A Way Forward to Promote the Farming Contracts Between Firms and Farmers in Cultivation Productions: A CASE STUDY Of VIETNAM

The purpose of this study is to assess the factors affecting the level of linkages between firms and farmers in crop development in Thai Binh province, Vietnam. The two main research methods used were exploratory factor analysis (EFA) and regression analysis with survey data which was conducted from 300 farmers, 10 firms, and 15 managers related to contract-farming. The results indicated that there are 5 factors that significantly affect the level of participation of farmers in the linking process, the order of impacts from strong to weak is as follows: (1) Commitment and sharing of benefits and risks between farmers and firms in the linking process, (2) Issues related to management skills and enterprises capacity of link firms, (3) Price issues that farmers are interested in in the process of implementing links with firms, (4) Policy environment Institutions related to farmer contracts between firms and farmers, (5) Household awareness about the benefits brought about by farmer contracts with firms. Based on results, a number of specific solutions have been proposed to enhance the efficiency of farming contracts between firms and farmers in cultivation production.


Introduction
Vietnam's agricultural sector is still small, fragmented and traditional agriculture. Therefore, cooperation and alignment in the industry are more necessary than ever (Tran Quoc Nhan, 2012). However, the results of linkage in Vietnam were not as expected. Farmers are not interested in joining the agricultural association (Quang Sang, 2019). In the total 50,000 firms involved in the agricultural sector, only about 1,000 are directly linked to farmers (Bao Loan & Duong Thanh, 2019). The farmer contract between firms and farmers was not strong and sustainable because there was no mechanism for sharing profits and risks (Nguyen Van Chi, 2020). The proportion of commodity agricultural products that are consumed through contracts between farmers and firms are still low (Quynh Anh, 2020).
Thai Binh is a Vietnamese administrative unit and a coastal province in the Red River Delta. By the end of 2018, Thai Binh ranked 13/63 provinces and cities in terms of population, ranked 29th in Gross Regional Domestic Product (GRDP) and ranked 8th in speed GRDP growth (Thai Binh Statistical Office, 2018). The area of food crops in 2017 was about 171,9 thousand hectares, of which the rice area was 158,7 thousand ha (General Statistics Office, 2017). However, the current implementation of farming contracts between firms and farmers in Thai Binh province has been still very limited.
The government perceives contract farming as an excellent opportunity to boost rural economic development (Hélène et  The above studies analyzed different topics of linkage in Vietnam. However, there are many research gaps that can develop: The process of links between business and farmers in each locality are different. On the other hand, to our understanding, there has not been any scientific and comprehensive and systematic research on how to link businesses and farmers in the development of crop production in Thai Binh province. Against this background, this study contributes to the literature in three ways. First, we generalized the situation of agricultural production linkages between farmer households and businesses in Thai Binh province. Secondly, we modeled the factors that influence people's participation and analyzed the impact of each of them. Thirdly, we formulate policy recommendations to improve linkage efficiency in production in Thai Binh and can be applied to other localities. Hirschman (1958) divided the concept of farmer contracts based on industry and interdisciplinary relationships: Links include backward linkages and forward linkages. The backward linkages arise from the demand for input supply of a newly established industry, while forward linkages arise from using that sector's output as the input of the pull-along activities.

Literature reviews
Contract farming, as an institution in agriculture, has a long history. Various forms of this institutional arrangement were employed by United States multinationals in Central America at the beginning of the 20th century, and by the Japanese to secure sugar production in Taiwan from 1885 (Runsten & Key, 1996).
During the period 1930-1950, contracting was used increasingly in many food and fibre sectors. The fruit and vegetable canning sectors expanded in the United States and Europe (Little & Watts, 1994;Clapp, 1994). From the late 1950s, Mexican growers increasingly supplied the American markets with fruit and vegetables under contract (Watts, 1994), and in the period 1960-80 there was a significant increase in contracting for vegetables, fruit, nuts and seed crops (Kilmer, 1986). Contract farming has also spread rapidly in Asia, Latin America and Africa owing to the higher returns earned by high-value export crops and the impact of new technologies (Clapp, 1994;Staatz & Eicher, 1998). Some places, forms of contracting are dependent on specific institutions such as marketing orders and bargaining cooperatives (Sporleder, 1992). The contract could specify the price, quantity, quality, the provision of agribusiness inputs (Sporleder, 1992;Runsten & Key, 1996). Reardon and Barrett (2000) have observed that the agricultural industrialization process in many developed countries has brought a number of effects in regulating the supply chains that make them more closely connected. Boehlje (2000) argues that in many developed countries, agricultural production is changing from an industry dominated by family-based, small-scale farms or firms to one of larger firms that are more tightly aligned across the production and distribution value chain. The signing of contracts allows farmers to overcome barriers to integration into the world economy (Carney, 1988;Clapp, 1994;Jackson & Cheacher, 1994;Little & Watts, 1994;Royer, 1995;Pasour, 1998;Delgado, 1999;Vellema, 2000). Contracting could also improve access to capital and credit (Hudson, 2000).
Contracting farmers can reduce production costs and increase production and income as a result of their use of new technology and their access to company inputs (Watts, 1994;Clapp, 1994). Contracting farmers can stabilise output of the product and stabilise income. At the same time, contracts may simplify production and marketing decisions, thus improving the farmer's effectiveness (Hudson, 2000). Contracting farmers can increase profit opportunities through a greater product range and differentiated products (Pasour, 1998), or by diversifying out of traditional crops in developing countries in order to grow high-value crops and thereby increase their income (Williams, 1985;Levin, 1988;Delgado, 1999). Saggi (2002), Giroud & Scott-Kennel (2006), Unctad (2001) have concluded that horizontal linkages shows the interaction between foreign and local enterprises in the production of goods and services in the same production phase. The most important effect is called the spillover effect ( Hussain & Planning (2000) have concluded that there are two main types of linkages. The first type is called economic link (E-farm contracts), including activities related to the production and trading of goods and services. The second type is called knowledge link (K-farm contracts), including human activities in creation and transfer of knowledge, thus creating the knowledge spreading effect. Mesquita & Lazzarini (2008) concluded that linkages have a combination of vertical and horizontal links, which appear in the development and integration process when the relationship between households, facilities and enterprise is a combination of cooperation and competition.
Otsuka et al. (2016) found that although contract farming contributes to the improvement of farmers' income by introducing new crops and production methods, there is room for strengthening its effects on poverty reduction through policy. Olounlade et al. (2020) found evidence of significant negative effects on rice production income at a 1% level. Hung Anh & Bokelmann (2019) found out formal institution brings better market access for coffee farmers, but main issues are regarding opportunistic behavior, imperfect market knowledge, traditional farming habits, and contract noncompliance, which have resulted in a lower preference for the market of processors/exporters.
Through the theory of contract farming, it can be seen that contract farming gives contract farmers many benefits such as easy market access and more stable market, the contract could specify the price, quantity, quality, the provision of agribusiness inputs. Through an economic contract, the parties must be responsible and obliged to implement the terms of the contract. These are also the scientific bases for us to build a hypothetical research model for this research.

Research methodology 3.1. Conceptual Research Framework
The process of developing the survey table and conducting data collection survey has been carried out in the following main stages.
To start this analysis, at first define the factors related to the level of participation of farmers in linkage with firms in Thai Binh province. There were six factors that were expected to affect the level of participation of farmers as follows: (1) Awareness of farmers about benefits of farmer contract with firms; (2) Commitments in the process of linking with firms; (3) Sharing benefits and risks between farmers and linked firms; (4) Management skills and business competence of linked firms; (5) Policy environment -Institutions related to the farmer contract between firms and farmers; (6) Price issues that farmers are interested in in the process of implementing farmer contracts with firms.
The second stage is that design of preliminary survey. After that discussion groups and expert interview to adjust the scale, complete questionnaires and determine the location and selecting interviewees. After official investigation, the collected data were entered into SPSS software and start processing data. Cronbach's Alpha test is constructed to examine the reliability for each component. Exploratory factor analysis is used for reducing the number of variables to a smaller set of underlying summary variable or component. The regression analysis helps to provide an estimation of the relationship among the variables and the impact of independent.

Data Collections
There are 31 independent variables and two dependent dichotomous variables that form a database. These data were obtained from 300 samples according to the following formula: If the model has m scales, Pj is the observed variable number of j-th scale. The ratio of the sample to an analytical variable (k) is 5/1 or 10/1. In this study, the author used 6 scales with a total of 31 variables, with the number of samples selected from 5-10 times the number of variables, so the total number of samples is from 155-310 samples.
This study used data from a panel survey of random 300 farmers in three selected districts of Thai Binh province (Dong Hung district, Vu Thu district and Kien Xuong district). The set of questionnaires was divided into two parts, the first part involved the respondent background such as gender, age, education level, occupations, monthly income, agricultural production area, major crops… while the second part included 5 Likert scale ranging from strongly disagree (1) to strongly agree (5) direct questionnaire about factors affecting the level of participation of farmers, such as: awareness of farmers, commitments in the linking process, sharing benefits and risks, policy environment and price. Besides, the author also interviewed in-depth 10 enterprise and 15 managers to find out information about markets and policies in the province.

Data analysis
The sample data were entered into computer software (SPSS). It is statistical software which eases to compile and analyze data. We can compile

Nguyen Van Song, Vu Ngoc Huyen, Nguyen Thị Thuy, Dinh Van Tien, Thai Van Ha, Nguyen Xuan Diep, Vuong Thi Khanh Huyen
or entry collected primary data or secondary as same as Microsoft Excel. Two main analyses in this study are Exploratory Factor Analysis (EFA) and Multivariate Regression Analysis.  Exploratory factor analysis is used for reducing the number of variables to a smaller set of underlying summary variable or component. This method is one of a statistical method to uncover the underlying structure of a relatively large set of variables (Norris & Lecavalier, 2010). In Exploratory Factor Analysis (EFA), firstly it is looking at the Kaiser-Meyer-Oikin (KMO) value that measure of sampling adequacy for the variables is close to 1.0 (Latif et al., 2012;Latif et al., 2018;Kaiser, 1974). The values more than 0.6 consider reliable as recommended by Nunnally (1978). According to Cristobal et al. (2007), the item with corrected item -total correlation lower than 0.30 are not acceptable. Similarly, the Bartlett's Test of Sphericity to see the sufficiency correlation items to proceed in the analysis when it is significant at p < 0.001, which means that there are some relationships between the items. Since both of the tests were significance, it is suggesting that the data is appropriate to proceed with data reduction procedure to group the items into acceptable components (Habidin, 2013). The number of factors is determined based on the Eigenvalue index (The value of the variance divided by each factor)representing the variance explained by each factor. According to Cerny & Kaiser (1977) factors with Eigenvalue less than 1 will be excluded from the model. Variance explained criteria: the total variance explained must be greater than 50%. Using Principal components extraction method with varimax rotation and stopping points when extracting elements with Eigenvalues greater than 1 with observed variables. Then, all items are loaded into respective factor with a minimum factor loading of 0.4 (Habidin, 2013;Hatcher, 1994). A reliability test is conducted to ensure all factors are accepted as being reliable for the research in order to enhance the accuracy of the assessment and evaluations (Habidin, 2013).
After completing the Exploratory Factor Analysis, the variables do not guarantee values

EFA analysis
Overview of theory Design of preliminary survey

Situation of farmer contract between firms and farmers in Thai Binh province 4.1.1. Links between farmers and farmers
According to Department of Agriculture and Rural Development of Thai Binh Province (2017), the province had 112 groups and affiliated groups in livestock production and aquaculture, 109 livestock groups, 3 aquaculture groups and 13 models of husbandry cooperation groups. Collaborative groups have linked on breeds, feed, veterinary drugs, animal husbandry processes, lower production costs, improved product quality, and more concentrated product creation. Some cooperative groups have signed contracts with input suppliers and outlets. A number of large-scale self-invested farm owners are forming a chain of production with agreements to supply animal feed, veterinary drugs and consumption of livestock products.
Links between farmers and farmers in Thai Binh Province have been a horizontal linkages type. Farmers have joined together to create areas of specialized production of crop products with a larger scale and larger area. These links have helped them to sign contracts directly with the companies supplying raw materials in large quantities, higher discount, and guaranteed of raw materials quality. Therefore, the product quality has been more uniform, input production cost has been more economical.

Links between firms and farmers, between cooperatives and farmers
Thai Binh province has 20 enterprises participating in linking with cooperatives and farmers; 4 livestock enterprises are linking with owners of livestock farms; 2 enterprises are linking with seafood production households. The form of these links are vertical linkages. Some typical link models in Thai Binh province: (1) Farmer contract model of Japanese rice production and high-quality rice. The area given for this farmer contract model accounted for 50.1% (4,892ha), being by far the largest one among farmer contract models.

Link between farmers and associations
Currently in Thai Binh province, the way of link between farmers and associations are often through contracts. In 2019, Thai Binh province had 80 communes in association with production with 20 companies and enterprises. Products contracted include: Rice seed, commercial rice, corn, millet, cucumber, squash, peanuts, spinach, potatoes...
Regarding the development of high-tech agriculture, there are some large corporations in renting land from farmers to develop clean agriculture, typically TH TRuemilk Group. TH TRuemilk invests in the production of clean and organic foods, such as rice, vegetables, tubers and fruits. According to plan, TH TRuemilk Group will invest in developing hi-tech agriculture project with scales up to 3000ha and total investment of 3,000 billion VND.

Assess factors affecting the farmer contracts between firms and farmers 4.3.1. Cronbach's Alpha Analysis
Cronbach's Alpha analysis results for the observed variables are described in the following table:

MT04
The role of the state and local authorities in promoting the farmer contract between farmers and firms (credit support, science and technology support) .645 .759

MT05 The role of associations in promoting farmer contracts between farmers and firms (Farmers' Union, Business Association,)
.444 .814 Price issues that farmers are interested in in the process of implementing farmer contracts with firms (GC), Cronbach's Alpha = 0.747 29 GC01 Farmers receive price support in case of crop failure or high price of inputs .615 .615 30 GC02 Linked firms offer reasonable buying prices to farmers .590 .590

GC03
The company guarantees to make the purchase price as committed and not reduce the purchase price in case of crop success .769 .769 The level of participation of farmers in the linking process (TG), Cronbach's Alpha = 0.925

TG01
Farmers are interested in associating with firms in production and consumption of products .872 .

33
TG02 Farmers are willing to link with firms .872 .

Source: Summary of the author (2019)
The summary results in Table 4 show that most of the scales had Cronbach Alpha coefficient in the range of 0.7-0.8 so the scales in this study were highly reliable. The results shown in Table 4 indicate that there were 3 observed variables excluded due to the failure of Cronbach Alpha coefficients and the total correlation (NT04, MT05, GC03).

Exploratory Factor Analysis (EFA)
After assessing the reliability of the scales, the author conducted an exploratory factor analysis to evaluate the convergence value and discriminant value of the independent and dependent variable scales. In this study, the KMO value was 0.909 (KMO > 0.7) and Bartlett's test of sphericity had a significant value = .000 (less than .05). These results suggest that factor analysis was suitable for these statements. In the extraction of the EFA, based on the initial eigenvalues greater than 1, the analysis output showed five components were the factors in this EFA. The percentage of explained variance was 77.563%. In all statements, the factor larger than 0.5 suggests that the statements can optimally explain variances of their related factors, and thus the questions are significant.
Thus, the factors after implementing EFA discovery factor ensure the ability to represent the original survey data and be eligible to perform multivariate regression analysis. Five factors were drawn from the implementation of the EFA analysis, including: Commitment and sharing of benefits and risks between farmers and firms in the linking process (X1), Household awareness about the benefits brought about by farmer contracts with firms (X2), Policy environment -Institutions related to farmer contracts between firms and farmers (X3), Issues related to management skills and enterprises capacity of link firms (X4), Price issues that farmers are interested in in the process of implementing links with firms (X5). Thus, the EFA factor analysis results have changed the group of variables compared to the original theoretical model.

Exploratory Factor Analysis for dependent variables
Result of the Kaiser Meyer Olkin value was .905 (KMO > 0.7) and Bartlett's test of sphericity had a significant value = .000 (less than .05). These results suggest that factor analysis was suitable for these statements. There are one dimension or component emerged from the EFA procedure based on the computed Eigenvalue >1.0. The extracted variance reached 93,617%. This has confirmed the ability to converge and perform well of the observed observation variables.

Pearson's Correlation Coefficients
Pearson's correlation coefficient helps to check the collinearity of independent variables. Table 5 shows the correlation analysis between the level of participation of farmers in the linking process and other factors resulted with the moderate positive and significant correlation at p < 0.01. The correlation coefficient of the level of participation of farmers in the linking process (Y) with the variable X1 has moderate correlation (r = 0.734, p < 0.01) which was followed by that of X2 (r = 0.515, p < 0.01), X4 (r = 0.488, p < 0.01), X5 (r = 0.465, p < 0.01), and X3 (r = 0.422, p < 0.01). Here the absolute value of Pearson correlation coefficient is less than 0.8, it shows collinearity is very less likely to exist.

Multivariate regression analysis
Multivariate regression analysis was performed with 5 independent variables and the results of multivariate regression analysis were as follows: The F-value of ANOVA Table 6 measures the statistical significance of the model. From above table it can be concluded that the fitted model is significant as P-value of F statistics is 0.00 and it is less than level of significance level (α = 5%). The value of VIF is 1< VIF < 5 ; it specifies that the variables are moderately correlated to each other. The small values of VIF corresponding to the variables show that there is no problem of collinearity.
The R 2 = 0.634, it means that 63.4% of the variation in the dependent variable Y is explained by the independent variable in the regression model, the remaining 36.6% was due to non-model variables and random errors. The adjusted R 2 = 0.628 gives the idea of how well the model generalizes. The difference between the R 2 and adjusted R 2 is 0.325 -0.316 = 0.006; it means if the model was derived from the population rather than a sample it would account for approximately 0.6% less variance the outcome. Based on the regression analysis results, the regression equation was obtained as it is shown below: Y= 2.102 + 0.732X1 + 0.091X2 + 0.231X3 + 0.222X4 + 0.281X5

Discussion
As described above, the degree of farmers participation in the link was classified according to five factors: Commitment and sharing of benefits and risks, household awareness about the benefits of contracts, policy environment, management skills and enterprises capacity of link firms and price issues. Regarding the factor of commitment and

Nguyen Van Song, Vu Ngoc Huyen, Nguyen Thị Thuy, Dinh Van Tien, Thai Van Ha, Nguyen Xuan Diep, Vuong Thi Khanh Huyen
sharing of benefits and risks, all 10 variables had factor loading scores between .721 and .973, and the top-3 factor loading scores were the farmers are committed to providing information for firms (CK04;.973), following the production process (CK03; .970), selling products to enterprises as the original commitment (CK05; .931). For the factor of household awareness about the benefits of contracts, there were 5 variables which had factor loading scores between .820 and .914, and the highest factor loading score was the farmers can stabilize the price of their products by joining the link (NT07; .914). For the factor of policy environment, there were 4 variables which had factor loading scores between .760 and .848. For the factor of management skills and enterprises capacity of link firms, there were 2 variables which had factor loading scores between .946 and .972. For the factor of price issues, there were 2 variables which had factor loading scores between .908 and .852.
The absolute value of β (Beta) in Table 5 indicates the order of importance of the independent variables. The variable with the highest β value is the relatively most important independent variable. On examining the contributions made by the independent variables in the model to the model, it was found that the Commitment and sharing of benefits and risks between farmers and firms in the linking process made the biggest contribution with the value of (β = 0.732). It was followed by the Price issues that farmers are interested in in the process of implementing links with firms, Policy environment -Institutions related to farmer contracts between firms and farmers, Issues related to management skills and firms capacity of link firms, Household awareness about the benefits brought about by farmer contracts with firms. In fact, these figures were 0.281, 0.231, 0.222, 0.091, respectively. All the factors in the regression model had a positive effect on participation of farmers (coefficient a_i > 0).

Conclusions and Implications
Based on data from the 300 farmers survey of Thai Binh province, using the method of economic statistical analysis, comparative method, exploratory factor analysis method. The study has analyzed the status of farmer contracts through basic links: First, the link between farmers and farmers in Thai Binh was vertical linkages. Farmers join together to create areas of specialized production of crop products with a larger scale and larger area. Second, farmer contracts between farmers and firms, between farmers and cooperatives have been vertical links. That both the farmer and the business are responsible for the product, information can be shared more easily, and product competitiveness is enhanced. Third, the research analyzed the farmer contracts between farmers and firms. This form will help develop the crop sector to achieve higher economic efficiency. The parties to the contract must be complied with the binding conditions as agreement, thereby improving the responsibility among the parties.
In order to strengthen linkage between firms and farmers in crop production. There are some possible solutions. First, completing the system and policy environment, especially the process of agricultural land accumulation. Second, strengthen the management capacity and business capacity of linked firms. Third, promoting the development of a value-added chain model. For example, reaching new markets, finding outlets for product consumption, bridging helps farmers reorganize the production of agricultural goods in accordance with the process, standards, and requirements for the quality of agricultural products that the market needs. Fourth, developing production links, concentrating agricultural land, applying science and technology into production, establishing production farmer contract chain of crop production.
Some limitations of this study should be noted. First of all, this study used 5-point Likert-type scales for measuring the dependent and independent variables. However, the small number of options may contribute to the low reliability coefficients for the variables of this study. Furthermore, the limitations of this study are limited research materials, limited literature review, limited knowledge on correlation and regression analysis. Research only summarizes theoretical concepts but it has not given out new concepts. It is recommended that future research has to focus on rich literature review and primary research on how correlation and regression can be effectively use in data analysis processes of quantitative methods.