TECHNICAL EFFICIENCY OF STATE-OWNED SUGARCANE PRODUCTION IN EAST JAVA

This study aims to (1) identify the factors that influence the production of plant cane and ratoon cane, (2) determine the level of production efficiency of plant cane, ratoon cane, and poll, and (3) identify the factors that influence the inefficiency of plant cane and ratoon cane production. The data used was secondary data sourced from the production data for the 20172018 planting season with some inputs: land area, fertilizers, herbicides, labor, age of plants harvested and data of land types. From the analysis, it was revealed that (1) factors influencing the increase of plant cane production were land area, ZA fertilizer, harvest labor, and types of fields. Meanwhile, the influential factors impacting the increase of ratoon cane were land area, SP36 fertilizer, ametryn herbicide, harvest labor, type of fields, and HGU land type while estate labor, mechanization, and dummy varieties affect decreasing on it, (2) sugarcane farming was technically efficient (3) factors affecting the inefficiency for plant cane are formal education and rank levels of plant officer. However the coefficient of the formal education variable was negative and the rank level coefficient was positive. A higher level of education will increase production, but a higher rank level of plant officer will decrease it. Improving education levels can be provided by giving mentoring or the provision of courses. In ratoon cane, there was no effect of technical inefficiency. So an increase in ratoon cane production can be done by increasing the use of production inputs.


INTRODUCTION
One of the main problems of the sugar industry in Indonesia is low productivity. A release from the Ministry of Agriculture shows that the sugar yield According to Tinaprilla (2011) the land area is considered the most responsive to sugarcane production.

METHODS
Descriptive-analytic was applied as the basic method in this study. The secondary data sourced from the stateowned plantation company located in East Java Province were analyzed. They effect of technical inefficiencies. The production function is separated into functions for the plant cane and ratoon cane. Based on the above equation, the plant cane production function is estimated as: Ln Y = Ln a 0 + a 1 LnX 1 + a 2 LnX 2 + a 3 LnX 3 + a 4 LnX 4 + a 5 LnX 5 + a 6 LnX 6 + a 7 LnX 7 + a 8 Ln X 8 + a 9 LnX 9 + a 10 LnX 10 + a 11 LnX 11 + d 1 D 1 + Where Y is the production of plant cane (tons) and X are the factors that expected to influence the sugarcane production: X 1 = land area (hectare), X 2 = ZA fertilizer (tons), X 3 = SP36 fertilizer (tons), X 4 = KCL fertilizer (tons), X 5 = seeds (tons), X 6 = ametryn herbicide (liters), X 7 = 2.4D herbicide (liters), X 8 = estate labor (man-days), X 9 = harvest labor (man-days), X 10 = mechanization (work machine days), X 11 = age of cane harvested (months), dummy type of field (D 1 ) where wetland (D=1) and others (D=0), dummy sugarcane variety (D 2 ) where Bululawang (BL) variety (D=1) and others (D=0), and dummy HGU land type (D 3 ) where HGU land (D=1) and not HGU land (D=0). Then a 0 -a 11 = coefficient of the variable X 1 -X 11 and d 1 -d 3 =coefficient of dummy variable The test was carried out by using the Frontier 4.1 program. The frontier production function model is described as (Coelli et al., 2005): Where Qi is the i-th estate production, β 0 is intercept, βi is the coefficient of the variable Xi, Xi is the factor that is expected to influence the production of Qi, ʋi is the effect of uncontrollable factors and μi is the Agro Ekonomi Vol. 31/Issue. 1, June 2020 4 Ln Y = Ln a 0 + a 1 LnX 1 + a 2 LnX 2 + a 3 LnX 3 + a 4 LnX 4 + a 6 LnX 6 + a 7 LnX 7 + a 8 Ln X 8 + a 9 LnX 9 + a 10 LnX 10 + a 11 LnX 11 + d 1 D 1 + d 2 D 2 + d 3 D 3 There are two other variables besides the variables above, they are errors caused by uncontrolled factors Where μi is the effect of technical inefficiencies and are expected to be influenced by Z 1 = age (year), Z 2 = work experience (year), Z 3 = formal education (year), Z 4 = household size (person), and D 5 = dummy factor of the rank level of plant officer which are categorized into rank III and above (D=1) and others (D=0). δ 0 -δ 5 =coefficient of the variable Z 1 -Z 5 .
According to (Coelli et al., 2005), Frontier 4.1 program will produce the estimation of log-likelihood, gamma (γ) and Σ 2 values. According to Battese & Corra (1977) the log-likelihood value by MLE method should be compared with the log-likelihood value by OLS method.
If the log-likelihood value by MLE method is greater than OLS, it means that the production function is good and matches the reality on the estate. The gamma value (γ) indicates how much variation in the error term component caused by the inefficiency effect. The value of Σ 2 indicates the distribution of the error term inefficiency (µi). The small value of µi means that µi is normally distributed. T-ratio for sigma squared (Σ 2 ) and gamma (γ) is also compared with t-tables at 99%, 95%, and 90% level of confidence to test whether they are partially significant to the analysis of frontier production functions.
The stochastic frontier produces two simultaneous conditions that influence the efficiency and inefficiency. Efficiency was measured by the approach from the output side. The measurement of technical efficiency from the output side was the ratio of the observed output to the maximum output. The technical efficiency at each i-th estate in terms of output was measured using the formula (Coelli, 1996): Based on the estimating variables used in this study, the equation becomes: Where: TE = technical efficiency of the i-th estate E (Y * | U 1 , X 1 , X 2 , ..., D 3 = observed output (i = 1,2, ..., n) E (Y * | U 1 = 0, X 1 , X 2 , ..., D 3 = maximum output (i = 1,2, ..., n) frontier production function is using The value of technical efficiency is between 0 ≤ TE ≤ 1.

Stochastic frontier production of plant cane
In this study, the OLS method was not raised because OLS is an estimation of the average production function, while this research focused on the stochastic frontier production function. Analysis of the estimation of the stochastic Land area (X 1 ) significantly influenced the plant cane production with an elasticity value of 0.774. It means a 1 percent increase in land area with another input is constant (ceteris paribus) will increase production by 0.774 percent. Increasing the land area of plant cane can be done by adding HGU land, by leasing land from owner, by co-working with regional or private companies, or through land conversion from annual crops to sugarcane. The additional land area or extensification is the fastest step in increasing production, but increasing sugarcane land in Java needs to consider in the availability of land because land conversion has the potential to reduce the production of other commodities such as rice, corn, horticulture, and others. It is in line with study by Susilowati & Tinaprilla (2012) and Mazwan & Masyhuri (2019).
Harvest labor (X 9 ) had a significant effect with an elasticity value of 0.641, meaning that a 1 percent increasing of these input and ceteris paribus will increase production by 0.641 percent. Other variables that significantly affected were ZA fertilizer (X 2 ) with elasticity value of 0.005 and 2.4D herbicide (X 7 ) with elasticity value of 0.011. Adding 1 percent of X 2 and X 7 (separately) will improve the production by 0.005 and 0.011 percent respectively, so it is necessary to pay attention to the costs before adding these inputs. According to Pakpahan & Purwono (2018), ZA fertilizer contains high nitrogen nutrients so it can increase the weight of the yield of sugarcane and ultimately can increase sugarcane productivity. Therefore, many farmers apply ZA fertilizer more than other fertilizers. There were some variables with negative elasticity value and significant, they are seed (X 5 ), ametryn herbicide (X 6 ) and estate labor (X 8 ). It shows that the use of these inputs is excessive and must be reduced. The number of seeds used in this study was 7.86 tons/ha which according to the study, the amount should reduced.

Harvest labor became a significant
In conventional system prevailing in India, about 6 -8 tons seed cane/ha (nearly 10% of total produce) is used as planting material, which comprises of about 25-30 cm stalk pieces having 2-3 buds (Jain et al., 2010). The amount of ametryn used in this study was 3.73 liters/ha. Compared to the studies undergone by Puspitasari et al. (2013), the use of ametryn should be reduced. Puspitasari et al. (2013) stated that the use of a single herbicide ametryn (dose 3 liters/ha) once or twice is more effective in controlling weeds and able to increase the vegetative growth of sugarcane.
Some dummy variables show the difference in production between D=1 and D=0. In the type of field (D 1 ) variable, there was a difference between wetland and others. Wetland production was 7.5% higher than that of dryland. Dryland soils produced less than wetland because the extension of the stem was not optimal.
Based on research by (Mastur, 2016) where the production of BL was 9.8% smaller than that of non-BL. Study by Riajaya & Kadarwati (2016) stated that timely planting using varieties with the appropriate type of land typology will increase the productivity of sugarcane and sugar. variable, there is a difference between wetland and others. Wetland production was 16.6% higher than that of dryland.
Likewise with the varieties variable (D 2 ) where the production of BL was 7.2%, smaller than that of non-BL. The HGU land type shows that HGU land production was 26.3% greater than that of the others. The HGU land productivity was 66.87 tons/ha and the others were 62.05 tons/ha. HGU productivity was 7.7% greater than the others.
As figured in table 2, the loglikelihood function of the MLE in the ratoon cane was 213.691, while the value of the log-likelihood MLE was 187.802. It shows that the production function with the MLE method in the ratoon cane is good and following the conditions in the estate. Table 2 shows that based on the t-test, the sigma squared (Σ 2 ) value was significant at the 99% confidence level and the value of  S e c o n d , i t s h o w s t h a t t h e opportunity to increase actual production according to its potential production becomes smaller. In this study, the opportunity to increase production was 4.45%. Therefore, to increase sugarcane production, land extensification and increasing the production inputs are needed.

Fa c t o r s a f f e c t i n g p r o d u c t i o n inefficiency
As figured in   Source: Secondary data analysis *** = significant at 99%, ** = significant at 95%; * = significant at 90% ns = not significant of sugarcane farming was the level of education. The dummy variable rank level of plant officer was significant and had a positive value. It means that the higher the plant officer rank level, the higher inefficiency will be.
In ratoon cane, based on the results of the estimation of the MLE production function, there was no effect of inefficiency there so that the coefficient of inefficiency was meaningless.
Increased production on ratoon cane can be done through increased production factors, including increasing land area, increasing the use of SP36 fertilizer , ametryn herbicide and harvest labor. The expansion of wetland and the addition of HGU land can also be done to increase the production of ratoon cane. On the other hand, an increase in ratoon cane production can be done by improving ratoon maintainence techniques which are in accordance to Kadarwati et al. (2015) research proclaiming that it can increase productivity by 16.20 tons/ha. negative and the rank level coefficient was positive. A higher level of education will increase production, but a higher rank level of plant officer will decrease it. In ratoon cane, there was no effect of technical inefficiency. So, an increase in ratoon cane production can be done by increasing the use of production input or by ratoon maintain techniques.

CONCLUSION AND SUGGESTION
Inefficiencies in the production of plant cane can be reduced by increasing the level of education of plant officer.
Improved education levels can be provided by giving mentoring or the provision of courses. Another thing that can be done is by allowing plant officer to benchmark to other similar companies.