Socio-economic appraisal of small and marginal farmers of Jammu for sustainable livelihood

Introduction

In the state of Jammu, the agriculture and allied sectors contribute to 25.81% of the Gross State Domestic Product (Govt. of Jammu, 2023). Nearly 70% of the population directly or indirectly rely on this sector for their livelihoods. The predominant cropping systems in the Jammu region are rice-wheat (63.15%) and maize-wheat (76.41%) for irrigated and rain-fed areas, respectively. Other farming activities encompass various combinations of mono or multiple cropping, horticultural crops, agro-forestry, livestock, poultry, fishery and goat/sheep rearing (Bazaz, 2013) [3] . However, comprehensive studies on farming systems in the Jammu division are currently scarce. With a growing emphasis on the farming system approach, it became evident that understanding primary farming systems, their economics and resource productivity of crop and non-crop enterprises in this region is crucial. Such insights would enable both academicians and policymakers to formulate targeted, region-specific recommendations, as opposed to generic ones.

Agriculture, the backbone of Jammu and Kashmir's economy, supports a substantial 80% of its population. The state is divided into three agro-climatic zones: Jammu, Kashmir and Ladakh. Each of these zones has distinct geo-climatic conditions that dictate cropping patterns and productivity levels. While rice dominates in Kashmir zone, with maize, barley and wheat as secondary crops, the Jammu region excels in both maize and wheat production (Kaloo & Choure, 2015) [9] . Jammu's terrain comprises a mix of plains bordering Punjab and hilly areas. According to the Agriculture Census of 2020-2021, maize and wheat production prevails in the Jammu region, accounting for about 65% of the total area and contributing significantly to overall production. Despite yields that might not be exceedingly high, the Jammu region plays a substantial role in the national production of other cereals and millets. This is particularly remarkable given the dense population, resulting in limited per capita arable land. Prior to commencing farming system research activities, a baseline survey was conducted in 2019-2020. This survey aimed to comprehend existing crops, varieties, cropping patterns, input usage, production costs, socioeconomic and agro-climatic conditions, challenges and potentials affecting the present farming systems. The insights gained from this survey paved way for a well-informed research program, targeting increased productivity, reduced environmental degradation, sustainable land use and an enhanced quality of life for farming families.

Materials and Methods

Study area and Sample size:

The present study was carried out across organic, inorganic and integrated farming systems (IFS) located in Jammu, Samba, Reasi and Udhampur districts of the Jammu division. A total of 480 farmers participated in the study. Specifically, 160 farmers from organic, inorganic and integrated farming systems (IFS) (20 farmers from each of the 8 selected villages) were sampled for characterization. Necessary measures were taken to ensure the unbiased collection of data from the selected farmers. Face-to-face interviews were conducted using a recall-based approach. Additionally, focus group discussions (FGDs) were held in each of the selected locations to further enrich the insights obtained. Primary data was collected over a four-year period from 2018 to 2022, encompassing three rice crop seasons including. The data collection spanned various farming operations, including livestock rearing, poultry farming, homestead enterprises as well as non-farm activities. In addition to primary data collection, secondary sources of information were also consulted. Handouts, reports, publications, notifications and relevant documents related to this study were utilized to enhance the understanding context of the research.

Climate: Jammu experiences a humid subtropical climate (Javid et al., 2024) [7] . The region witnesses extreme summer temperatures that can reach as high as 46 °C (115°F), while winter temperatures occasionally drop below 4 °C (39°F). The hottest month is June, with average highs around 40.6 °C (105.1°F), while January stands as the coldest month, with average lows reaching approximately 7 °C (45°F). Annual precipitation averages around 42 inches (1100 mm), with majority of rainfall occurring between June and September. However, even during winter months, the area can experience significant rainfall. During winter, dense smog can lead to inconveniences and temperatures plummet to as low as 2 °C (36°F). Subsequent to the scorching hot season, the monsoon arrives, bringing heavy downpours accompanied by thunderstorms. Rainfall during this period can accumulate to approximately 669 mm (26.3 in) in the wettest months.

Characterization survey: Surveys were conducted between 2018 and 2022 to gather data for the study. In each chosen village, a minimum of 20 households were randomly selected using the transect method for individual interviews, resulting in a total of 480 participating farmers. Due to the ethnic diversity in the area, local translators were recruited from the community to facilitate effective discussions and interactions with the farmers. Once the research objectives were presented to the farmers, data collection took place using a semistructured questionnaire. The questionnaire covered a range of topics including socio-demographic and economic characteristics of the agricultural respondents (such as age, gender, education, household size, livestock, irrigation, years of experience in various farming systems like organic, inorganic and integrated farming, agricultural training, household income sources, farm resources and involvement in farmers' associations). Other areas of focus included the crop production system, cultural practices (such as area under cultivation, number of crop plots, types of cultivated crop varieties, bird control, frequency of fertilizer applications, labor types, weeding frequency, yield and the number of oxploughs used), and production constraints. Alongwith individual surveys, the altitude and geographical coordinates of the crop fields were recorded using GPS (Global Positioning System). The information gathered from the surveys was subjected to descriptive and multivariate statistical analysis. The sociodemographic profile of the surveyed rice farmers and the farm characteristics underwent Pearson chi-square test and ANOVA using IBM SPSS version 23.0 statistical software. This analysis aimed to compare the variations among the different surveyed regions. The significance level was set at 0.05, and means were differentiated through the Student Newman Keuls test (Newman, 1939) [12] . The classification of crop farming systems in the study area was carried out in two main steps: First, a Factorial Analysis on Mixed Data (FAMD) was conducted, resulting in an intermediate data representation. Subsequently, a Hierarchical Cluster Analysis (AHC) was performed based on the representative factors extracted from the FAMD. Discriminant variables of the generated clusters were identified through canonical discriminant analysis. The differentiated crop farming systems were described and compared using the final fit package. To visualize the distribution of rice farming systems across the surveyed villages, GPS data from rice fields were utilized. The collected data formed a matrix with 480 rows representing the surveyed crop farmers and 23 columns representing variables of both quantitative and qualitative nature (Table 1). This matrix was analyzed through crosssorting between the variables of interest (such as rice and wheat) and the other 28 variables, using the approach proposed by Xie and Ermon (2019) [18] . This approach entails providing means and standard deviations for continuous quantitative variables, frequency tables for discrete qualitative variables and conducting univariate tests on each variable. The model comprised twenty-three explanatory variables, and initially, the variance inflation factor (VIF) was assessed for each variable to gauge collinearity. As per Hossain et al.

(2015) [6] VIF values between 0 and 5 indicate no significant multicollinearity, values between 5 and 10 imply moderate multicollinearity and values exceeding 10 indicate high multicollinearity among predictors. Considering the presence of collinearity in numerous explanatory variables, a stepwise variable selection approach was employed prior to model adjustment to mitigate collinearity issues (correlations) between these variables. Here, π(x) signifies the probability of farmers adopting improved rice and wheat varieties based on their socio-cultural characteristics. This probability was expressed as a function of socio-cultural characteristics using the formula (Eq. 1):

The coefficients α0, α1,..., α15 were estimated using the maximum likelihood method through R software version 4.0.3. The independent variables (x1, α2,..., x15) are described in Table 1. The extent of susceptibility (likelihood) to adopt improved varieties, based on the selected factors, was gauged using odds ratios. Variables that demonstrated significant impact within the final model were identified via an overall test on the model.

Resource Use Efficiency Analysis

To assess the resource productivity of different farming systems for enhancing farmers' economic conditions and to evaluate the specific factor's contribution in conjunction with other factors that drive changes in output levels, a multiple regression analysis was employed. The Cobb-Douglas production function, presented below, was fitted to the data:

Marginal Value Productivity (MVP)

The productivity of different inputs used in basmati rice production was examined by calculating the marginal value productivities of inputs, which will be estimated at geometric mean levels of inputs.

Where, MVP (Xi) is the marginal value productivity of i th resource. bi is the regression Coefficient (estimated).

GM(Y) is the Geometric Mean of Output (yield). GM(Xi) is the Geometric Mean of the Input

Results and Discussion

The distribution of various inputs within the total variable cost reflects the cost structure, which is influenced by the level of technology and utilization of modern inputs. Disparities in input usage across different farm sizes highlight discrepancies in the level of technology implemented on each respective farm. Understanding the cost and return structure is essential for evaluating the necessity of interventions to bridge gaps and enhance the overall economic viability of farming (Vroegindewey and Hodbod, 2018) [17] . Regarding the structural characteristics of agricultural farms, it's notable that men (71.15%) were predominant in crop production within the study area. Among the surveyed farmers, a significant proportion (63.54%) lacked formal education, with an average age of 40.85 years farmers in the Jammu region exhibited less experience in crop production compared to those in other regions (Table 2). Additionally, household sizes were significantly larger in the Samba, Reasi and Udhampur regions compared to the Jammu region. On average, surveyed farmers cultivated an area of 0.9 hectares in Udhampur region featuring the largest plots sown on average. Agriculture stood as the primary income source for the surveyed farmers across all regions. While a majority (73.05%) owned the land where they cultivated rice and wheat, there remained a portion (21.2%) facing issues of land access and resorting to cultivating on rented lands. There were also limited instances of community-owned cultivation plots. In Jammu and Samba regions, crop production relied heavily on family labor, while in the Reasi region, a substantial portion of farmers (34.4%) hired external workers. The scarcity of farm mechanization was evident in all regions. Factors affecting the utilization of improved rice-wheat systems were observed among farmers who adopted at least one improved rice-wheat variety. These farmers were notably older (p < 0.05) and belonged to larger households when compared to those who didn't adopt any improved varieties at all (refer to Table 3). The decision to utilize at least one improved rice-wheat variety was significantly (p < 0.05) associated with factors such as off-farm income, hired farm labor, farmer training in agriculture, membership in agricultural associations, prioritizing rice as the main crop, fertilizer usage, interaction with government extensions, NGOs, and international institutes, as well as the region in which the farmers were situated. The yield of basmati rice from marginal farms was subjected to regression analysis against various production factors, namely human labor, machine labor, seed, manure & fertilizer, plant protection chemicals, and irrigation charges. These factors were considered as explanatory variables. Upon examining the data presented in Table 4, it becomes evident that the production function for basmati rice on small farms, with an R 2 value of 0.97, was highly statistically significant. This signifies that 97 percent of the total variation in yield could be attributed to the aforementioned variables. Furthermore, the regression analysis indicated that the coefficients of human labor and manure & fertilizers were significantly significant at a probability level of 1 percent. Additionally, the coefficient for seed was significant at a 5 percent probability level, with values of 1.087, 0.378, and 0.378, respectively. Conversely, machine labor and irrigation charges exhibited negative significance at the 5 percent and 1 percent levels, respectively, with corresponding values of -0.412 and -0.313. The coefficient for plant protection chemicals, on the other hand, was negatively non-significant. The analysis also revealed that the marginal value productivity of human labor, seed and manure & fertilizers were positive, with values of 6.253, 19.456 and 5.236, respectively. Conversely, the marginal value productivity estimated for machine labor, plant protection chemicals, and irrigation charges was negative, with values of -4.056, -0.423 and -13.256, respectively.

The various components comprising the per-hectare cost of cultivation are detailed in Table 5. The total variable cost for the sample farms, encompassing costs associated with postharvest operations, amounted to Rs. 37,510 per hectare. Within this, variable input costs accounted for approximately 97 percent, while the remaining 8 percent constituted postharvest/marketing expenses. When analyzed by category, the lowest cost of Rs 36,655 per hectare was observed on medium-sized farms, while the highest cost of Rs 39,267 per hectare was observed on marginal category farms. This indicates that as farm size increases, the cost of basmati production and marketing tends to decrease. The proportionate share of variable input costs in the total cost was fairly consistent across all farm categories, ranging from 95.57 to 96.28 percent across all groups. Labor costs contributed significantly to the total variable input cost, amounting to Rs. 18,520 on marginal farms, Rs. 19,260 on small farms, and Rs. 19,835 on medium farms, respectively. Machinery costs, on the other hand, totaled Rs. 6,691 per hectare overall, with values of Rs. 6,750 on medium farms, Rs. 6,620 on small farms, and Rs. 6,580 on marginal farms, respectively. Similarly, pesticide/weedicide costs averaged Rs. 2,373 per hectare across all farms, with figures of Rs. 2,415 on medium farms, Rs. 2,230 on marginal farms, and Rs. 2,305 on small farms. Notably, the usage of labor intensifies as farm size decreases (Nene 2005) [11] . Farmers practicing rainfed and irrigated lowland rice farming tend to cultivate improved varieties compared to those practicing rainfed upland rice pro-ductions. Moreover, farmers practicing irrigated lowland rice production used more sowing than transplantation. Little difference was observed between the proportion of farmers practicing the different rice production systems in terms of ploughing, soil fertility management or pest management Table 6. In terms of the costbenefit ratio, the highest ratio was observed on marginal farms at 1.88, followed by 1.69, 1.63, and 1.61 for small, medium, and large farms, respectively. The average costbenefit ratio across all farms was 1.70.

Identification of Farming Systems in the Intermediate Zone

The data presented in Table 9 highlights that 43.2 percent of farmers were engaged in livestock-based farming systems, 56.8 percent were involved in cereal-based farming systems and a mere 0.6 percent were practicing fruit-based farming systems. This data underscores the prevalence of cereal-based farming systems in the intermediate zone of the Jammu region. The primary reason for farmers continuing to adopt livestock and cereal-based farming systems lies in the constraints posed by small land holdings. These limitations restrict their ability to transition to other enterprises. Farming in this region remains primarily subsistence-oriented, compelling farmers to cultivate crops that fulfill their family's needs. Notably, only one respondent possessed a substantial landholding, thereby adopting a fruit-based farming system. This demonstrates the potential shift to other profitable crops if farmers have larger land holdings that allow for increased income generation. This observation is consistent with Agrawal (2012) [2] , who found a predominance of cereal-based farming systems in the midwestern plains of the Jammu region. Similarly, Kumar et al.

(2012) [10] noted that the majority of households in the waterscarce region of Jammu were engaged in the wheat-mustard-based farming system. Livestock-based farming systems accounted for 26.62 percent of the total land holdings, indicating smaller land holdings compared to cereal-based farming systems. The data implies that respondents' land holdings directly influence their choice of farming system. Greater land holdings correlate with a higher likelihood of adopting cereal-based farming systems, while smaller holdings are associated with livestock-based farming systems. This relationship stems from the higher income generation potential of cereal crops when cultivated on larger expanses of land. The classification of farming systems was based on gross returns from various enterprises. Interestingly, fruit-based farming systems exhibited the highest land holding per household at 2.67 percent of the total land. This suggests that fruit-based farming is primarily practiced by larger farmers who possess ample land beyond fulfilling their family's needs. The number of enterprises varied across different farming systems, ranging from three to five. This finding aligns with the results of Prasad (2009) [13] , who observed similar variability in the number of enterprises within identified farming systems in the Jammu region.

Farm management practices by the sample farmers

The agronomic management includes variety selection, seed rate, seeding date, transplanting fertilizer application date, organic manures addition, weed control, water management, pesticide application, and harvesting. Agronomic managements of the cultivated crops have been indicated in the Table 10.

Cobb-Douglas Production Function for Resource Productivities of Major Farming Systems in Various Districts

The Cobb-Douglas production function was employed to assess the resource use efficiency of major farming systems in different districts of the Jammu region, as presented in Table 11. The production elasticity indicated that a one percent increase in the area under other crops (AOTH) across all districts led to a gross income increase ranging from 3.239% for livestock-based farming systems to 0.870% for cerealbased farming systems. Specifically, for Jammu, Doda and Reasi districts, this relationship was positively significant across all farming systems. However, in Kathua and Rajouri, the relationship was found to be non-significant. Regarding the elasticity coefficients for cereal-based farming systems across all districts, they were found to be positive but nonsignificant (0.378). This suggests that the contribution of dairy animals to increased gross income in cereal-based systems is negligible. On the other hand, for livestock-based farming systems, the elasticity coefficient was negative and non-significant (-0.853).

Annual income of sample farmers

It is observed from Table 12 that the average total income of the inorganic farms was Rs. 140211 and for organic farms, it was Rs. 126689. The income figure of inorganic farming is higher than the national average of Rs. 118542. Considering the farming systems, farmers practicing F-L-M-V system earned highest annual income Rs. 167085 under inorganic farming and in case of organic farming, farmers practicing F-L-M-V system got highest annual income Rs. 148521 than other farming systems. This result is slightly similar with Uddin and Takeya (2005) [16] who found that income from non-agricultural enterprises is higher income Rs. 61172 for farmers in organic farming compared to those of inorganic farming income of Rs. 84868 in respect of farm sizes. It was calculated that Change in total annual income no farm activities of the inorganic farms was increased by the amount of Rs. 22962 in the study areas due to intervention which is statistically significant at 10 percent level.

Sources of income Inorganic farming

Organic farming Avg. F-L-M-V L-R-W-P L-R-W-M L-W-R-P L-W-P F-L-M-V L-R-W-P L-R-W-M L-W-R-P L-W-P Farm activities

Food security index of the sample households

The food security index for inorganic farm households was 1.21, while for organic farm households, it stood at 1.29. This index indicated a higher level of food security for both inorganic farming (1.21) and organic farming (1.29) among households categorized as food-secure. In contrast, the index was 0.92 for inorganic farming and 0.85 for organic farming for households categorized as food-insecure. Both the headcount indices for inorganic and organic farm households in the study areas suggest a state of food security. Among the sampled households, 80 percent of inorganic farms and 85 percent of organic farms managed to meet the required daily calorie intake of 2186 kcal and 2241 kcal per capita, respectively. Conversely, 1.56 percent of households practicing inorganic farming and 1.21 percent of households engaged in organic farming were categorized as food insecure (refer to Table 13). Notably, the average calorie intake of food-secure households surpassed the national average of 2200 kcal. Specifically, for inorganic farming households, the average intake was 2793.83 kcal, and for organic farming households, it was 2910.76 kcal. The food insecurity gap/surplus index provides additional insight. Food-secure households exceeded the food poverty line by 0.06 percent in the case of inorganic farming and by 0.07 percent for organic farming. Conversely, food-insecure households fell short of the required calorie intake by -0.1 percent in the context of inorganic farming and by -0.09 percent for organic farming. Adekoya (2009) [1] made a significant discovery regarding the role of inorganic rice farming in boosting food production. Their research demonstrated that the inorganic farming system outperforms rice monoculture in terms of resource utilization, diversity, productivity, as well as the quality and quantity of the food produced.

Contribution towards food and nutritional security

The provision of food and nutritional security to farming families stands as a fundamental objective of integrated farming systems. These well-integrated and complementary farming systems play a pivotal role in addressing the dietary needs of farm households, either partially or in full, contingent on the extent of integration and the scale of the farm. These systems constitute the foundation of Indian agriculture, serving as the primary source of staples for countless farm families. The integrated approach of these systems, such as the incorporation of short-duration vegetable crops, pulses, and livestock feed, allows efficient utilization of land and time. Additionally, these systems enable diverse applications of water resources, encompassing activities like fish, duck, or poultry rearing alongside crop irrigation.

Recognizing their significance, these systems hold a critical position in realizing future food security and nutritional requirements for India's burgeoning population. A noteworthy example is the integration of fruits, vegetables, livestock, and mushrooms (F+L+M+V) as well as the integration of livestock, water resources, and poultry (L+W+P). These integration strategies, as depicted in Table 14, contribute significantly to fulfilling staple food needs for farm families. Additionally, the integration of homestead farming with livestock, such as the case in Jammu (1 ha), showcases a sustainable model capable of supporting a family of six members throughout the year. This model offers a diverse range of cereals, pulses, oilseeds, fruits, vegetables, milk, meat, eggs, and mushrooms, as documented by Kumar (2012) [10] . Indicators: H-High, M-Medium, L-Low

Conclusion

Among the 18 million households in Jammu, roughly 83% are small farmers, with some being landless (Rather 2024). These small farmers practice diversified farming to meet household needs and mitigate risks, making the most of their farming systems. Our research findings, spanning various locations within the state, underscore the greater advantages of integrated farming over solitary cropping systems for small and marginal farmers. Integration of crops, livestock, poultry, fruits, and vegetables has consistently demonstrated elevated rice grain equivalent yields, augmented net returns and heightened energy efficiency. Remarkably, these benefits have been achieved without compromising soil quality or system performance. Moreover, this integrated approach has emerged as an environmentally sound practice, leveraging resource recycling of farm waste to curtail reliance on chemical fertilizers, pesticides, and weedicides. The challenges faced by small and marginal farmers distinguish them from their larger counterparts. These farmers rely on agriculture to meet their household needs, often facing resource constraints and exhibiting limited education and technical acumen. Consequently, the development of suitable agricultural technologies becomes imperative for this demographic. Integrated farming emerges as a viable strategy, surpassing specialized cropping systems. It addresses the multifaceted demands of these farmers, contributing to their well-being and socio-economic advancement. These integrated systems, tailored to the socio-economic and agroclimatic conditions, ensure an interdependence of resources, enhancing sustainability, overall production and household income. In this context, achieving sustainable food production amid population growth, economic expansion, changing consumption habits, urbanization and climate change is a formidable challenge. Solutions must involve using less land, water, labor and chemicals, while focusing on improving productivity and well-being for small-scale farmers. The interplay between components within farming systems highlighted by Shyam et al., (2023) [15] and others as a means to enhance productivity reduce environmental degradation and improve the lives of impoverished farmers play an active role in identifying research problems and testing potential solutions, thereby contributing to meaningful progress. The outcomes derived from this study carry potential implications for both farmers and researchers, offering insights into agricultural diversification. Furthermore, the findings provide a valuable foundation for effective agricultural planning by government bodies and policymakers. As we look ahead, it is imperative that future research delves into the interplay between farm size, social dynamics, environmental outcomes and the well-being of laborers, farmers and consumers.

Acknowledgement

AcknowledgmentsThis study received funding from the ICAR-IIFSR (Indian Institute of Farming System Research) grant. The authors extend their gratitude to the Chief Agronomist, Senior Scientist, Principal Scientist, Technical, RA, SRF, and the other staff at the IFS Centre, SKUAST, Jammu. Additionally, the authors would like to express their appreciation to the chiefs of surveyed villages, heads of farmers' associations, and the agricultural producers who actively participated in this study. Their collaboration and willingness to share their knowledge have been invaluable.

References

1. Adekoya AE. Food insecurity and coping strategies among farming households in Oyo State, Nigeria. Journal of Food Agriculture and Environment. 2009;7(3-4):187191.
2. Agrawal A. Aggregating fragmented land holdings. Agriculture Today. 2012;15(3):49-50.
3. Bazaz NH. Crop diversification in Jammu and Kashmir: Pace, pattern and determinants. IOSR Journal of Humanities and Social Science. 2013;11:1-7.
4. Douglas P. Cobb Douglas production function. Quarterly Journal of Economics. 1928;42(3):393-415.
5. Government of Jammu. Economic Survey 2022-23. Jammu and Kashmir Government; 2023. p. 327.
6. Hossain S, Ahmed SE, Doksum KA. Shrinkage, pretest and penalty estimators in generalized linear models. Statistical Methodology. 2015;24:52-68.
7. Javid M, Ahmad K, Rahmani AR, Ahmad F, Nigam P. Species diversity and composition of macrophytes in Shallabugh Wetland, Ganderbal, Jammu and Kashmir, India. International Journal of Plant & Soil Science. 2024;36(6):436-443.
8. Kachroo J, Kachroo D. Economic analysis of fine rice production under subtropical agro-climatic zone of Jammu region of Jammu and Kashmir State. Agricultural Situation in India. 2007;8(2):413-417.
9. Kaloo MJ, Choure T. Present status and future prospectus of agriculture in Jammu and Kashmir. IOSR Journal of Humanities and Social Science. 2015;20(11):62-67.
10. Kumar M. Marginalized farmers and bank loans in J&K. Daily Excelsior. 2012 Feb 18.
11. Nene YL. Rice research in South Asia through ages. Asian Agricultural History. 2005;9(2):85-106.
12. Newman D. The distribution of range in samples from a normal population, expressed in terms of an independent estimate of standard deviation. Biometrika. 1939;31(12):20-30.
13. Prasad GR. India's food production: Review and paradigm shifts desired. Indian Journal of Fertilizers. 2009;5(12):55-56, 59-61.
14. Rather KA. Evaluation of household awareness and utilisation of Jammu and Kashmir Food Entitlement Scheme. Economic & Political Weekly. 2024;59(5):37.
15. Shyam CS, Shekhawat K, Rathore SS, Babu S, Singh RK, Upadhyay PK, et al. Development of integrated farming system model-A step towards achieving biodiverse, resilient and productive green economy in agriculture for small holdings in India. Agronomy. 2023;13(4):955.
16. Uddin MT, Takeya H. Economic analysis of integrated farming by agricultural enterprise in Mymensingh district of Bangladesh. Journal of Agricultural Development Studies. 2005;16(1):40-49.
17. Vroegindewey R, Hodbod J. Resilience of agricultural value chains in developing country contexts: A framework and assessment approach. Sustainability. 2018;10(4):916.
18. Xie SM, Ermon S. Reparameterizable subset sampling via continuous relaxations. arXiv. 2019;1901.10517.