Các yếu tố quyết định hiệu quả kỹ thuật của các nông trại trồng lúa ở Kiên Giang, Việt Nam

Nội dung text: Các yếu tố quyết định hiệu quả kỹ thuật của các nông trại trồng lúa ở Kiên Giang, Việt Nam

1. DETERMINANTS OF TECHNICAL EFFICIENCY OF RICE FARMS IN KIEN GIANG PROVINCE, VIETNAM CÁC YẾU TỐ QUYẾT ĐỊNH HIỆU QUẢ KỸ THUẬT CỦA CÁC NÔNG TRẠI TRỒNG LÚA Ở KIÊN GIANG, VIỆT NAM TS. Nguyễn Hữu Đặng Trường Đại học Cần Thơ Abstract The aim of this study is to determine technical efficiency and determinants of technical efficiency of rice farms in Kien Giang, Vietnam, based on a cross-sectional data collected in 2015 from 302 rice farmers in Kien Giang province. The Cobb-Douglas stochastic frontier production function, incorporating inefficiency effects was employed to analyze the data, using the FRONTIER 4.1. The results revealed that the technical efficiency was ranged 64.22-99.85%, average of 84.97%. Significant factors that were found to positively affect rice output per farm were area, phosphate fertilizer, and rice variety while seed and nitrogen fertilizer were negatively related to the rice output per farm. Significant determinants of technical efficiency were positively related to technical efficiency were gender, education attainment, farm size, training, membership of farmers’ association. Key Words: Technical efficiency, rice farms, determinants of technical efficiency, stochastic frontier production function. Tómtắt Mục tiêu của nghiên cứu này là xác định hiệu quả kỹ thuật và các yếu tố quyết định hiệu quả kỹ thuật của các nông trại trồng lúa ở Kiên Giang, Việt Nam dựa trên số liệu năm thu thập từ 302 nông dân trồng lúa ở Kiên Giang năm 2015. Chức năng sản xuất biên giới ngẫu nhiên Cobb-Douglas, các hiệu ứng kết hợp không hiệu quả được sử dụng để phân tích dữ liệu, sử dụng FRONTIER 4.1. Kết quả cho thấy hiệu quả kỹ thuật đạt từ 64,22% đến 99,85%, trung bình đạt 84,97%. Các yếu tố quan trọng được nhận thấy là có ảnh hưởng tích cực đến sản lượng lúa trên mỗi trang trại là diện tích, phân lân và giống lúa, trong khi hạt giống và phân bón nitơ có ảnh hưởng tiêu cực đến sản lượng lúa trên mỗi trang trại. Các yếu tố quyết định hiệu quả kỹ thuật có ý nghĩa tích cực liên quan đến hiệu quả kỹ thuật là giống, trình độ học vấn, quy mô trang trại, đào tạo, thành viên hiệp hội nông dân. Từ khóa: Hiệu quả kỹ thuật, nông trại lúa gạo, các yếu tố quyết định về hiệu quả kỹ thuật, chức năng sản xuất biên giới ngẫu nhiên. 1. Introduction Technical efficiency, which reflects the ability of the firm to obtain maximum output from a given set of inputs (Farrell, 1957). It indicates that technical efficiency is the ratio of the actual output over the maximum output. There are a number of studies on the determinants of technical efficiency. Kalirajan and Flinn (1983) found that the practice of 950
2. transplanting rice seedlings, incidence of fertilization, years of farming, and number of extension contacts had significant influence on the variation of the estimated rice farm technical efficiencies in the Philippines. In addition, Najma and Atul (1996) found that technical efficiency was higher for the high-yielding variety (HYV) Boro crop as compared to the traditional Aman crop of rice farmers in Bangladesh. Adam et al. (2003) revealed that farm-level specialization was found to have a positive effect on efficiency while land fragmentation was detrimental to efficiency in Chinese grain sector. In addition, Tijani (2006) the levels of technical efficiency largely ranged from 29.4 percent to 98.2 percent in the rice farming in Osun State, Nigeria. Surender (2007) indicated that small-size farms are more efficient than medium- and large-size farms. Idiong (2007) showed that farmers’ educational level, membership in a cooperative/farmers’ organization and access to credit significantly and positively influenced the farmers’ efficiency. Ayinde et al. (2009) found that farm size, hired labor, fertilizer, seed, age, gender, household size and amount of credit were the significant determinants of technical efficiency of rice farmers in Nigeria. Jyoti et al. (2010) the farm size and female workers were positively related with technical efficiency. Rice production in Vietnam is mostly concentrated in the Mekong Delta, which is located in the Southern part of Vietnam, consisting of 13 provinces and covering 12 percent of the total country’s land area. The Mekong Delta covers more than four million ha of natural land area, three-fourths of which is agricultural land, and the rest is comprised of rivers and other uses. The Mekong Delta plays a key role in the country’s food security and export. It contributed about 90 percent of the country’s rice export in volume. Kien Giang province located in the south of the Mekong River Delta, its economy is agriculture based economy as the share of agricultural sector is 38 percent while those of industrial and service sectors are 26 percent and 36 percent, respectively. Rice cultivation is the most important subsector of Kien Giang province since it plays a crucial role in employment creation, income generation especially from rice exports, poverty reduction, and food security for the region and for the country as a whole. It has 733.850 ha of rice production area, account for around 10 percent of the regional rice production area, considered as the largest rice production province in the Mekong River Delta in terms of rice production area and output. However, it is difficult to expand rice production by increasing rice land area or crop intensification since almost all the agricultural land in Kien Giang have been utilized. There are also limitations related to crop intensification such as soil erosion, pest infestation, and other issues concerning sustainable development in agriculture. Therefore, promoting policies aimed at sustainable growth in rice yield will be the basis for sustainable development in the rice cultivation in Kien Giang province in the future. Moreover, rice production in Kien Giang recently has been confronted with problems such as the rapid increase in labor cost and other material input costs, which in turn, caused the decrease in the farmers’ levels of input use. A reduction in input use may have negative impacts on rice yield and the productive efficiency of rice farmers as well. These lead to question that how is the level of technical efficiency of rice farms and what factors affect the 951
3. farm’s technical efficiency. Thus, this study is aim to estimate technical efficiency and identify determinants of technical efficiency of the rice farms in Kien Giang province. 2. Methodology 2.1 Stochastic frontier production function The stochastic frontier production function was independently proposed by Aigner, Lovell, and Schmidt (1977) and Meeusen and van den Broeck (1977). The original specification involved a production function specified for cross-sectional data which had an error term with two components, one to account for random effects and another to account for technical inefficiency. Following Battese (1992), the stochastic frontier production function can be expressed in the following equation: Yi =−fx(iii ;β )exp( V U ) (1) th where i = 1, 2, , N and Yi represents the possible production level for the i sample unit; fx(;)i β is a suitable function (e.g., Cobb-Douglas or Translog) of the vector, th xi of inputs for the i unit and a vector; β is a vector of parameters to be estimated; and N represents the number of the units involved in a cross-sectional survey. This model is such that the possible production Yi is bounded above by the stochastic quantity, fx(i );exp( vi ) , hence, the term stochastic frontier. Besides, V is the symmetric error term accounting for random variations in output due to factors outside the control of the farmer such as weather, disease, bad luck, and measurement error whereas U represents the technical inefficiency relative to the stochastic frontier, which assumes only positive values. The distribution of the symmetric error component V is assumed to be independently and 2 identically distributed as N(0,δv ). However, the distribution of the one sided component u is assumed to be half 2 normally (u > 0) distributed as N(0,δu ) and, thus, measures shortfalls in production from its notional maximum level. If u = 0, then the farm lies on the frontier obtaining maximum output given variable and fixed inputs; but, if u > 0, then the farm is inefficient and makes * losses or the production lies below the frontier function and the distance of Yi and Y measures the extent of the farmers’ technical inefficiency (Coelli et al, 2005). Therefore, the larger the one sided error is, the more inefficient the farm becomes. Technical efficiency. The technical efficiency of an individual producing unit is defined in terms of the ratio of the observed output of the corresponding frontier output, given the available technology (Coelli et al, 2005). Thus the technical efficiency of unit i in the context of the stochastic frontier production function is the following expression. TEUi =−exp(i ) (2) * TEYi ==ii/ Y fx ( i ;ββ )exp( V i − U i ) / fx ( i ; )exp( V i ) =− exp( U i ) (3) * Yi is an observed output and Yi is the frontier output. X i , β s , and Vi are as defined earlier. In this case, Yi achieves its maximum value of fx(i ;β )exp( Vi ) if and only if TEi = 1. Otherwise, TEi < 1 provides a measure of the shortfall of observed output from maximum feasible output in an environment characterized by stochastic elements that varies across producers. 952
4. 2.2 The Empirical Model Using panel data gathered from the two surveys, this study employed the stochastic frontier analysis following the single-stage estimation procedure developed by Battese and Coelli (1995, 2005). The advantage of using stochastic frontier model is that it can help in understanding the causes of productivity changes over time. The stochastic frontier production function would be estimated by the Cobb-Douglas or the translog forms as follows: - The Cobb-Douglas stochastic frontier production form: 7 lnYi = β0 + ∑ β j ln X ji + β8 Di +Vi −U i (4) j=1 - Translog stochastic frontier production form: 7 1 7 7 7 (5) lnYi = β0 + ∑ β j ln X ji + β8Di + ∑∑β jk ln X ji ln X ki + ∑β j8 ln X ji * Di +Vi −U i j=1 2 jk==1 1 j =1 where, βj: regression coefficients of the explanatory variables in the estimated stochastic production function, where j = 1, 2 7; Yi: rice production output (kg/farm). Xji are factors contributing to rice output per farm, consisting of: X1i: land area (ha/farm); X2i: amount of seed used (kg/farm); X3i: amount of nitrogen used (kg/farm); X4i: amount of phosphate used (kg/farm); X5i: amount of potash used (kg/farm); X6i: amount pesticide used (g/farm); X7i: human labor used (man-days/farm); Di: other factors contributing to rice output per farm such as: D1: rice variety dummy (1 = improved variety; 0 = traditional variety). Vit: random variable assumed to be independently and identically distributed (iid) 2 N (0, σv ) and independent of Ui; Ui: non-negative random variable that is assumed to account for technical inefficiency in production. The subscripts j, i refer to the jth input used of ith farm. Simultaneously estimated with the frontier model was the rice farmer level technical inefficiency (TIE) model. The TIE model for the rice farm is expressed mathematically as follows: 8 TIEi = Ui = δ0 + ∑δ j Z ji + ξi (6) j=1 where, δj: regression coefficients of the explanatory variables in the estimated technical inefficiency model, where j= 1, 2 8; Zji: factors contributing to technical inefficiency such as, Z1i: gender of farmer dummy (male = 1; female = 0); Z2i: age of the farmer (years); Z3i: education attainment of farmer (years of schooling); Z4i: experience of the farmer in rice farming (years); Z5i: membership in farmers’ association (member = 1; not member = 0). Z6i: farm size dummy (area ≥ 0.6 hectare = 1; area < 0.6 hectare = 0); Z7i: credit access dummy (with credit = 1; no credit = 0); Z8i: attendance in training on rice production dummy (with training = 1; no training = 0); ξi: error terms, assumed to be 2 independently and identically distributed with mean = 0 and variance = σξ ; and the subscripts j, i refer to the jth characteristic of ith farm. 953
5. - Test for the appropriate functional form (i.e., Cobb-Douglas vs. Translog): the appropriate functional form was determined using the following selection criterion: (i) overall significance of the estimated equation based on the generalized Likelihood Ratio (LR) test, (ii) the number of significant variables based on the t-test, (iii) consistency of signs of the MLE coefficients with economic theory, and (iv) absence of multicollinearity. The likelihood ratio statistic (λ) used for the generalized Likelihood Ratio (LLR) test is given as follows: λ = -2[(L (H0) - L (H1)] (7) where, L (H0): value of the log-likelihood function of a restricted frontier model (or the Cobb-Douglas) as specified by a null hypothesis, H0; L (H1): value of the log- likelihood function of an unrestricted frontier model (or translog model) as specified by the 2 alternative hypothesis, H1. The LR test statistic (λ) has approximately a chi-square (χ ) distribution with the number of degrees of freedom equal to the difference between the parameters involved in H0 (Cobb-Douglas) and H1 (translog). The critical value was obtained from the normal χ2 table. The decision for this test was to reject the null 2 hypothesis (H0) if λ is greater than the critical χ value and vice versa. - Test for the appropriate frontier estimators (OLS vs. MLE): Using the same statistical testing procedure (generalized LR test) as testing for appropriate functional form mentioned above. However, L (H0) in the formula refers to the value of the log-likelihood function of the OLS frontier model as specified by the null hypothesis, H0, while L (H1) is the value of the log-likelihood function under the alternative hypothesis, H1 (i.e., MLE model). Similarly, the test statistic λ has approximately a chi-square distribution. The degree of freedom is equal to the number of parameters involved in the inefficiency model plus one (k +1), where k is the number of parameters or restrictions or explanatory variables except the intercept. The critical χ2 value was obtained from the Kodde and Palm (1986). The decision rule for this test is to reject the null hypothesis (H0) if λ is greater than the critical χ2 value and vice versa. Anyway, another test would be able to employ. The value of gamma parameter may lie between zero and one. A value of γ = 0 indicates that technical inefficiency is absent and the OLS is a more adequate estimation procedure to describe the parameters in the model. A value of γ close to one means that there exists technical inefficiency in the model, or if γ = 1, all the deviations from the frontier are entirely due to technical inefficiency and the MLE adequately characterizes the data. LR results for the functional and frontier estimation method tests were automatically derived by using the FRONTIER 4.1 computer program. 2.3. Data The data in this study is cross – sectional data collected by directly interviewing 302 rice farmers in three districts of Kien Giang province, namely Chau Thanh, Giong Rieng and Tan Hiep. About 100 rice farmers per each district were selected by random sampling. The data collection includes quantity of input use, paddy yield in the first crop of 2015 and other data related to the farm’s specific characteristics. 954
6. 3. Results and discussion 3.1. Farm‘s specific characteristics On average, the rice farmer-respondents have 8.74 years of schooling, 23.15 years of rice farming experience, 0.87 ha of rice farming area. This indicates that education attainment of the farmers are quite low that would be logically a somewhat barrier in adaption new production technology. The average distance from the main rice field to the farmer‘s house is 3.28 km, which implies that most of the farmers are living near their rice fields. There is 68 percent of rice farmer-respondents accessed the formal credit while another 32 percent were self-financing for their rice farming; 71 percent of rice farmer- respondents participated in rice production training while another 29 percent did not join any training related to rice farming over last three years; and 61 percent of rice farmer- respondents are member of local farmer’s association (Table 1). Table 1. Specific characteristics of 302 sample rice farmer-respondents in Kien Giang province, Vietnam Std. Farm ‘s characteristics Unit Average Dev. Gender dummy 1: male; 0: female 0.87 0.12 Educational attainment Year 8.74 3.32 Rice farming experience Year 23.15 11.24 Farm size dummy Ha 0.87 0.13 Credit access dummy 1: borrowed; 0: not 0.68 0.18 Training dummy 1: Participated; 0: not 0.71 1.91 Farmer’s association membership 1: member; 0: not 0.61 0.16 dummy Distance (largest field – house) Km 3.28 1.34 Source: Author’s survey in 2015. 3.2. Input use and yield of the sample rice farmer-respondents The average amount of seeds used by the sample rice farmer-respondents was 190.54 kg/ha while that in Mekong Delta, on average, was 142.7 kg/ha (Dang, 2017). This could be attributed to that most of the farmers (69.87%) applied the broadcast sowing, which need more amount of seed than that of the line sowing while number of farmers adapted line sowing account for 30.13 percent. The sample farmer-respondents applied several types of fertilizers. The most commonly used fertilizers were urea, ammo-phos (or Di-Ammonium Phosphate), complete fertilizer (contains nitrogen, phosphorous, and potassium) and muriate of potash, among others. In terms of active fertilizer ingredient form, on average, the sample farmer respondents applied 97.6 kg/ha of nitrogen fertilizer, 78.2 kg/ha of phosphate fertilizer and 33.8 kg of potash fertilizer. In addition, sample farmer-respondents applied several types of fertilizers in both liquid and powder pesticides. In terms of active pesticide ingredient and by converting the 955
7. liquid pesticides into powder pesticide, on average, sample farmer-respondents applied 1.835.03 g/ha (~ 1.835.03 ml/ha); the lowest level of pesticide application was 1,357.21 g/ha while highest one was 2,612.14 g/ha. The labor use was ranged 22.37-38.26 man- day/ha, an average of 31.5 man-day/ha. Recently, there was a marked reduction in the use of labor for harvesting operations due to the increased adoption of mechanical harvesters. Paddy yield of the sample rice farmers-respondents was, on average, 6,907.30 kg/ha; the lowest level of paddy yield was 6,224.27 g/ha while highest one was 9,692.13 kg/ha. Table 2. Mean levels of input use per hectare and paddy yield, 302 sample rice farmer-respondents in Kien Giang provinces, Vietnam ITEM Maximum Minimum Mean Std. Div. Seed (kg/ha) 135.51 243.37 190.54 42.31 Fertilizers by ingredients: Nitrogen (kg/ha) 147.13 61.54 97.6 33.24 Phosphate (kg/ha) 101.21 62.12 78.2 27.89 Potash (kg/ha) 53.28 0 33.8 16.12 Pesticide by active ingredients (g/ha) 2,612.14 1,357.21 1,835.30 343.21 Labor (man-days/ha) 38.26 22.37 31.5 8.19 Rice yield (kg/ha) 7,692.13 6,224.27 6,907.30 697,23 Source: Author‘s survey in 2015. 3.2. Results of the stochastic frontier production analysis 3.2.1. Testing results for appropriate functional form and estimator The result of LR test indicated that the translog functional form was more appropriate than the Cobb Douglas since the value of likelihood ratio statistic (λ) was 122.413, which was greater than that of critical value (60.097). Therefore, the Ho was rejected. However, except the interaction and square variables in the tranlog model, the Cobb Douglas resulted in more significant variables than the translog model based on T- test. Moreover, the signs of coefficients of variables in the Cobb Douglas were more consistent than those of the translog model. In addition, based on the result of testing for multicollinearity, the translog model contained serious multicollinearity problem. Hence, the Cobb Douglas functional form was chosen to analyze the data. Besides, gamma parameter γ was close to 1 (0.924), which indicated the existing of technical inefficiency in the model. Thus, the MLE was adequately characterizes the data. 3.2.2. Results of the stochastic frontier production analysis The results of the frontier production function revealed that the area, seed, nitrogen fertilizer, potash fertilizer and rice variety are found significantly to affect rice output per farm at one or five percent probability level, while the potash fertilizer, pesticides and labor were found to have no significant effects on rice output per farm at 10 percent probability level. In a Cobb-Douglas frontier production function, the regression coefficients are already the output elasticity. For instance, the regression coefficient of area of 0.91 956
8. indicates that a one percent increase in cultivated area would result in a 0.91 percent increase in rice output per farm, ceteris paribus. With regard to seed and nitrogen fertilizer usages, the study found that the farmers might be overuse of seed and nitrogen fertilizer as their coefficients are exhibited negative signs with rice output per farm. Phosphate fertilizer, on the other hand, positively influenced rice output per farm. The regression coefficient of phosphate of 0.036 indicates that a one percent increase in phosphate fertilizer would increase rice output per farm by 0.036 percent, other factors held constant. Similarly, the regression coefficient of variety is positive (0.070), implying that rice farms planted to improved varieties have a higher rice output per farm than those planted to conventional varieties, other factors held constant. Table 3. MLE of the Cobb-Douglas stochastic production and technical inefficiency functions, sample rice farmer-respondents in Kien Giang province, Vietnam Variable Para- Std. Variable name Coefficient T-ratio symbol meter Error Frontier Production Function Constant β0 6,862 0.260 26.416 ln A Area (kg) β 1 0.913 0.097 9.431 ln S Seed (kg) β 2 -0.120 0.063 -1.911 ln N Nitrogen (kg) β 3 -0.186 0.094 -1.976 ns ln P Phosphate (kg) β 4 -0.018 0.074 -0.249 ln K Potash (kg) β 5 0.036 0.014 2.607 ns ln LP Pesticide (g) β 6 -0.001 0.016 -0.047 ns ln L Labor (man-day) β 7 -0.020 0.367 -0.056 * DV Variety dummy β 8 0.070 0.040 1.746 Technical Inefficiency Function Constant δ0 -0.182 0.054 -3.357 Z1 Gender dummy δ1 -0.338 0.127 -2.668 ns Z2 Age of farmer (years) δ2 0.001 0.005 0.220 Z3 Education attainment (years) δ2 -0.012 0.005 -2.490 ns Z4 Farming experience (years) δ3 -0.095 0.082 -1.155 Z5 Membership dummy δ4 -0.013 0.006 -2.159 Z6 Farm size dummy δ5 -0.136 0.056 -2.430 ns Z7 Credit access dummy δ7 -0.287 0.194 -1.483 * Z8 Training dummy δ8 -0.082 0.044 -1.849 Variance Parameter σ2 0.014 0.003 5.428 γ 0.924 0.039 23.497 Log-likelihood function 152.545 LR test of the one-sided error 28.738 Mean technical efficiency (%) 84.970 Note: , , and * indicate statistically significant at 1%, 5%, and 10% probability level, respectively; and ns denotes insignificant. Source: Author estimates. 957
9. Determinants of technical efficiency: The average technical efficiency was 84.97 percent, which implies that with the recent input level, the rice sample farmer-respondents could be able to increase their rice output by 15.03 percent by improving technical efficiency factors. This is to examine the effects of socio-economic and farm-specific factors on technical efficiency of the sample rice farmer-respondents. A negative sign of the regression coefficient of an explanatory variable in the technical inefficiency function indicates that the variable improves technical efficiency. A positive sign means the opposite. The factors which were found positively affect technical efficiency of the rice farmer-respondents were gender and education attainment of the farm operator, farm size, participation in rice production training programs, and membership in a farmers’ association. The positive relationship between education attainment and technical efficiency might also be attributed to that the higher educated farmers adopted new production technology better than the lesser educated farmers. Likewise, the regression coefficient of participation in training dummy has a negative sign, which indicates that the sample rice farmer-respondents who participated in training programs on rice production which were conducted by the staff of the Department of Agriculture and Rural Development and some NGOs were more technically efficient than those who did not attend the afore-mentioned training programs. The explanation is that the sample rice farmer-respondents who attended training programs on rice production learned more about new technological developments and therefore were able to adopt better farm management practices in rice production. Thus, they tended to have more efficient use of resources than those who were not able to attend any training at all. This finding confirms the results of Seyoum et al. (1998), Wilson et al., (2001), and Seidu (2008) who reported that farmers who sought technical information and had adequate extension contact were associated with higher levels of technical efficiency. Similar findings were also found Kelvin, et al. (2008) in rice farming in Bangladesh. Similarly, the regression coefficient of membership in a farmers’ association dummy exhibited a negative sign and is statistically significant at one percent probability level. This suggests that the farmers who are members of farmers’ association would be more technically efficient than non-members, which might also be attributed to that the members of association have better chance to exchange production experience among the members and more frequency in participate in training program conducted by extension workers that help them have more efficient use of resources than those who were non-members. This finding is consistent with the finding of Idiong (2007) in his study of small-scale rice farms in the Cross River State of Nigeria. Likewise, the farm size was found positive effects to technical efficiency. This finding might be attributed to that with the larger farm, the farmers tends to spend more efforts on new production technology than those have smaller farm. On the other hand, age and farming experience of the farm operator, and credit access dummy had no significant effects on technical efficiency at ten percent probability level. 958
10. Distribution of technical efficiencies: The predicted technical efficiencies of the sample rice farmer-respondents differed substantially ranging from 64.22 percent to 99.85 percent. About 13.91 percent of the total sample farmer-respondents belonged to the most efficient category (95 - 100%). Only few (3.9%) of the sample farmers had technical efficiencies below 70 percent. Majority (41.72%) of the sample rice farmer-respondents belonged to the category (90 - >95%), indicating that most of the rice farmer-respondents were very technically efficient (Table 4). Table 4. Distribution of technical efficiency of 302 rice farmer-respondents, selected provinces in Kien Giang province, Vietnam, 2015 TECHNICAL EFFICIENCY No. of Farmers Percent (TE, %) < 70 6 1.99 70-<75 12 3.97 75-<80 18 5.96 80-<85 25 8.28 85-<90 73 24.17 90-<95 126 41.72 95-100 42 13.91 Total 302 100.00 Average 84.97 Minimum 64.22 Maximum 99.85 Std. Dev. 6.93 * Significant at 10 percent probability level. Source: Author estimates 4. Conclusions and Recommendations The study is to determine the technical efficiency and determinants of technical efficiency of selected rice farms in Kien Giang province, Vietnam based on a cross- sectional data collected in 2015 from 302 rice farmers in Kien Giang province. The Cobb- Douglas stochastic frontier production function, incorporating inefficiency effects was employed to analyze the data, using the FRONTIER 4.1. The results revealed that the average technical efficiency was 84.97%. With the recent input level, the rice farmers could be able to increase their rice output by 15.03 percent by improving technical efficiency factors. Significant factors that were found to positively affect rice output per farm were area, phosphate fertilizer, and rice variety while seed and nitrogen fertilizer were negatively related to the rice output per farm. Significant determinants of technical efficiency were positively related to technical efficiency were gender, education attainment, farm size, training, membership a farmers’ association. In order to further improve the rice yield and technical efficiency of rice farms, the study recommends to the rice farmers to reduce amount of seed usage; using improved rice variety; improving fertilizer management focusing on efficient use of fertilizer; and 959
11. increasing the farm size as possible. In addition, the study recommends to the local government to intensify extension services particularly the conduct of training programs; providing continuous support for massive propagation and dispersal of high-yielding varieties in cooperation with the private sector; strengthening farmers’ association; improving the level of education of farmers through short technical training; and developing agricultural land right market./. REFERENCES Aigner, D., C. Lovell and P. Schmidt, 1977, Formulation and Estimation of Stochastic Frontier Production Function Models. Journal of Econometrics 6: 21-37. Anuradha, N. and Y. Zala, 2010, Technical Efficiency of Rice Farms under Irrigated Conditions in Central Gujarat. Agricultural Economics Research Review 23: 375- 381. Adam, Z., E. Wallace and R. Scott, 2003. Technical Efficiency of Chinese Grain Production: A Stochastic Production Frontier Approach. Paper prepared for presentation at the American Agricultural Economics Organization Annual Meeting: 27-30. Ayinde, O., M. Adewumi and V. Ojehomon, 2009. Determinants of Technical Efficiency and Varietal-Gap of Rice Production in Nigeria: A Meta-Frontier Model Approach. Paper prepared for presentation at the International Organization of Agricultural Economists Conference, Beijing, China, 16-22. Battesse. G., 1992, Frontier Production Function and Technical Efficiency: A Survey of Empirical Applications in Agricultural Economics. Agricultural Economics Review 7: 185-208. Battese, G. and Coelli, T., 1992, Frontier Production Functions, Technical Efficiency and Panel Data: With Application to Paddy Farmers in India. Journal of Productivity Analysis 3: 153-169. Battese, G. and Coelli, T., 1995, A Model for Technical Efficiency Effects in a Stochastic Frontier Production Function for Panel Data. Empirical Economics 20: 325–332. Chau, L. 2004, Factors Affecting the Yield and Technical efficiency of Rice Production in Ha Tay Province, Vietnam. Agricultural Technology and Science Review 2: 70-75. Benoit, G., 2015, Technological Change: What do Technology and Change stand for. Project on the Intellectual History of Innovation. Working Paper No. 24. Coelli, T., Rao, D., O’donnell, C., and Battese, G., 2005, An Introduction to Efficiency and Productivity Analysis. Springer Science Business Media, Inc. (2005): 41-83. Dang, N.H., 2017, Technical effiency and technological change of rice farms in Mekong Delta, vietnam. Proceedings of the 11th Asia-Pacific Conference on Global Business, Economics, Finance and Business Management (AP17Thai Conference) ISBN: 978- 1-943579-72-3. 960
12. Farrell, M. J., 1957, The Measurement of Productive Efficiency. Journal of the Royal Statistical Society, Series A 120: 253-290. Kalirajan, K.P. and Flinn, J.C., 1983, The Measurement of Farm-Specific Technical Efficiency, Pakistan Journal of Applied Economics, 2, 167-180. Idiong, I., 2007, Estimation of Farm Level Technical Efficiency in Small Scale Swamp Rice Production in Cross River State of Nigeria: A Stochastic Frontier Approach. World Journal of Agricultural Sciences 3: 653-658. Jyoti, K., Arti, S., and Dileep, K., 2010, Technical Efficiency of Dryland and Irrigated Wheat Based on Stochastic Model. Agricultural Economics Research Review 23: 383-390. Mansfield, E., 1968, the Economics of Technological Change, New York: Norton. Najma, R. S. and Atul, A. D., 1996, Stochastic Frontiers and Technical Efficiency Distributions: An Analysis Based on Rice Farming Data for Bangladesh. Canadian Journal of Economics, Special issue. Rola, A. and Quintana Alejandrino, J.T., 1993, Technical Efficiency of Philippine Rice Farmers in Irrigated, Rainfed Lowland and Upland Environments: A Frontier Production Function Analysis. Philippine Journal of Crop Science 18, 56-69. Schmidt, P. and Lovell, C. A. K., 1979, Estimating Stochastic Production and Cont Frontiers When Technical and Allocative Inefficiency Are Correlated. Journal of Econometrics 13: 83-100. Surender, S. 2007, A Study on Technical Efficiency of Wheat Cultivation in Haryana, India. Agricultural Economics Research Review, Vol. 20: 127-136. Tijani, A., 2006, Analysis of the Technical Efficiency of Rice Farms in Ijesha Land of Osun State, Nigeria. Agrekon 45. Yao, R.T. and Shively, G. E., 2007, Technical Change and Productive Efficiency: Irrigated Rice in the Philippines. Asian Economic Journal 21(2):155–168. Xiaosong, X. and Scott, R. J., 1998, Efficiency and technical progress in tradtional and modern agriculture: evidence from rice production in China. Agricultural Economics 18: 157-165. 961