072	−1.340	−1.436	1.120

*, **, *** indicates statistical significance at 10%, 5%, and 1% levels, respectively. ¹ “Probability to Invest” is the probability of sparing wheat for the corresponding social capital motive ($dPr\left(y>0\right)/dX)$. ² “Actual Amount Invested” is the actual amount of wheat spared for the corresponding social capital motive ($dE(\left.y\right|y>0)/dX)$.

compares the average marginal effects for probability of sparing wheat and actual amount of wheat spared for social capital motives. For probability of sparing wheat, growing corn had a positive and highly statistically significant effect on all social capital motives. Hence, growing corn is a very influential variable across all social capital motives. Next, off-farm income had a positive and statistically significant effect on three social capital motives. Owning sheep and goats had negative statistically significant effect on ethics/promise-keeping and belonging motives. This raises concern for policy, as the Ministry of Agriculture and Forestry provides significant support to increase the number of sheep and goats per farmer and the number of producers [70]. Hence, these support programs should be accompanied with efforts to increase and sustain social capital among farmers and promote sustainable agriculture.

Broadband internet access had negative statistically significant effects on altruism and goodwill motives. Policies to enhance rural broadband internet infrastructure should incorporate a social capital motive. For the amount of wheat spared for social capital motives, the non-farming neighbors variable had positive and statistically significant effects on three of the social capital motives. Hence, by creating linkages between farmers and non-farmers, such as through agritourism activities, the government can increase social capital among farmers and contribute to sustainability.

6. Conclusions

Accomplishing community-based efforts requires cooperation among individuals willing to contribute to the collective goals of the community. Using household survey data and a farmer survey in Turkey’s Konya region, this study investigated the factors contributing to social capital motives and the implications for involvement in community-based initiatives. While some Turkish government programs and nongovernmental organizations provide financial support and training to establish farm cooperatives, establishing a culture rooted in working together and helping one another is critical to achieve cooperation at the community level.

Our results indicate that the more a farmer relies on himself or herself and resources available outside of the community, the less likely the farmer will be to invest in community efforts. This could explain why some previous studies found a social capital decline in countries over time. The more isolated or self-sufficient individuals become, the less invested in their own social capital and thus the less investment in community social capital. A lack of individual (farmer) interest in building one’s own social capital could especially be a problem for community-based efforts, such as managing common properties, enhancing community infrastructure, and reaching community consensus toward local policies.

To maximize the chance of accomplishing goals, the institutions, organizations, or firms need local infrastructure and policies to align with their goals for individual farmers. It is more efficient for the local farmers to contribute toward community priorities and needs than for outsiders to coordinate community change. Recent trends toward digitalization allow institutions, organizations, and firms to consider messaging that boosts individual farmer social capital as a way to enhance desired community outcomes. Otherwise, a focus on independent thinking could decrease social capital in rural areas and yield long-term negative outcomes.

We found evidence that the agriculture industry’s structure influences relationships and social capital of the individual. Hence, a region’s type of agricultural production should be considered while targeting the building of community social capital. Our results indicated that cooperative membership may not be a good proxy for individual social capital buy-in. Hence, programs to promote regional social capital could focus on production activities (e.g., joint equipment use or marketing efforts).

With respect to future research, additional work should connect social capital attributes with farmers’ use of sustainable agricultural practices. It could identify how different social capital motives influence farmers’ use of conservation practices. Another important extension involves analyzing the influence of family size and structure (i.e., number of children or dependents) on farmers’ decision to invest in social capital motives. Lastly, conducting an experimental study among farmers to measure social capital attributes would be beneficial for internal validation of farmers’ endowment allocation across different social capital motives and for their own use.