INTRODUCTION

Agricultural cooperatives play an important role for small-holder farmers by offering a wide range of opportunities in accessing agricultural inputs, technology, market information, communications, credit and links to the marketing channels. Although cooperatives can create competitiveness for small-holder farmers, the formation of cooperatives have not developed very well in many developing countries (Balgah, 2019). As the Cooperative Societies Law is not in line with the international standards regarding with the autonomy, rules and regulations, the people’s trust in cooperatives’ functions is still low in Myanmar (MOALI, 2018).

After the new civilian government took power in 2011, they encouraged the development of bottom-up cooperative societies in the farmers’ community with support from Non-governmental Organizations (NGOs) in Myanmar. In the Second Five-Year Plan of Agricultural Policy (2016), the Ministry of Agriculture, Livestock, and Irrigation (MOALI) emphasized the policy goal of organizing smallscale farmers, livestock keepers and fisher folks into groups or cooperatives, with an emphasis on enhancing women’s participation (MOALI, 2018). Improving agricultural productivity and technical efficiency among the smallholder farmers is crucial for rural development and poverty reduction. Therefore, the establishment of agricultural cooperatives has been advocated as a potential policy tool in developing countries (Ma, Renwick, Yuan and Ratna, 2018).

In 2015, the United Nations Office on Drugs and Crimes (UNODC) initiated the coffee cooperative named Green Gold with over 900 former opium farmers from 60 villages in Southern Shan State of Myanmar in collaborations with the Ministry of Agriculture, Livestock and Irrigation and the Ministry of Cooperatives. The main objective of establishing Green Gold Cooperative project is to switch the opium poppy growing into sustainable coffee cultivation in the former opium fields in Hopong and Loilen Townships (UNODC, 2015). Following this, Indigo Mountain Cooperative was established in 2017 with 44 farmers from 77 villages in Hopong, and Shwe Taung Thu Cooperative was formed in Ywangan township in 2018 with the help of Winrock International and USAID (Evenson, 2019, 2021). Additionally, cooperatives formed by farmers themselves have emerged in these areas one by one during recent years, reflecting a growing local initiative to support and expand coffee cultivation.

Under the initiation of the coffee cooperative projects by the NGOs, these organizations have helped link cooperatives with foreign companies to facilitate product exports. With the assistance of the UNODC, the Green Gold Cooperative contracted five-year partnership (2017-2022) with the French company Malongo for the commercialization of its products in 2016 (UNODC, 2018). Shwe Taung Thu Cooperative has also exported its coffee to various countries such as the US, Canada, Dubai, the UK, France, Switzerland, Japan, Singapore, Australia, and Thailand over the past six years. Indigo Mountain has also exported its products to Netherlands, Australia, and Japan. Under the fair-trade conditions, coffee farmers receive fair prices through their cooperatives and have hope for sustainable livelihoods in the future. Market accessibility is important especially for smallholder farmers to enhance their income and profitability.

Moreover, farmers receive technical guidance, processing and quality assistance, and financial assistance from their cooperatives. Farmers are trained in several areas, such as: (i) determining the optimal time to begin harvesting, (ii) conducting rigorous quality assessments on the drying beds, (iii) upholding stringent storage and warehousing standards, (iv) delivering regular training in Good Agricultural Practices (GAP) farming, harvesting, and processing techniques, (v) encouraging climate resilience, (vi) creating organic fertilizer through farm waste composting, (vii) providing guidance on mulching practices to combat droughtrelated challenges, and (viii) book-keeping and recordkeeping to aid members to track, calculate profits and losses, evaluate efficiency, and perform break-even analyses.

Since cultivating high-quality coffee could generate high income like cultivating opium, the interests of farmers have risen to switch opium poppy growing into coffee cultivation in mountainous area of Southern Shan State (Guckelsberger, 2019). The growing demand for high-quality coffee in the international market presents an opportunity for expanding cultivation. Although Myanmar has favorable conditions for coffee growing, coffee is not as wide as other main agricultural export products and the sector is still in its infancy.

Although farmers’ cooperatives can strengthen the bargaining power of small-scale farmers in both input and output markets and create the opportunities to integrate into the value chain, it is crucial to assess the current status of technical efficiency and income among coffee cooperative farmers in Myanmar. Most of the coffee farmers still rely on traditional ways for planting and harvesting, and they face challenges such as political conflict, insecurity in their regions, and insufficient infrastructure. As a result, they struggle with lower yields and quality (Basu, Dobermann and Macchiavello, 2019;Guckelsberger, 2019). Access to improved technology, extension services and necessary infrastructure is essential for enhancing productivity and technical efficiency among small-scale farmers. This study aims to evaluate whether cooperative farmers achieve higher technical efficiency and greater revenue from coffee cultivation compared to non-cooperative farmers in Myanmar.

Many previous studies have analyzed whether the cooperatives achieve pricing efficiency, scale efficiency, and technical efficiency in various contexts. Yoo, Buccola, and Gopinath (2013) study the pricing and scale efficiency in 166 Korean cooperative Regional Rice Processing Complexes (RPCs) and find that the larger cooperatives exhibit scale efficiency, whereas small and medium-sized firms tend to be scale-inefficient. Some empirical studies prove that farmers who are the members of agricultural cooperatives achieve higher technical efficiency than non-cooperative members. Cooperative members achieve higher technical efficiency compared to their counterparts because they have easier access to extension services and productive inputs at the farm level in Ethiopia (Abate, Francesconi, and Getnet, 2014). Similarly, among the apple farmers in China, cooperative members exhibit greater average technical efficiency than non-members (Ma et al., 2018). On the other hand, Hailu, Weersink, and Minten (2015) find that the cooperative membership does not have a significant effect on the technical efficiency for teff growers in Ethiopia.

Some previous studies have emphasized whether participating in cooperatives achieve higher prices for the products and generate higher farm income based on various case studies. Wollni and Zeller (2007) describe that farmers participating in specialty coffee markets and marketing cooperatives receive higher prices than those in conventional sector in Costa Rica. Bernard, Taffesse, and Gabre- Madhin (2008) analyze the impact of marketing cooperatives on the commercialization of cereals and find that cooperative members obtain higher prices than non-cooperative members in Ethiopia. However, the share of production sold by the members does not increase significantly. Moreover, larger farmers increase their output level due to the high price while smaller farmers reduce their production. Wang, He, Zhang, and Jin (2021) show that the net income level of cooperative members is higher than those of non-cooperatives members in Shennongjia region, China. In contrast, Hun, Ito, Isoda, and Amekawa (2018) find that participation in agricultural cooperatives does not have a significant impact on the yields and revenue of the paddy farmers in Cambodia. Ofori, Sampson, and Vipham (2019) demonstrate that membership in vegetable cooperatives does not significantly increase agricultural incomes, the value of outputs and the amounts of inputs compared to non-cooperative members in Cambodia.

Many studies have investigated the production efficiency of the Myanmar’s agricultural sector by specializing on various crops such as rice, mung bean, mango, soybean, groundnut, and sesame by using data envelopment analysis (DEA) and stochastic frontier analysis (SFA) (Win, 2009;Aung, 2011;Latt, Hotta, and Nanseki, 2011;Mar, Yabe, and Ogata, 2013;Tun and Kang, 2015;Soe, Takahashi, and Yabe, 2020;Wai and Hong, 2021;Aung and Lee, 2021). However, studies on the technical efficiency of the coffee have not been conducted yet, and there is also a need to analyze the role that cooperatives play in the Myanmar’s agricultural sector by measuring their technical efficiency, productivity and farm’s income. This study aims to contribute to the existing literatures on the technical efficiency and the importance of cooperatives for enhancing productivity and rural development, focusing specifically on the current situation of the coffee cooperatives in Myanmar.

MATERIALS AND METHODS

Stochastic Frontier Analysis (SFA)

This study applied the stochastic frontier analysis (SFA) proposed by Aigner, Lovell and Schmidt (1977) to estimate technical efficiency of coffee farmers. The stochastic frontier production function expresses the maximum attainable outputs from a given inputs with a fixed production technology by adding two new terms that are technical inefficiency and random shocks (Aigner et al., 1977;Kumbhakar and Lovell, 2000;Coelli, Rao, O’Donnell, and Battese, 2005). The general form of stochastic production function can be expressed as follow:

where y_i is the maximum attainable output, x_i is a vector of inputs, β is a parameter to be estimated, v_i is random error or shocks beyond the control in the production process (e.g. weather, pandemic) following independently and identically distribution (i.i.d) N(0, σ υ 2 ), u_i is the nonnegative technical inefficiency following i.i.d truncated at zero N(0, σ u 2 ).

This study uses log-linear Cobb-Douglas production functional form of the stochastic frontier model (Kumbhakar and Lovell, 2000;Coelli et al., 2005) with the following equation:

where ln y_i is the logarithm of red cherry yield per acre of individual farmer, ln x_ni includes the logarithm of inputs used such as total area of land dedicated to coffee cultivation by each individual farmer, the number of coffee plants planted per acre, the number of labors used in picking coffee per acre, the expenditure on organic fertilizer per acre, the cost associated with using a weeding machine per acre, and total fixed costs per acre. Total fixed costs include land clearing cost, labor cost for marking stick, labor cost for staking, labor cost for digging, labor cost for transplanting and cost for growing shade trees (e.g., silver oak).

In addition to the Cobb-Douglas production function, this study also uses the Translog production function (Christensen, Jorgenson, and Lau, 1973) and analyzes with the following equation:

The Translog production function is highly flexible because it includes both first- and second-order terms, allowing the relationship between inputs and output to capture more complex behavior than simpler functional forms like Cobb-Douglas.

Propensity Score Matching

To investigate the impact of cooperative membership on the technical efficiency, farm income and costs, this study applies propensity score matching (PSM) method. If the observable and unobservable factors that may influence on productivity and efficiency as well as joining decision of farmers are omitted to account, selection bias may occur (Dong, Mu, and Abler, 2019). Then, the estimation of impact of cooperative membership on the technical efficiency, farm income and costs will be biased due to selection bias. Propensity score matching (PSM) has great features to correct the selection bias and differences derived from observable factors between the cooperative members and non-members (Bernard et al., 2008;Abate et al., 2014;Ma et al., 2018). Considering the unobservable factors, this study follows Bernard et al. (2008). They assume in their study that the formation of cooperatives is exogenous to community or household unobservable characteristics (i.e. risk preference, entrepreneurial spirit of each household) because most of the cooperatives (63%) in Ethiopia are driven by government institutions and not by the members. Therefore, they assume that controlling the observable characteristics is sufficient to make a comparison between the members and non-members.

Following Bernard et al. (2008), this study also assumes that the establishment of coffee cooperatives in the study area is partly exogenous to the unobservable characteristics of coffee farmers because these cooperatives are initiated by the NGOs or international donor community to switch the former opium poppy growing fields into coffee growing ones by helping small-holder farmers to engage in the specialty coffee markets. This study selected only the cooperatives initiated by the NGOs rather than those established by farmers themselves in the study area. Therefore, controlling the observable characteristics of the individual farmers is assumed to be enough to compare the technical efficiency of the two groups. PSM is a good technique to make a comparison if the selection bias due to the unobservable factors are possible to be negligible (Khandker, Koolwal, and Samad, 2010).

The first step of PSM is generating the propensity scores by using the probit or logit model that is the probability that a farmer would join the cooperative while other observable factors are assumed to be equal among farmers. Then, the second step is to construct a comparison (control) group by matching the non-cooperative members and cooperative members based on their propensity score. There are a number of matching approaches such as nearest- neighbor matching, caliper and radius matching, stratification and interval matching, and kernel matching and local linear matching (Khandker et al., 2010). Finally, we can estimate the average treatment effect on the treated (ATT) which is the mean differences in outcome values of the two groups. The ATT can be estimated with the following equation:

where, ATT is the average effect of the treatment (cooperative) on the treated (cooperative members), p(X_i) is the propensity scores, X_i is a vector of the observed variables that are assumed to have influence on joining decision to the cooperative, Y_1i and Y_0i are outcomes (i.e. technical efficiency) for the cooperative members (treatment) and non-cooperative members (control), D_i = 1 is the cooperative member and D_i = 0 is the non-member.

Data Collection

Shan State is the state with the most coffee cultivation than other states and regions in Myanmar. It is the largest state and can be divided as Southern Shan State, Northern Shan State, and Eastern Shan State with the greatest ethnic diversity of all regions and states. Its rugged, hilly terrain and historical ethnic conflicts have profoundly influenced the socio-economic landscape of its inhabitants for centuries (UNDP, 2015). In Southern Shan State, three districts: (1) Taunggyi, (2) Loilen, and (3) Langkho and two self-administered zones: (1) Danu and (2) Pa-O are included.

The necessary primary data were collected from coffee farmers living in Ywangan township, Hopong township, and Hsihseng township by using survey questionnaire. Since the NGOs initiated the coffee cooperatives projects in these regions, they are purposively selected as the study area. The survey was conducted from 28 December 2023 to 28 February 2024 in 15 villages of three townships. Among these villages, (7) villages from Ywangan township, (1) village from Hopong township, and (7) villages from Hsihseng township are included. From each village, 5 to 20 farmers were randomly selected. In total, data from 217 farmers from 15 villages of three townships were collected in this study. Table 1 describes the sample size, the number of cooperative farmers, and the number of non-cooperative farmers by villages and townships.

Measurement of the Variables

In the first step of the propensity score matching (PSM) method, the propensity scores are estimated by using the logit model. In the logit model, the dependent variable will be the cooperative membership or not. The respondents were asked whether they are cooperative membership or not with the dichotomous question and coded their responses as (yes = 1 or no = 0).

The independent variables such as age, gender, township, educational level, family size, total land cultivated area, credit availability, assets ownership, and non-farm income that might affect farmers to participate in the cooperative were selected based on the previous studies (Bernard et al., 2008;Abate et al., 2014;Ma et al., 2018). Moreover, we include the marital status, level of processing, and training by NGOs. Level of processing refers to the stage of production process conducted by farmers to sell their products both in domestic and international markets. Cooperative farmers are provided technical assistance and trainings by the NGOs to produce sun dried natural, parchment, and green bean to be able to export to the international markets. Therefore, most of the cooperative farmers’ products are not at the raw stage (i.e. red cherry) but at the intermediate stage. Because 94 per cent of respondents have already received training supported by Department of Agriculture, it is not considered as explanatory variable that may affect the probability of joining the cooperative. Instead, only training by NGOs is considered as explanatory variable because most of the cooperative members are obtaining technical supports from NGOs.

Age is measured as the continuous variable. Gender is coded as “1” if the respondent is male and “0” if the respondent is female. If the respondents are living in Ywangan township, it is coded as “1” and “0” for respondents who are living in Hopong and Hsihseng. Ywangan township is a part of the Danu Self-Administered Zone and most of the local people living in Ywangan are ethnic Danu people. Hopong and Hsihseng townships are parts of the Pa-O Self-Administered Zone where most of the residents are ethnic Pa-O people. Local Pa-O people are now growing coffee in the fertile Indigo Mountains located in these townships. Education is measured as the completed years of education by the respondents. The marital status of farmer is coded as “1” if they are married and as “0” for otherwise. Family size is measured by the number of family members who are currently staying together with the respondent. Total land cultivated area is measured as the number of acres cultivated by the respondents. Credit availability is coded as “1” if farmers are able to access the credit and as “0” for otherwise.

Assets ownership is measured by the number of assets owned by the farmers such as phone, TV, tractor, solar, motorcycle, bicycle, hand tractor, cabinet, water pump, car, and transport tractor. Non-farm income is coded as “1” for farmers who have other businesses besides cultivating coffee and as “0” for otherwise. The level of processing is coded as “1” if farmer sell and process only red cherry, as “2” if they sell and process red cherry and parchment and as “3” if they sell and process up to the green bean stage. Training by NGOs is coded as “1” if farmers are provided technical training services by NGOs or as “0” for otherwise.

RESULTS AND DISCUSSION

Descriptive statistics

Table 2 shows the descriptive statistics of the variables used in the logit model and stochastic frontier model. It shows that 42 per cent of the sample represent the cooperative members. The average age of the respondents is about 50. Among the 217 respondents, 77 per cent are male and 95 per cent are married. Among the respondents, 35 per cent of them have received credit from the government, 64 per cent have other sources of income, 16 per cent have accessed to the training by NGOs, and 46 per cent live in Ywangan township respectively.

Logistic estimation of the propensity score and matching

As the first step of the PSM approach, we have to estimate the propensity scores that is the probability of joining a cooperative by using a logit model. According to the results of the logit model presented in Table 3, marital status, access to credit, non-farm income, level of processing, trainings by NGOs and Township (Ywangan) are significantly related with the probability of joining a cooperative. If a farmer is married, it is less likely to join the cooperative than a single. A farmer who is access to credit is more likely to participate in the cooperative. The government has provided credit to coffee farmers from the economic development fund to encourage coffee farmers’ collective activity and to increase the coffee cultivation in 2022-23 and 2023-24 (Myanmar News Agency, 2022;Doe, 2022). A farmer who receives income from other sources rather than farming is less likely to participate in cooperative. On the other hand, farmers who engaged in higher level of processing and who received trainings by NGOs have more probability of participating in the cooperative because NGOs have supported technical assistance and trainings to the cooperative farmers. Moreover, a farmer who lives in Ywangan is less likely to participate in cooperative than those in Hopong and Hsihseng.

The distribution of propensity scores between the treatment and control groups before and after matching are presented in Fig. 2. Before matching, it is clear that the two distributions between the two groups are quite different. Therefore, matching technique is necessary to ensure the validity of our analysis. In this study, two matching techniques are used, namely (i) nearest neighbor matching and (ii) kernel-based matching. For nearest neighbor matching, each treated sample is matched with the closest comparison neighbor from the sub-set of the control group based on the propensity score distribution (Bernard, 2008). For kernel-based matching, each treated sample is matched with the whole sample of the control groups with weights inversely proportional to the distance between treated and control samples based on the propensity score distribution (Bernard, 2008).

Table 4 shows the balancing test that compares the mean values of the treated and control groups before and after matching. As shown in table, the mean differences between the treated and control groups for the unmatched samples are statistically significant for most variables. This result means that the mean differences between the two groups are significantly different from each other to make a comparison. However, after matching by using nearest neighbor matching and kernel-based matching techniques, the significant differences between the mean values of the two groups are reduced except for some variables.

Estimation of the technical efficiency

The maximum likelihood estimates of the stochastic frontier model with Cobb-Douglas function and Translog function are described in Table 5. Where output is red cherry yield per acre of individual farmer and six inputs including total cultivated coffee acres of individual farmer, number of coffee plants planted per acre, number of labors used in picking coffee per acre, organic fertilizer cost per acre, use of weeding machine cost per acre, and total fixed costs per acre are shown in logarithms form. Moreover, the regional dummy variable (Ywangan) is also included to capture the effect of regional differences on the coffee productivity. Under Cobb-Douglas function, the results in a likelihood-ratio test statistic of 16.97 (p-value = 0.000), assuming that the null hypothesis of no technical inefficiency (H₀: σ u 2 = 0) is rejected. Similarly, the likelihood-ratio test statistics of 2.14 (p-value = 0.072) under Translog function also indicated that the null hypothesis is rejected. The results revealed that the inefficiency component of the error term is significantly different from zero, indicating a stat- istically significant presence of inefficiency.

According to the results of Cobb-Douglas function in Table 5, the total cultivated area of coffee has a significant negative effect on yield. In contrast, the number of plants, the number of labors, the costs of fertilizer and fixed cost all have significant positive effects on yield. The cost of using weeding machine does not show a significant effect. Under the Translog function, fixed cost shows a significant negative effect on yield. Ywangan shows a significant positive effect on coffee yield per acre in both functions. Based on the results of stochastic frontier analysis, technical efficiency of individual farmer is estimated. Technical efficiency is the ratio of observed output to maximum attainable output.

where y_i is the observed output, f(x_i; β).exp{υ_i} is the maximum frontier output with the effect of random shocks on each producer. The value of technical efficiency falls between o and 1. If TE = 1 means that y_i achieves the maximum possible output whereas TE < 1 means that there is a shortfall of the actual output y_i below its maximum frontier output (Kumbhakar and Lovell, 2000).

The average impact of cooperative membership on technical efficiency, income and cost

Table 6 shows the results of the Average Treatment Effect on the Treated (ATT), the average impact of being membership on the technical efficiency, income and cost of the farmers, based on the Propensity Score Matching (PSM) method. According to the results, the technical efficiency obtained from the Cobb-Douglas production function for cooperative farmers is significantly higher than that of comparable non-cooperative farmers in Kernelbased matching. However, it does not show a significant result in nearest neighbor matching. The technical efficiency obtained from the Translog production function for cooperative farmers is significantly higher in both Kernelbased and nearest neighbor matching. It means that the maximum attainable outputs from a given combination of inputs for cooperative farmers are higher than those of comparable non-cooperative farmers.

It may be because of the technical assistances provided not only by the Department of Agriculture but also by NGOs. Farmers are provided knowledge and skills about the coffee business, developing coffee nurseries, establishing sustainable plantations and finding markets and buyers by working together with the NGOs (UNODC, 2018). Myanmar government has also encouraged the development of coffee cooperatives and the expansion of cultivation area by providing the credit and necessary infrastructure such as pulping machine in recent years. All these factors might lead to cooperative members being in a better situation. This result is consistent with the previous findings by Abate et al. (2014), Ma et al. (2018), and Dong et al. (2019). They also found that the technical efficiency of cooperative farmers is higher than that of non-cooperative farmers.

Regarding with the total income, the cooperative members received higher income, on average 1284.20 (‘000 kyat), than non-cooperative members in Kernel-based matching. However, in the nearest neighbor matching, the total income of cooperative members and non-cooperative members are not different from each other. It may be because the cooperative members obtain higher price by exporting their products to world market. Wollni and Zeller (2007) also find that farmers participating in specialty coffee markets and marketing cooperatives receive higher prices than those in conventional sector in Costa Rica. The total income of farmers in this study does not include non-farm income and considers only income derived from coffee production.

On the other hand, the total variable costs of cooperative members are significantly higher in both matching techniques, on average 360.83 (‘000 kyat) in the nearest neighbor matching and 370.18 (‘000 kyat) in Kernel matching than their counterparts. Total variable costs include the costs of labor for picking red cherry, the costs of organic fertilizer and the costs of weeding machine. These costs are yearly costs incurred by farmers. As shown in Table 6, all of these costs for cooperative members are significantly higher than those of non-members. During the data collection period, the market price for picking red cherry is within the range of 1000-1500 kyat per viss and labor costs are paid based on the amount of red cherries picked. Cooperative members may be hiring more workers due to the large scale or higher labor intensity of their tasks. Moreover, they could be using more organic fertilizers to meet cooperative standards and more frequent use of weeding machine. It is assumed that cooperative members may have more advantageous in access to inputs that can contribute to a reduction in their production costs. However, the results of our study show that the variable costs of members are significantly higher than those of non-members. It is clear that cooperatives are not in a competitive position in the input market. Therefore, members still need to obtain necessary inputs at a reasonable price by purchasing collectively.

Moreover, this study also finds that there is no significant difference in net income between the two groups. Net income measures as the difference between total income and total variable costs. Although the members may receive higher price by exporting, they could not enjoy higher net income because their total variable costs are also higher than non-members according to this study. The finding of this study is different from the previous findings by Wang et al. (2021). They evident that cooperative members obtain higher net income than non-cooperative members. However, our finding is consistent with the findings by Hun et al. (2018) who find that being a membership does not have a significant impact on the paddy yields and farm’s revenue in Cambodia, and Ofori et al. (2019) who also find that being a membership in vegetable cooperatives does not significantly increase agricultural incomes, the value of outputs and the amounts of inputs in Cambodia.

If the variable costs for cooperative members were lower than those for non-cooperative members, they could achieve higher net income. Therefore, it is crucial for cooperatives to enhance their collective purchasing strategies and improve their systems for providing inputs to members. By managing procurement more effectively, cooperatives could potentially lower their variable costs, leading to an increase in net income for their members. If cooperatives can negotiate better prices for inputs due to bulk purchasing or improve efficiency in input usage, they could reduce their costs. This would enhance the financial benefits of cooperative membership.

Robustness test

This study uses OLS regression to check the robustness of the results obtained from the PSM. Table 7 presents the results of the OLS regression, which estimates the effect of participation in cooperative on different outcomes such as technical efficiency, total income, total variable cost, and net income. The findings indicate that participation in cooperative has a positive effect on total income and net income and this effect is statistically significant at 1% level. However, there is no significant relationship between participation in cooperative and technical efficiency or total variable cost. In the results of the ATT, participation in cooperative does not affect net income, however, it does have a significant impact on technical efficiency, total income and total variable cost. Therefore, the results of PSM and OLS are not exactly the same in all outcome variables. Both methods yield consistent results for total income, but they differ for technical efficiency, total variable cost and net income.

CONCLUSIONS

Coffee cultivation in Myanmar is less widespread compared to other Southeast Asian countries although coffee has been grown for many years ago and there has a favorable weather and geographical conditions. After political reforms in 2011, coffee sector has attracted international donor attention as an alternative cash crop to opium particularly in the hilly areas of Shan State. With the assistance of the NGOs, smallholder coffee farmers have formed cooperatives and begun exporting their products to the international market. Since that time, coffee farmers could enjoy the fair price and many farmers have been interested in coffee cultivation.

This study analyzes the impact of membership in agricultural cooperatives on technical efficiency, income and cost of the coffee farmers in Myanmar by using the cross-sectional survey data. Our findings show that level of technical efficiency of members is significantly higher than that of non-members. Although members achieve higher total income due to the fair price by exporting their products, their net income are not significantly higher. Because their total variable costs are significantly higher than that of non-members. The type of coffee farmers’ cooperatives in Myanmar combine supply and marketing functions. Although the marketing functions are gaining momentum, the supply functions of cooperatives still needed to be improved. According to the interview results with the cooperative members, the cooperative in Hsihseng township has just started collectively purchasing organic fertilizer from a local factory using a government loan. However, some cooperatives in other townships have yet to establish a collective purchasing system.

To improve the collective input purchasing systems of the cooperatives in Myanmar, several strategies can be implemented. First, cooperatives could negotiate bulk purchasing agreements with suppliers to buy at the discounted price for larger quantities of goods or services. Second, cooperatives could consider a long-term contract with suppliers for providing raw materials or services to secure buying at lower prices or discounts. Lastly, cooperatives could organize shipments for members by sharing the costs to reduce transportation costs and delivery times. These strategies could help save on fertilizer costs as well as labor costs. Moreover, cooperatives could share necessary machinery (e.g. weeding machine) among members to reduce individual costs.

Fischer and Qaim (2014) highlighted that collective action through farmer groups is a crucial strategy for smallholders to remain competitive in dynamic agricultural markets. In addition, they suggested that cooperatives that are too small may not be able to achieve significant benefits as well as too large could lead to coordination difficulties and free-riding behavior. Therefore, finding an optimal size for cooperatives is crucial for balancing the benefits of collective action. It would also be beneficial for cooperative members to receive training on cost-effective farming practices such as making natural compost from animal manure and agricultural residues to further reduce fertilizer costs.

According to the survey results, 94 per cent out of 217 farmers have received the trainings from Department of Agriculture and 16 per cent have obtained the technical assistance from NGOs. Due to these technical trainings, the level of technical efficiency of member farmers might be significantly higher. Abate et al. (2014) suggested that easier access to extension services and productive inputs could lead to the higher technical efficiency of cooperative farmers in Ethiopia. Based on the other countries’ experiences, Department of Agriculture and INGOs should continue to provide innovative farming technologies, market information and necessary inputs to enhance the performance of Myanmar’s coffee farmers.

Ma et al. (2018) suggested that policymakers should implement policy incentives to encourage smallholder farmers in China to join cooperatives to enhance apple productivity, since participating in cooperatives leads to an increase in technical efficiency. Wang et al. (2021) also highlighted the role of cooperatives in China’s poverty reduction efforts by increasing farmers’ income. Moreover, Dong et al. (2019) found that participation in farmer producer cooperatives (FPCs) significantly boosts income and technical efficiency for small-scale farmers. However, a low participation rate could not fully benefit to many members. Correspondingly, Myanmar government should also provide incentives and support for smallholder farmers to promote cooperative membership, as this could help improve technical efficiency and productivity, especially in areas where farmers face barriers to forming cooperatives. In recent years, Myanmar government has introduced agricultural loans with favorable interest rates for coffee farmers. Coffee cultivation is a long-term investment and requiring 3-4 years to begin fruit production. The initial investment cost for coffee is high and the profit will be able to make after 7-8 years. To encourage coffee cultivation and to remove the financial constraints facing by smallholder farmers, access to credit is essential. In addition to offering credit, providing improved technology, extension services and necessary infrastructure could serve as an incentive strategy to encourage cooperative membership and enhance productivity and technical efficiency among smallscale farmers.

To strengthen the competitiveness of Myanmar coffee farmers, it is essential to focus on building networks that link farmers with both domestic and international markets. Cooperative networks can provide farmers with better access to market information and technical assistance. Wollni and Zeller (2007) suggested that providing market information and liquidity to small-scale farmers are important because these factors prevent them to join the cooperatives. Their findings also showed that participation in specialty coffee marketing channels and cooperatives increases the prices received by coffee farmers in Costa Rica. In the present, the management boards of cooperatives in Myanmar are building a network connection with coffee buyers, exporters, and other stakeholders in the coffee industry to ensure the smooth operation of the business. They also actively participate in the global coffee community through international trade fairs and virtual expos. Establishing direct connections between cooperatives and international coffee buyers will enhance Myanmar’s presence in the global coffee market.

In summary, strengthening agricultural cooperatives is crucial, as they play a pivotal role in both supply and marketing functions for coffee farmers. Improving cooperative input purchasing systems will help reduce costs and enhance competitiveness. Additionally, increasing the adoption of new technologies through targeted training and practical support is vital for boosting productivity and technical efficiency. Investment in infrastructure, such as roads, processing facilities, will facilitate better market access and efficiency. Addressing political instability and improving regional security are also critical for maintaining stable production and ensuring farmer safety. Expanding access to affordable credit will support farmers in managing the high initial investment costs associated with coffee cultivation.

Moreover, coffee cultivation contributes to environmental conservation because coffee is normally cultivated under shaded trees and growing shaded trees is a prerequisite condition. Shifting from opium poppy cultivation to coffee farming supports the sustainability of both the community and the region. Therefore, the roles of government and NGOs are crucial in building a more resilient and prosperous coffee sector that benefits smallholder farmers while also contributing positively to environmental and social sustainability.

There are some limitations in this study. First, the data used in this study are cross-sectional with 217 observations. Second, the data were collected only from coffee farmers in Southern Shan State. Coffee is primarily grown in Southern and Northern Shan State and Kayin State in Myanmar. Because of the limited time and the current political instability especially in Nothern Shan State and Kayin State, this study focuses solely on townships in Southern Shan State where coffee is widely cultivated. Third, if this study had applied the Instrumental Variables (IV) approach or Difference-in-Differences approach, clearer causal effect could have been estimated. However, due to data availability challenges, we tried to mitigate the selection bias by using the Propensity Score Matching method. In the future, this study could be expanded with more observations and panel data to get more reliable results.

적 요