THE ECONOMIC PERFORMANCE OF INDONESIAN RICE-BASED

International Journal of Administrative Science & Organization, January 2014 Bisnis & Birokrasi, Jurnal Ilmu Administrasi dan Organisasi

Volume 21, Number 1

The Economic Performance of Indonesian Rice-based Agribusiness JOKO MARIYONO Lecturer and Researcher, Faculty of Economics - University of Pancasakti, Tegal, Indonesia [email protected] Abstract. Inefficiency is one of major causes of low performance in Indonesian rice production. This study measures the technical efficiency of rice production in five Indonesian regions and examines its determining factors. A stochastic frontier production function is used to reflect best practice production given certain levels of input use with equal amounts of technology. Unbalanced panel data on input-output rice production consisting of 358 farm operation in 2003, 2008 and 2013 are employed for estimating frontier production functions. The results indicate that variation in rice production across the five main regions is due primarily to variation in technical efficiency. Sources of variation within technical inefficiency include household characteristics, composition of labour and tractor use. Of the five regions investigated, rice production on Java is the most efficient. Technical efficiency of rice production increases over time in all five regions but remains low overall. This study concludes that there is considerable room for productivity improvements in Indonesian rice-based agribusiness through increases in technical efficiency. Keywords: farm level panel data, rice agribusiness, stochastic production frontier, technical efficiency Abstrak. Inefisiensi merupakan salah satu penyebab utama rendahnya kinerja agribisnis bebasis padi di Indonesia. Studi ini mengukur efisiensi teknis produksi padi di lima wilayah Indonesia dan meneliti faktor-faktor yang menentukan efisiensi. Fungsi produksi frontier stokastik digunakan untuk menduga produksi terbaik pada tingkat penggunaan input dan teknologi tertentu dengan jumlah yang sama. Data panel input-output produksi padi yang terdiri dari 358 agribisnis padi pada tahun 2003, 2008 dan 2013 digunakan untuk memperkirakan fungsi produksi frontier. Hasil penelitian menunjukkan bahwa variasi produksi padi di lima wilayah utama terutama disebabkan oleh perbedaan efisiensi teknis. Sumber variasi dalam inefisiensi teknis meliputi karakteristik rumah tangga petani, komposisi tenaga kerja dan traktor digunakan. Dari lima wilayah penelitian, produksi padi di Jawa adalah yang paling efisien. Efisiensi teknis produksi padi meningkat dari waktu ke waktu di semua lima wilayah namun tetap rendah secara keseluruhan. Penelitian ini menyimpulkan bahwa ada cukup peluang untuk memperbaiki produktivitas agribisnis padi Indonesia melalui peningkatan efisiensi teknis. Kata kunci: agribisnis padi, efisiensi teknis, produksi frontier stokastik, panel data tingkat petani

INTRODUCTION Indonesian agriculture is still one of sectors that occupies a central position in the national economy. The agricultural sector is the main foundation providing food for 245 million people in Indonesia, about 87% of raw materials of small and medium industries, as well as contributing about 15% of gross domestic product (GDP). In addition, the agricultural sector absorbs about 33% of the workforce and become a major source of income for about 70% of households in rural areas (Haryono, 2013). Recently, the agricultural sector is still important, despite gradual declines in the contribution to economy and employment, as depicted in Figure 1. Politically, rice is a strategic product. Both shortages of rice in domestic markets and highly volatile rice prices hold the potential to generate domestic political instability. The shortage of supply of rice into domestic markets has become a more pressing problem in the Indonesian economy, not only because of its status as the main staple food, but also because the price of rice plays a major role in forming expectations on inflation and economic stability. The recent developments in rice production have energised the ongoing debate in Indonesia regarding the government’s interventions in the rice market. Such

Figure 1. Share of Agricultural Sector to GDP and Employment Source: Indonesian Statitical Agency (BPS) 2013, analysed. interventions concerning the quantities and prices of rice imports are politically sensitive especially because rice is a staple food and accounts for a large share of both consumers’ budgets and total employment. In the future, the agricultural sector also remains one of the mainstays of the national food and energy security. The need for food and energy will continue to increase in line with the rate of population growth and increased prosperity. As a large country, food security is a key pillar of national stability, thus becoming one of the main targets of agricultural development which cannot be

36


compromised. Until now, rice is still a major component of national food security, so that the rice self-sufficiency remains a primary indicator of food security. The government continue to increase national rice production to 5% per year and is targeting a surplus of 10 million tons of rice in 2015 (Haryono, 2013). The target of production capacity improvement program is to enhance the national capacity to increase food production that can respond the dynamics of food demand of the population and encourage equitable distribution of food supply. It is expected to optimize the utilization of natural resources to achieve food security based on domestic resources (Suryana, 2008). Crucial problems of Indonesian rice production include low production efficiency (Haryono, 2013) and low productivity (Mulyani and Sarwani, 2013). This leads to a condition in which rice production is relatively low and uncompetitive compared to other rice producing countries. As a result, there is no incentive for farmers to continue operating rice farms, and imported rice will dominate the Indonesian market. It is more realistic for Indonesia to be more competitive in rice production by increasing the performance in terms of efficiency and productivity. This study aims to explore sources of variation in the productivity of Indonesian rice-based agribusiness. Understanding whether technical efficiency has been achieved will be important for agricultural policy makers in deciding whether or not upgrading existing technology or introducing new technology is necessary – or optimal – in order to increase rice production. If the level of technical efficiency is still low, it is likely that an increase in rice production can be achieved by improving technical efficiency using existing technology. Because, efficiency means to the extent of efforts required to accomplish the desired result (Rosdiana, 2011), so using the technology it’s important to increasing the rice production. This analysis utilises a stochastic frontier production technique to estimate technical efficiency and to determine sources of inefficiency. Many previous studies investigated technical efficiency in Indonesian agriculture, using various methods and data. Factors affecting efficiency include the type land as Makki et al. (2012) indicate that the farmers in tidal swamp land have better efficiency than those in upland; and farming practices as shown by very recent studies by Laksana and Damayanti (2013) showing that rice production with system rice intensification is more efficient than usual practices, and irrigation infrastructure is the dominant factor encouraging farmers to adopt the system; and by Hidayah et al. (2013) indicating that rice farming system with integrated plant and resource management approach in the research areas are efficient and profitable. Up to now, information on temporal patterns of technical efficiency in Indonesian rice-based agribusiness is still limited. This study complements and fills the gaps of previous studies in multiple respects. Firstly, this study uses panel data sets at farm level, which can reduce the analytical effects of specific characteristics embodied in each farm and farm operator. In most cases examining the technical efficiency of rice production, efforts have been made to isolate sources of technical efficiency. The intention of

Volume 21, Number 1

these efforts is to provide a basis for the development of polices which could improve technical efficiency. Many factors affecting efficiency in rice-based agribusiness have been determined, often based on locally specific factors. In most cases, the managerial characteristic of farm operators including age (representing farmers’ experience) and level of education (representing capacity to adopt technology) is of interest. This study seeks to improve the previous studies by examining additional sources of variation – including labour composition, mechanisation and geographical characteristics — which have not previously been examined. Lastly, this study uses more updated and wider coverage of data. For these reasons, it is expected that this study will provide updates on technical efficiency in Indonesian rice-based agribusiness RESEARCH METHODS This study uses a method of stochastic frontier analysis. The method has been widely used as a measure of performance of firm in many sectors, including agricultural sector (e.g. Tijani, 2006), manufacturing sector (e.g. Prabowo and Cabanda, 2011) and banking sector (e.g. Abidin and Cabanda, 2007) and multi-sectors (e.g. Ikram et al. 2012). In agricultural sector, the method has been recently used to analyse performance of agriculture in Africa (e.g. Fatoba et al. 2009; Enwerem and Ohajianya 2013; Erhabor and Ahmadu 2013), South Asia (e.g. GheeThean et al. 2012; Alam et al. 2011), South East Asia (e.g. Villano and Fleming, 2006), the Middle East (e.g. Moradi et al. 2013), and Europe (e.g. Kumbhakar et al. 2014). The recent global use of such method indicates its credibility in representing performance of firm in various sectors. Stochastic frontier production is defined as function model in which the disturbance term (Ԑ) is composed of two parts, a systematic component (v) and a one-sided component (u) (Villano et al., 2010). For panel data, a functional form of a stochastic production function can be specified as:

Yit = f (X it , β , t ) exp{ε it } .......................................(1) for ί = 1,2, … and t = 1, 2, … where Υ is output, Χ is a vector of inputs is time trend to capture technological change and β is a vector of parameters to be estimated. The error term (Ԑ) is, then defined as:

ε it = vit − u it ...................................................(2)

The systematic component vit, which captures random variation in output due to factors outside the control of the farmer, is assumed to be independently and identically distributed (iid) as N (0,σ2), and independent of uit, which specifies the technical inefficiency relative to the stochastic frontier. Most of the empirical literature assumes that ui has a non-negative (one-sided) half-normal distribution with N (0,σv2). Based on the assumption that ui and vi are independent, the parameters of the production frontier can be estimated using a maximum likelihood method and econometric software. Furthermore, given a multiplicative production frontier for which the production function

MARIYONO, THE ECONOMIC PERFORMANCE OF INDONESIAN

is specified, the farm-specific technical efficiency of the ith farm in the tth period is defined as the ratio of the conditional expectation of output, given the inefficiency effect, relative to its expectation if the inefficiency effect is zero, as: ϕ it =

E (Yit | u it , X it ) E (Yit | u it = 0, X it

)

= exp{− u it

}

..............(3)

Technical inefficiency can be considered as the unobserved effects embodied in producers. When data are collected over a sufficiently long time, technical efficiency may indeed vary over time concurrent with other changes in the (efficiency-related) characteristics of producers (Druska and Horrace, 2004; Feng and Horrace, 2007). This study uses a primal approach – i.e. the direct estimation of the production function– with functional form of a transcendental logarithmic (translog) production technology. Thiam et al. (2001) concludes that using more flexible functional forms results in a more accurate technical efficiency estimate. More flexible functional forms reduce the error terms (Ԑit = vit - uit), which results in higher estimates of technical efficiency. Considering that a higher rate of efficiency represents a better estimate, a primal approach is more accurate than the dual because ‘studies using the primal approach leads to significantly higher TE estimates than those obtained from dual frontiers’ (Thiam et al., 2001, pp. 241). The translog production function in this study, is specified as: .....(4) where (k,j)=1, 2, …,5 for land, capital, labour, material and agrochemicals respectively, βkj =βjk for k ≠ j , t is the time index, is logarithmic operation. A time index is included in the model to account for smooth technological progress (Kompas and Che, 2006). Factors affecting inefficiency of farm can be analysed using a multiple regression, formulated as:

......(5) where AG is age of farmers, ED is education, FM is number of family members, NP is the number of plots, ST is status of land, SZ is size of total area, SH is share of hired labour, MC is a dummy for mechanisation, and JV is a dummy for Java. The underlying principles for including those variables that determine level of efficiency are as follows. Age represents experience of farmers, older farmers being more experienced and thus expected to be less inefficient. Education represents human capital and skill and thus more educated farmers are expected to be less inefficient. Family member represents the size of households. Larger households are expected to be more capable of dealing with problems in farming and thus less inefficient. The number of plots represents land fragmentation. More fragmented land will be more difficult for farmers to manage and thus are expected to be more inefficient. Hired labour represents professionalism and thus more hired labour employed on farms is expected to be less

37

inefficient. Mechanisation represents the adoption of technology and thus mechanisation is expected to render farms less inefficient. Java is included in the model as a source of variation because of Java’s status as the “rice laboratory” of Indonesia. New technology and policies related to rice-based agribusiness have often been implemented on Java first, and thus farms located on Java are expected to be less inefficient. A time trend is not included in the model because of the inclusion of the age of the farmer variable, which increases over time at the same rate as time trend. Level of education and number of family members are also likely to increase overtime when the period under examination is sufficiently long. The time trend would likely be strongly correlated with those variables if it were included in the model. To capture the time-varying technical inefficiency, the temporal pattern of technical inefficiency needs to be modelled a quadratic function of time as:

u it = δ 0i + δ 1i t + δ 2i t 2 .....................................(6) where δ0i, δ1i, δ2i ( =1, 2, …,n) are the producer-specific parameters to be estimated. This model has advantages in that it is flexible and allows inefficiency to vary across time and between producers. In this model, the average rate of change in technical inefficiency across time can be identified. Importantly, there is no inconsistency in this two-stage approach because in the second stage predicted efficiency is dependent merely on time trend which is identically distributed among producers rather than on a number of producer characteristics. This model is consistent with a method in which technical efficiency is independently and identically distributed in the stochastic frontier (Karagiannis et al., 2002). This study uses a database which is established from longitudinal surveys conducted in five rice-producing regions of Indonesia. The regions include Lampung, West & East Java, West Nusa Tenggara, South Sulawesi and North Sulawesi. The regions were selected to provide diverse representation of Indonesian rice production. The unbalanced panel data consist of 358 farm operations in 2003, 2008 and 2013. Several villages were selected in each region based on agro-climate characteristics and farmers cultivating rice were drawn using purposive random sampling. Once farmers were selected, they became respondents of the survey and were interviewed sequentially. The quantitative data were supported with qualitative data collected from informal discussion with key informants. These information related to habits and social culture that might affect the performance of agribusiness such as exchange and voluntary labour during planting and harvesting seasons; share cropping and share tenancy, quality of land, land fragmentation and habits of local people working in informal sectors during out of planting and harvesting seasons. Farmers’ perception on soil fertility of land was also discussed. The number of variables observed in the data varied widely across survey years. This is because the survey sought to accommodate variations in farming which are spatially and temporally specific. For example, certain fertilisers were not used in some regions but always used

38


in others. In some regions, voluntary labour was common during early planting and harvesting seasons but it was uncommon in others. For the purpose of this study, therefore, the data were aggregated to avoid problems of missing data. Note that in Indonesian agricultural practice, including rice-based agribusiness; it was common for farmers not to use fertilizers, pesticides and tractors at all. In the absence of such inputs the rice production was still positive. However, with a translog production technology, the production with no such input will be zero and econometric estimation will be impossible as logarithm of zero is undefined. It is suggested that the problem can be handled by summing the quantities of individual fertilisers and replacing the zero level of input use with a small positive value. Villano and Fleming (2006) use a quadratic functional form rather than a translog model to overcome this problem. The results show that both methods provide very similar measures of output elasticity with respect to inputs and estimates of technical efficiency. The translog model, however, provides more precise estimates than the quadratic model as the log-likelihood for the translog model is much greater than that for the quadratic model (and the variance of the technical inefficiency effects in the stochastic frontiers for the translog model is also greater than that for the quadratic model). The description and unit measurement of aggregated input and output variables and technical inefficiency variables can be seen in Table 1 and Table 2 respectively. The hypothesis established here is to find factors affecting level of efficiency. The test for a technical efficiency is formulated as: H0 : δ 0

= δ 1 =  = δ 9 = 0 H : H is not true 1 0

(H.1)

inefficiency effect is constant. The test is formulated as: H0 : δ 1

=  = δ 9 = 0 H : H is not true 1 0

(H.2)

Testing for these hypotheses is conducted using a likelihood ratio test (LR-test) as described in Verbeek (2003). That is, L R = −2(L LR

H0

−L

H1

)

............................................(7)

where LL(H0) and LL(H1) are the values of the likelihood function under the null and alternative hypotheses, respectively. The value of LR asymptotically has a chisquare distribution if the null hypothesis is true. RESULT AND DISCUSSION For convenience, the outcomes of analyses are presented in three sub-divisions: results of testing for the hypotheses, technical efficiency and factors determining level of efficiency; and profiles of rice-based agribusiness grouped by level of technical efficiency. Table 3 shows tests for the technical efficiency model. The hypothesis that δ0 = δ1 = ... = δ9 = 0 is rejected. This indicates that inefficiency is dependent on producer characteristics. The test also rejects δ1 = ... = δ9 = 0, meaning

Volume 21, Number 1

Table 1. Data on Input and Output Used to Estimate Production Efficiency Variable

Description

Unit measurement

Rice production

un-husked production

kilogram

Area (A)

Total rice-sown area

hectare

Labour (L)

Total labour comprises family, voluntary and hired labour, used for six stages of farming

man-working day

Capital (K)

Capital consists of tractors and animals mainly used in land tillage

tractor-working day

Materials (M)

Total material used in rice production comprises seed, water irrigation, and green manure

monetary term*

Chemical fertilisers and pesticides. monetary Fertilisers consist of Urea, Triple term* Super Phosphate (TSP), Ammonium Sulphate (ZA) and Potassium Chloride (KCl). Pesticides comprise solid and liquid formulations Note: *) Monetary value is at 2003 constant price, deflated using food-crop price index. Chemicals (X)

Table 2. Data on Factors Determining Technical Efficiency Variable

Description

Unit measurement

Age

Age of farmer

year

Education

Education of farmer, years spent in formal education

year

Member

Number of household members, includ- persons ing the farmer

Plot

Number of blocks of land cultivated with rice

unit

Status of land

Fraction of privately owned land cultivated with rice. = 1 if totally owned land, = 0 if purely rented land

[0,1]

Area

Total area of rice cultivation

hectare

Share of hired labour

Share of hired labour, = 100 if fully hired, = 0 if fully unpaid labour

[0,100]

Mechanisa- Dummy for using tractor, = 1 if using tion tractor, = 0 otherwise

dummy

Java

dummy

Dummy location, = 1 if located on Java, = 0 otherwise

that the constant of the effect of inefficiency should be included in the model of the technical inefficiency effect. Table 4 shows parameter estimates of technical inefficiency effects. Individually, the factors that significantly improved level of efficiency include age of household head, education of household head, share in hired labour, size of farm, use of mechanisation and farm location. Older farmers enhance the performance of rice production. This is likely a result of older farmers having more experience and knowledge of rice growing activities than younger farmers. Older farmers may


also be more willing to embrace agricultural production practices that increase technical efficiency. They may also be simply more reliable in performing production tasks. Perhaps for these reasons, technical efficiency increase as age increases. Education level enhance the efficiency, meaning that a higher level of educational attainment results in better technical efficiency. Consistent with many studies on productivity and growth, educational attainment can be perceived as a proxy for human capital. Number of family members of the household is able to boost efficiency, by mean that more members in the household mean more labour is available for carrying out farming activities in a timely fashion and therefore the production process is more efficient. Number of plots can improve efficiency. The number of plots represents land fragmentation, which is expected to have a negative effect on efficiency. It is possible that more plots do not necessarily indicate that each plot is small, such that the more plots farmers have the larger those farms might be. Farm size is also has positive effect on efficiency. This means that a large farm is more technically efficient than a small one. Larger farms could be the result of either more plots or a larger single plot. Despite its insignificance, the status of land that Table 3. Testing for Production Fontier and Inefficiency Effects Hypothesis

Formulation

H.1

Source of inefficiency

δ 0 = δ1 =  = δ 9 = 0

H.2

Constant effect

δ1 =  = δ 9 = 0

z-count Decision

dasdfgval

27.29

reject

5.71

reject

Table 4. Parameter of Technical Inefficiency Model Coefficient

z-count

Constant

δ0

2.4751

4.44a

Age

δ1

-0.0252

-3.64a

Education

δ2

-0.0718

-2.57b

Member

δ3

-0.0464

-1.33n

No. Plots

δ4

-0.9066

-2.95b

Status of Land

δ5

-0.0761

-0.32n

Size

δ6

-0.6181

-1.88c

Share labour

δ7

-0.0055

-2.05b

Mechanisation

δ8

-1.3739

-3.48a

Java

δ9

-1.7632

-3.50a

1.0970

5.92a

0.8811

38.48a

σ γ

2

Log-likelihood -645.56 LR-test 137.47a Mean technical efficiency 0.6755

Note: a) significant at 1%; b) significant at 5%, c) significant at 10 %; n) not significant

39

represents the proportion of privately owned land to total farm land improve level efficiency. This is a common phenomenon whereby farmers rent out less fertile land and concentrate their farming on the more fertile land. Consequently, farms operated on less fertile land will be less efficient. This is consistent with the findings of Jamal and Dewi (2009) pointing out that sharecropping is inefficient for the tenants; and it is still a common type of land transaction in Java. To improve efficiency, the local government can help farmers by enabling them to lease land through fixed rental or other more favourable arrangements. Labour share enhance the level of efficiency. This share represents the proportion of hired labour to total labour employed on the farm. The employed labour that is not paid can include family, exchange and voluntary labour. Farms with a high proportion of hired labour are more efficient. Farmers are able to supervise the hired labour such that it works effectively and efficiently. On the other hand, it is unlikely and disrespectful for farmers to control exchange and voluntary labour to do the same extent. Mechanisation is able to increase level of efficiency. Farmers who use tractors are more efficient in producing rice. Tractor provides more effective work in large farm, which is consistent to the fact that large farms are more technically efficient. Rice-based agribusiness located in Java is more technically efficient than those located in other regions. One of the factors behind this is that Java is considered as a rice-bowl area, in which the government policy has conducted a lot of intensification programs and agricultural infrastructure has been well developed. Socially, tradition, culture in and the density of people in Java that is higher than that out site Java enables farmers to work closely in a so-called “gotong royong”. Average technical efficiency is depicted in Figure 2. Overall, the efficiency rate of Indonesian rice production was around 0.67. A very recent study by Mariyono (2014) using aggregate data across provinces of Indonesia shows a similar rate. This is lower than in Bangladesh, which show efficiency rate of 0.78 (Shantha et al., 2013). Of all the regions, average technical efficiency of rice-based agribusiness on Java is highest. This is not surprising because Java has the best quality land, in terms of soil fertility and climate, and as such is considered the most suitable location for non-tree cultivation of crops, including rice (Strout, 1983). Java also has better irrigation infrastructure than other islands. Touré et al. (2013) shows that rice production with better irrigation infrastructure perform better. In fact, the technical efficiency of Javanese rice production is 0.78, around 0.14 higher than the other regions in this study. Despite its performance relative to these other regions, however, Java’s agricultural facilities and highly fertile paddy fields mean that an efficiency statistic of 0.78 could still be considered low. In 2003, rice-based agribusiness outside Java was less efficient compared with that on Java. Based on the estimated technical efficiency, the rank order from most to least technical efficiency of rice-based agribusiness in 2003 is: Java, West Nusa Tenggara, South Sulawesi, Lampung and North Sulawesi. In 2008, the rank order was still the same as that in 2003. In 2013, however,

40

International Journal of Administrative Science & Organization, January 2014 Bisnis & Birokrasi, Jurnal Ilmu Administrasi dan Organisasi 0.85 0.8 0.75 Technical efficiency

the rank order dramatically changed. The coefficients for West Nusa Tenggara and South Sulawesi were not significant (and were even positive for South Sulawesi). This means that rice-based agribusiness in both regions was as technically efficient as that on Java. In other words, the technical efficiency of rice-based agribusiness in both regions had caught up with that on Java. The rank order in 2013 is South Sulawesi, Java, West Nusa Tenggara, and Lampung. Technical efficiency in rice-based agribusiness does not vary much across the other regions. In all regions, technical efficiency tends to increase over time. The differences in technical efficiency across regions and years (compared to Java) are given in Table 5. The fact that most of these coefficients are significantly negative indicates again that rice-based agribusiness on Java is the most technically efficient of all the regions. In Figure 3, it is clear that the technical efficiency of rice-based agribusiness in all regions except Java has continually increased. In 2013, the technical efficiency of rice-based agribusiness on Java fell due to a sharp decrease in capital use (capital had been substituted for by non-agriculture-experienced labour). As mechanised capital is an important factor in determining technical efficiency, and farms with mechanisation are more technically efficient, this substitution had some effect on technical efficiency. The catch-up of technical efficiency in West Nusa Tenggara and South Sulawesi therefore derives from two sources. The first is a fall in technical efficiency of rice-based agribusiness on Java in 2013 due to a sharp reduction in mechanised capital. The second is continual increase in technical efficiency in South Sulawesi and West Nusa Tenggara. Table 6 shows the dynamics of technical efficiency. Estimated with a linear form, technical efficiency significantly increases at a constant rate of 0.0227 every five years. However, when estimated using a purely quadratic form technical efficiency increases at an increasing rate. As described in Figure 2, overall technical efficiency is increasing at a decreasing rate, so it is reasonable to estimate the dynamics of technical efficiency in the form of a general quadratic function. The result indicates that the coefficient of the linear time trend is positive and the coefficient of the quadratic time trend is negative. This is an indication

0.7

0.65 0.6

0.55 0.5 2003

2008

2003 2008 2013 Overall

0.8 0.7

Lampung North Sulawesi

0.4 0.3 0.2 0.1 0

Nusa Tenggara

North Sulawesi

West Nusa Tenggara Overall

Table 5. Regression of Technical Efficiencies on Dummy Regions Regions

2003

2008

2013

Overall

Constant (=Java)

0.7716 (27.94)a

0.8268 (26.32)a

0.7325 (18.76)a

0.7839 (42.37)a

Lampung

-0.1822 (-5.41)a

-0.2229 (-5.96)a

-0.0923 (-2.03)b

-0.1757 (7.95)a

West Nusa Tenggara

-0.0803 (-2.56)b

-0.1234 (-3.48)a

-0.0223 (-0.50)n

-0.0840 (-4.01)a

North Sulawesi

-0.2031 (-4.46)a

-0.2255 (-4.48)a

N/A

-0.1990 (-6.11)a

South Sulawesi

-0.1372 (-4.16)a

-0.1257 (-3.11)a

0.0224 (0.46)n

-0.1043 (-4.58)a

R2

0.1122

0.1244

0.0632

0.0928

F-stat

10.62

10.62

3.78

20.77a

No. Obs.

341

304

172

817

a

a

b

Note: number in parentheses is t-ratio, a) significant at 1%; b) significant at 5%, c) significant at 10%; n) not significant; N/A: no observation Table 6. Regression of Technical Efficiency on Time Trend Variable

Linear Coeff.

Quadratic 1

z-count Coeff.

Constant

0.6348 39.32a

t

0.0227 2.75b

Quadratic 2

z-count Coeff.

0.6543 63.93a

z-count

0.6177

13.72a

0.0438

0.83n

0.0056 2.65b

-0.0054

-0.41n

R2

0.01

0.01

0.01

F-test

7.54b

7.00a

3.85b

Note: ) significant at 1%; ) significant at 5%, c) significant at 10%; n) not significant

0.5

West & East Java

Java South Sulawesi

Figure 3. The trend of technical efficiency

a

0.6

Lampung

2013

Year

t2 0.9

Volume 21, Number 1

South Sulawesi

Region

Figure 2. Comparison of Technical Efficiency Among Regions

b

that technical efficiency increases at a decreasing rate. It is important to note that both coefficients are individually insignificant, but jointly significant. This is because the time trend is small and the data are an unbalanced panel, such that both linear and quadratic time trends are highly correlated and cause a multicollinearity problem. The average technical efficiency of producers does not vary by region, but the individual technical efficiency among producers varies considerably. This indicates that within regions there is a large variation in technical

41


efficiency. This was always a likely outcome because of rice-based agribusiness’s sensitivity to ecological situations such as weather and pest infestations. This ecological situation clearly varies among regions and across time. For instance, when there is a pest outbreak in the middle or late stage of rice cultivation, rice production will be very low. In such a case, large amounts of inputs have been used and as a consequence the technical efficiency is low. Following Kompas and Che (2006), it is a relatively uncomplicated task to analyse technical efficiency rankings. Since there is a wide range of individual technical efficiencies, the rankings are grouped into ‘very low’ (less than 0.60), ‘low’ (0.60 to 0.75), ‘high’ (0.76 to 0.85) and ‘very high’ (greater than 0.85). The number of rice producers in each group is 219, 235, 280 and 83 respectively. The characteristics of rice-based agribusiness indicated by average values in each technical efficiency group are given in Table 7. There are a number of features that emerge from these profiles of rice-based agribusiness. First, high and very high technical efficiency groups of producers are more educated and more experienced operators. Second, ricebased agribusiness in both these groups also operates at a large scale either on single or multiple plots of land. The larger scale operations are the more technically efficient, where the levels of use of all inputs except irrigation and organic materials are higher. Third, the use of a high proportion of hired labour dominates the characteristics of the higher technically efficient groups. This is an indication that hired labour is more effective than voluntary labour. Finally, a high level of use of capital with a high proportion of tractors tends to be associated with high and very high technical efficiency. It is obvious that use of tractors is more effective than the use of animals for that same purpose, particularly in large-scale rice-based agribusiness. All of the coefficient correlations, except for unpaid labour and organic materials, are positive. The positive coefficient indicates that more technically efficient farms are correlated with a number of farm characteristics. Employing volunteer labour leads to low technical efficiency because, as mentioned above, workers are unlikely to be as effectively controlled by farm managers. Consequently, these workers are likely to be less effective. The use of organic materials also leads to low technical efficiency because the land actually receives organic material regularly from the biomass of plants at harvesting (eg. farmers usually leave dried rice stalks in the paddy field after harvest). Table 7. Summary Characteristics by Efficiency Groups Average value of farm characteristics

Unit

Efficiency group <0.60 (219)

0.600.75 (235)

0.760.85 (280)

>0.85 (83)

Correlation with tech. efficiency

Farmer Age

year

45.66

46.69

51.51

53.27

0.189a

Education

year

4.28

4.35

4.76

4.78

0.090c

Family member

#

4.90

4.92

4.85

4.86

0.021n

Output Total output

kg

784

1707

2859

5989

0.450a

Area

ha

0.50

0.58

0.70

1.04

0.211a

Number of plots

#

1.08

1.10

1.28

1.89

0.273a

Owned land

%

0.89

0.93

0.92

0.97

0.051n

Total capital

day

2.00

3.57

6.03

4.19

0.126a

Animal

day

1.63

2.41

2.52

2.01

0.074b

Tractor

day

0.41

1.15

3.51

2.19

0.142a

Total labour

day

50.45

48.26

62.91

98.32

0.181a

Family labour

day

39.75

34.89

40.86

41.43

0.045n

Unpaid labour

day

1.00

0.38

0.44

0.51

-0.049n

Hired labour

day

9.70

12.99

21.61

56.39

0.254a

Share hired labour

%

18.73

25.11

32.63

53.13

0.229a

Total material

Rp

20679

27453

38514

46638

0.224a

Seed

kg

28.56

38.19

49.52

59.86

0.178a

Irrigation

Rp

952

39594

10967

26119

0.252a

Organic materials

Rp

3796

135

134

70.48

-0.033n

Total agrochemicals

Rp

36656

55023

178838

125422

0.053n

Fertilisers

kg

83.15

124.63

205.59

322.30

0.197a

Pesticides

Rp

7610

12964

11820

21046

Land

Capital

Labour

Material

Agrochemicals

0.357a

Note: ) significant at 1%; ) significant at 5%; ) significant at 10%; n) not significant a

b

c

CONCLUSION The productivity of Indonesian rice-based agribusiness is on average still low, particularly in areas outside of Java. Because of the fact that most Indonesian people rely on rice for dietary energy requirements, it is important to raise productivity. There are two choices for achieving this: adopting new technology or raising the level of technical efficiency. Adopting new technology will be effective if the process of production with existing technology is technically efficient. However, if the production with the existing technology is still technically inefficient, improving technical efficiency is often a more appropriate policy objective. Thus, it is important to estimate the technical efficiency of rice-based agribusiness. After levels of technical efficiency are determined, factors producing the differences in technical efficiency can be isolated and efforts made to redress these factors. This study uses stochastic frontier production functions to indicate that technical efficiency still plays a key role in affecting Indonesian rice production. Average technical efficiency in the regions examined is around 0.68. The important factors that significantly increase technical efficiency are: farmer’s experience, educational attainment, size and number of plots, hired labour and mechanisation. More experienced and educated farmers

42


can increase technical efficiency because they will be more capable of implementing and correctly using existing technology. Mechanisation and hired labour lead to high technical efficiency because tractors are highly productive and hired labour works more effectively than voluntary labour. Regional characteristics that have positive effects on technical efficiency on Java include intensification programs and irrigation management. This higher level of technical efficiency appears where various extension programs have been implemented, and, as a result farmers in this region operate with more technical efficiency than those in other regions. Technical efficiency is increasing at a decreasing rate. The implication is that the average production function is approaching the production frontier. Since rice-based agribusiness is still technically inefficient, there is enough room for improvement in the productivity of rice farms by increasing technical efficiency for rice production with current agricultural technology. Rice-based agribusiness on Java, the rice-bowl of Indonesia, is considered the most desirable region for efficiency-related improvements because it already has good agricultural infrastructure and institutions, as well as good soil fertility in paddy fields. Despite Java having the highest technical efficiency of the regions studied, potential for improvements there remains, especially given the region’s worrying trend of decreasing technical efficiency. Another alternative, Indonesia has no other option for achieving national food security than to manage its available and suitable sub-optimal lands for food production. Efforts to increase productivity have become technically more difficult and economically less feasible for farmers. Nonetheless, it should be realized that suboptimal lands have many different characteristics and potentials. Therefore, technology development should be prioritized to create relevant technologies for each distinctive character of suboptimal land, financially affordable by local farmers, and in accordance with local communities’ preferences and social culture. There are two approaches could be simultaneously and interactively implemented. Firstly, optimizing physical, chemical, and (micro) biological soil conditions, coupled with effort in improving water resources management to increase effectiveness of irrigation or drainage network and efficient use of water and other resources. Secondly, selecting suitable agricultural commodities and developing crop cultivars adaptable to each specific characteristics of suboptimal lands. For maintaining sustainability of suboptimal land management, all technical and technological efforts should be evaluated not only based on their potential economic benefits, but also needed to consider their ecological impacts and socio-cultural values of the local community (Lakitan and Gofar, 2013). REFERENCES Abidin, Z. and Cabanda, E., 2007. Frontier Approaches to Production Efficiency of Commercial Banks in Indonesia, Manajemen Usahawan Indonesia, Vol. 36 No. 6, June. Alam, Mohammad Jahangir; Huylenbroeck, Guido Van;

Volume 21, Number 1

Buysse, Jeroen; Begum, Ismat Ara, and Rahman, Sanzidur 2011. Technical Efficiency Changes at the Farm-level: A Panel Data Analysis of Rice Farms in Bangladesh. African Journal of Business Management Vol. 5 No. 14, July. BPS, 2013. Ekonomi dan Perdagangan, Badan Pusat Statistik, Jakarta. (available at http://www.bps.go.id, data were downloaded on 31 January 2014) Druska, Viliam. and Horrace, William .C. 2004. Generalized Moments Estimation for Spatial Panel Data: Indonesian Rice Farming, American Journal of Agricultural Economics, Vol. 86 No. 1, February Enwerem V.A. and Ohajianya D.O., 2013. Farm Size and Technical Efficiency of Rice Farmers in Imo State, Nigeria. Greener Journal of Agricultural Sciences, Vol. 3 No. 2, February. Erhabor, P. O. and Ahmadu, J. 2013. Technical Efficiency of Small-Scale Rice Farmers in Nigeria. Research and Reviews: Journal of Agriculture and Allied Sciences, Vol. 2 No 3, July. Fatoba, I. O., Omotesho, O. A. and Adewumi, M. O., 2009. Economics of Wetland Rice Production Technology in the Guinea Savannah of Nigeria. Journal of Development and Agricultural Economics Vol. 1 No. 9, December. Feng,Qu and Horrace, William C. 2007. Fixed-Effect Estimation Of Technical Efficiency With TimeInvariant Dummies. Economics Letters Vol. 95 No. 2, May. Ghee-Thean, L., Ismail, M. M. and Harron, M. 2012. Measuring Technical Efficiency of Malaysian Paddy Farming, Journal of Applied Sciences, Vol. 12 No. 15, October. Greene, W.H., 1993. The econometric approach to efficiency analysis, in: H.O. Fried, C.A.K. Lovell, and S.S. Schmidt, (Eds.), The Measurement of Productive Efficiency: Techniques and Applications, Oxford University Press, Oxford, 68-119. Haryono, 2013. Strategi Kebijakan Kementrian Pertanian dalam Optimalisasi Lahan Suboptimal Mendukung Ketahanan Pangan Nasional. In: Herlinda S, Lakitan B, Sobir, Koesnandar, Suwandi, Puspitahati, Syafutri M.I, Meidalima D (Eds.) Prosiding Seminar Nasional Lahan Suboptimal “Intensifikasi Pengelolaan Lahan Suboptimal dalam Rangka Mendukung Kemandirian Pangan Nasional”, Palembang 20-21 September 2013. ISBN 979-587-501-9, pp.1-4 Hidayah, Ismatul; Hanani, Nuhfil; Anindita, Ratya and Setiawan Budi 2013. Production and Cost Efficiency Analysis Using Frontier Stochastic Approach, A Case on Paddy Farming System With Integrated Plant and Resource Management (IPRM) Approach In Buru District Maluku Province Indonesia, Journal of Economics and Sustainable Development, Vol.4, No.1, January. Ikram, M. Abdul Majid; Prasmuko, Andry; Anugerah, Donni Fajar; Kurniati, Ina Nurmalia 2012. Analysis of Sectoral Efficiency and the Response of Regional Policy. Bulletin of Monetary Economics and Banking, January. Jamal, Erizal; Dewi, Yovita Anggita, 2009. Technical Efficiency of Land Tenure Contracts in West Java Province, Indonesia, Asian Journal of Agriculture and Development, Vol. 6, No. 2, June.


Karagiannis, G., Midmore, P. and Tzouvelekas, V., 2002. Separating Technical Change from Time-Varying Technical Inefficiency in the Absence of Distributional Assumptions, Journal of Productivity Analysis, Vo.18, July. Kompas, Tom and Che, Tuong Nhu 2006. Technology Choice and Efficiency on Australian Dairy Farms, Australian Journal of Agricultural and Resource Economics, Vol. 50 No. 1, March. Kumbhakar, Subal C., Lien, Gudbrand, and Hardaker J. Brian 2014. Technical efficiency in competing panel data models: a study of Norwegian grain farming, Journal of Productivity Analysis, Vol. 41 No. 2, April. Lakitan, B. and Gofar, N. 2013 Kebijakan Inovasi Teknologi untuk Pengelolaan Lahan Suboptimal Berkelanjutan, In: Herlinda S, Lakitan B, Sobir, Koesnandar, Suwandi, Puspitahati, Syafutri M.I, Meidalima D (Eds.) Prosiding Seminar Nasional Lahan Suboptimal “Intensifikasi Pengelolaan Lahan Suboptimal dalam Rangka Mendukung Kemandirian Pangan Nasional”, Palembang 20-21 September 2013, ISBN 979-587-501-9, pp.5-14 Laksana, Satya and Damayanti, Arie 2013. Determinants of the Adoption of System of Rice Intensification in Tasikmalaya District, West Java Indonesia. Working Paper in Economics and Development Studies No. 201306 , Department of Economics Padjadjaran University Makki, Muhammad Fauzi; Ferrianta, Yudi; Suslinawati, Rifiana 2012. Impacts of Climate Change on Productivity and Efficiency Paddy Farms: Empirical Evidence on Tidal Swamp Land South Kalimantan Province – Indonesia. Journal of Economics and Sustainable Development Vol. 3, No.14, December. Mariyono, Joko 2014. Rice Production in Indonesia: Policy and Performance. Asia Pacific Journal of Public Administration, Vol. 36 No. 2, June. Moradi, Ebrahim; Pahlavani, Mosayeb; Akbari, Ahmad and Bashrabadi, Hossain Mehrabi 2013. Comparative Analysis of Stochastic Frontier Partially non-parametric and Stochastic Frontier Parametric Methods Case Study: Measuring Cost Efficiency in Wheat Production in Iran. International Journal of Agricultural Management & Development, Vol. 3 No. 2 June Mulyani, A. and Muhrizal Sarwani, M. 2013, Karakteristik dan Potensi Lahan Sub Optimal untuk Pengembangan Pertanian di Indonesia. In: Herlinda S, Lakitan B, Sobir, Koesnandar, Suwandi, Puspitahati, Syafutri

43

M.I, Meidalima D (Eds.) Prosiding Seminar Nasional Lahan Suboptimal “Intensifikasi Pengelolaan Lahan Suboptimal dalam Rangka Mendukung Kemandirian Pangan Nasional”, Palembang 20-21 September 2013, ISBN 979-587-501-9, pp.270-280 Prabowo, Handono E. T. and Cabanda, Emilyn 2011. Stochastic Frontier Analysis of Indonesian Firm Efficiency: A Note, International Journal of Banking and Finance, Vol. 8 No. 2, October. Rosdiana, Haula. 2011. Evaluating State Levies for Higher Education. Journal Administrasive Sciences & Organization: Bisnis & Birokrasi. Vol. 18 No. 2, May 2011. pp. 166 – 177. Shantha, A. Aruna, Ali Asan, B.G.H and Bandara, R.A.G, 2013. Technical Efficiency of Paddy Farming Under Major Irrigation Conditions in the Dry-Zone of Sri Lanka: A Parametric Approach. Australian Journal of Basic and Applied Sciences, Vol. 7, No. 6, April. Strout, Alan M. 1983. How Productive Are The Soils Of Java? Bulletin of Indonesian Economic Studies, Vol. 19 No. 1, January. Suryana, Ahmad 2008. Menelisik Ketahanan Pangan, Kebijakan Pangan, dan Swasembada Beras. Pengembangan Inovasi Pertanian, Vol. 1 No.1, Agustus Thiam, A., Bravo-Ureta, B.E. and Rivas, T.E. 2001. Technical Efficiency in Developing Country Agriculture: A Meta-Analysis, Agricultural Economics, Vol. 25 No. 2-3, September. Tijani, A. A. 2006. Analysis Of The Technical Efficiency Of Rice Farms In Ijesha Land Of Osun State, Nigeria, Agrekon, Vol 45, No 2, June. Touré, Ali A., Groenewald, Jan; Seck, Papa; Diagne, Abdoulaye Aliou 2013. Analysing Policy-Induced Effects On The Performance Of Irrigated Rice. African Journal of Agricultural and Resource Economics, Vol. 8 No. 1, July. Verbeek, M., 2003. A Guide to Modern Econometrics. John Wiley & Sons, LTD, Chichester. Villano, R., Boshrabadi, H. Mehrabi, and Fleming, E. 2010. When Is Metafrontier Analysis Appropriate? An Example of Varietal Differences in Pistachio Production in Iran. Journal of. Agricultural Sciences and Technology, Vol. 12, No. 4, October. Villano, Renato and Fleming, Euan 2006. Technical Inefficiency and Production Risk in Rice Farming: Evidence from Central Luzon Philippines, Asian Economic Journal, Vol. 20 No. 1, March.

THE ECONOMIC PERFORMANCE OF INDONESIAN RICE-BASED

Recommend Documents