MEASUREMENT OF TECHNOLOGICAL CHANGE IN PERENNIAL CROPS: Case in Rubber Farming in Batang Hari Regency - Jambi

Aulia Tasman

Keywords: technological change, total factor productivity, perennial crops, rubber farming.

Abstract:

This study aimed to measure the technical efficiency and technological change in rubber farming. Data collection was done in Batang Hari regency, Jambi from farmers belonging to the traditional group and to government estates. This cross section data were used to simulate the behavior of the time series data, based on potential yield and age of tree.

The farmers are experiencing positive technological change in group age 1with values ranging from 13.9% to 15.1%, and 2.3% to 4.6% in group age 3. Any attempt to increase productivity should be geared more tothe introduction of new technology, i.e., HYVs more widely for all groups of farmers, and the training of farmers on the package of knowledge.

INTRODUCTION

In terms of production of Indonesian rubber, the smallholders registered an overall growth of 3.0% per year, from 714,500 tons in 1980 to about 1.0 million tons in 1993. On the overall, production of government estates slightly increased by an average of 1.1% per year during the same period, from 186,000 tons in 1980 to 211,600 tons in 1993. Production from the private estates had increased at higher rate during the last five years, averaging 7.0% per year, although on the overall, production increased by an average of 3.7% per year, increasing from 120,000 tons in 1980 to 183,000 tons in 1993.

By type of producers, the government estates still had the highest average yield per hectare of 1,132 kilograms in 1993, reflecting a decline from 1989 to 1993 by around 2.0% per year. Yield of the private estates had stabilized at around 1,100 kilograms per hectare during the last five years. Yield of smallholder was the lowest at 639 kilograms, with slight improvement from its former level of 542 kilograms in 1980.

Comparing the yields among three type of rubber farming practices, the smallholder estates have the lowest yield. Several questions can therefore be raised such as:

(1) Why is there a big difference between the yields of the smallholder compared to the yields of government and private estates?

(2) What are the dominant factor sources of lower yield of the smallholders?

(3) Could lower yields be caused by slow technological change?

To answer these problem, the study will be focused in one particular regencies in Jambi province. The study will conduct a comparative analysis to the type of rubber farming between the smallholder and the government estates in Batang Hari regency, while the private estates is not included because the private estates for rubber plantation does not yet established in this regency.

This study aims to provide information to policy makers to be used in formulating an appropriate policy.

The general objective of this research is to assess the production models in rubber farming and based on the findings, develop the appropriate policy in improving farmer welfare, and to meet other government objectives such as to increase employment and farmer income. Specific objectives of this study are:

1. to analyze the nature of technological change.

2. to offer some appropriate policy implications relevance to the research findings.

REVIEW OF LITERATURE AND METHODOLOGY

1. Construction of Life Time Matrix of Perennial Crops

Precise estimation of economic parameters necessitates availability of data on prices and quantities of inputs and outputs for the entire life span of perennial crops, which in some cases exceeds many decades. Moreover, Chand (1994) said that many inputs applied to perennial crops in one period affect the output in following periods also. Any data collected by conducting survey can cover only some years of the total lifetime of such crops and it is nearly impossible to get data on lifetime input use, output and corresponding prices from the farmers growing these crops, especially the farmers who do not keep farm records to supply such fast information. There are at least two main problems related to perennial crops. First, the consequences of inputs applied in the initial period cannot be seen directly for the next periods. Second, it is difficult to determine precisely the future inputs and outputs.

Chand (1994) proposed other procedures/methodologies to study the economics of perennial crops by constructing a lifetime matrix for a given data. The data on quantity and value of inputs and output are obtained for each age year/group by dividing the total life of perennial crop in homogeneous periods. Based on these quantities, a single value of each parameter under study such as yield, return, profitability, etc. is obtained. The advantages of this approach are:

(1) It gives complete information of output and inputs, and the distribution for total life span of perennial crop for each producing unit rather than getting a single value for each variable.

(2) The implicit restrictive assumption of previous approach is that in the past and in future; the individual has the same value of variable and the restrictive assumptions can relaxed.

(3) The statistical tools which require data on individual observation can be applied. Similarly, the tools of production economics such as resource use efficiency, production function analysis, factor share can also be applied to analyze the production behavior.

(4) Estimates of expected cost of production, return and profit can be obtained for individual units based on total life span. Intuitively the estimates are expected to be more reliable sense compared to those obtained by using the previous approaches.

The tools of economics as applicable in the case of annual crops cannot be applied directly as such in the case of perennial crops until these have been postulated into economic problems. Precise estimation of economic parameters necessitates availability of data on prices and quantities of inputs and output for the entire life span of the perennial crops of more than three years. Chand (1994) proposed an alternative methodology to estimate production parameters by including some additional variables in the production function. Let:

· N the total life of perennial crops

· n the units for each age year or age group

· Sj the average of sub-sample at the jth year

· Yji the value of variable at the ith sample unit which belongs to jth year

· Ki the index for the value of variable for individual i which is the unit of sub-sample in j the age year,

then the computation of the ratio (index) between the value of variable for the individual and sub-sample average is

(1)

where Ki indicates that the value of variable Y for ith unit is Ki times the sample average. Thus it can be deduced that ith unit used Ki times more input or produced Ki times more output compared to the average of the sample. Based on this, the missing values of variable for ith unit in the past and the future life of the perennial crop can be obtained by multiplying the average value of each sub-sample by Ki.

2. Technological Change

Technological change refers to the changes in a production process that comes about from the application of scientific knowledge. These changes in the production process can be realized in various ways at the firms levels; through improved methods of utilizing existing resources such that a higher output rate per unit of input is obtained, often referred to as disembodied technological changes; through changes in input quality, referred to as embodied technological change; or through the introduction of new processes and new inputs (Antle and Capalbo, 1988).

Technology is a stock concept indicating the body of knowledge that can be applied in the production process. Brown (1968) defined technological change in terms of changes in the four characteristics of the abstract technology which a given production function embodies:

(1) Changes in the efficiency of a technology where the output is augmented for given set of inputs, but where the relationship between the inputs and the degree of returns to scale are altered;

(2) Changes in the returns to scale as the result of modifications in the technology; and

(3) Changes in the ratio of the elasticity of production with respect to different factors which alter the marginal rate of substitution between different factors.

Conventionally, the change or improvement of a technology can be observed as a change in factor productivity index in terms of a certain input (Y/Xi), or a change in total productivity index in terms of output per unit of combined input (Y/X), regardless of the nature of technological change, neutral or non-neutral. Basically, both are average products. The earlier term can be expressed as APL = Y/L and APC = Y/C, indicating average product of labor and capital respectively, while the latter can be expressed as AP = (Y)/(hL + kC), where Y is output, L is labor, C is capital, and h and k are weights for labor and capital, respectively. For changes in these ratios, the comparison between any two periods is taken (Swastika, 1995).

If disembodied technological change occurs in an existing production process, then it can be modeled in terms of a shift in the production surface. Embodied technological change introduces other measurement problems such as the measurement of input and output quality changes. If technological change occurs through the introduction of new processes and inputs, then the production possibilities set are both multi-product and multi-factor. In some respects it is simpler to use the dual cost or profit function (given outputs and factor prices) or an increase in profits (given output an input prices). When the technological change involves the adoption of new inputs or production of new output, for example, firms may be observed at corner solutions, where some inputs are not used or some outputs are not produced. Since the dual cost and profit functions are based on the assumption for maximizing behavior, they provide measures of technological change at the firm's optimal input and output level (Antle et al., 1988).

Many of the theoretical literature on technological change as well as most empirical research, are focused on the aggregate level of the industry or sector. In the aggregate approach, an aggregate production function is often postulated to be in the form Y = F(X, t), where Y is aggregate output and X = (X1, X2,., Xn) is the total amount of each type of input used, and t is the state of technology. Assume that F satisfies the neoclassical regularity conditions: F is a positive, increasing, concave function of X and is increasing and differentiable in t (Antle and Capalbo, 1988).

RESULTS AND DISCUSSION

The number of respondents is 310 farmers which are classified into the following: 50 are pure traditional farmers, 47 are partial traditional farmers, 53 are PRPTE farmers, 51 are NES -Durian Luncuk farmers, and 109 are NES - Bajubang farmers.

The estimated parameters of the modified translog production frontier function are highly significant different from zero. The coefficient of determination (R2) is 0.8631. It means that that 86.3% of the variations of output were determined by the behavior of changes in factor inputs and technology, while around 13.6% were determined by other factors outside the model. The F-ratio of the model is also highly significant from zero at 1% level, with an F-ratio value of 27.8644.

The presence of technical change can be identified from the estimation of stochastic frontier production function model using equation (6) by evaluation of the coefficients of group age 1, 2 and 3, and the interaction with variable inputs. These parameters can be positive, zero and negative in sign. If parameter bt is positive, then the production frontier will shift upward, indicating the improvement of technology. If bt is negative, then the production frontier will shift downward, indicating a decline of technology. Then if bt is zero, the production frontier will neither shift upward nor downward.

The magnitude of parameter btt (the square term of proxy time) indicates the rate of change of the production frontier, and the sign can be positive, zero or negative. If the sign of btt is positive, zero, or negative , then the rate of change of production frontier is increasing, constant, or decreasing, respectively.

The significance of the coefficients of Group age 1, 2 and 3, indicates that technological change has significantly changed the intercept of the frontier; otherwise, if it not significant, it means that they do not significantly alter the frontier intercept and only change the slope of the frontier.

The coefficient of T1 is not significant, which means that the presence of technological change in group age 1 only changes the slope of the frontier, not the intercept—the technology is the non-neutral type. In group age 2, the technological change exists and is significantly different from zero at 5% level, with value of 0.0052. It shows that the production frontier shifts upward with the improvement of technology and at the same time TFP also increases in group age 2. Coefficient of T3 is not significant, indicating that the technological change in group age 3 only changes the slope of the frontier, and not the intercept of the function – the technology again is the non-neutral type.

The coefficient of the square terms indicates the rate of change of the production frontier. The positive, negative or zero value of the coefficient means that the rate of shifting the production frontier is increasing, decreasing or constant, respectively. All the coefficients of squared terms, in periods Group age 1, 2 and 3, are not significant, that there will be no statistical significance in the rate of change of production frontier whether it is increasing, constant, or decreasing.

The change in elasticity of production over time can be derived by taking a second derivative of output with respect to time,-- parameter bit . If their values are positive, then there is an increase in the use of ith input, and thus the technological change is input-using. Otherwise, if the values are negative, then the technological change is input-saving.

The coefficients of LSIZT1, LHERT1, LTWKT1, and LNWDT1 are not significantly different from zero, however, its positive sign indicates that technological change is biased toward farm size, herbicide use, tapping week, and number of times of weeding in group age 1. The same pattern of biased technological change for these input is shown in group ages 2 and 3, with the coefficients of LSIZT2 & LSIZT3, LHERT2 & LHERT3, LTWKT2 & LTWKT3, and LNWDT2 & LNWDT3 showing a positive sign, which indicates that the technology is biased toward farm size, herbicide, tapping week, and number of times of weeding.

Using the same method of analysis, that the coefficient of LFERT1 has a positive sign indicates that technological change is input-using toward fertilizer in group age 1. The fertilizer-saving technological change is likely present in group ages 2 and 3, as indicated by the negative coefficients of LFERT2 and LFERT3, although they are not significantly different from zero.

The biased technological change toward stimulant is input-saving in group age 1, indicated by negative coefficient of LSTMT1. The behavior is toward biased technological change for stimulant in group ages 2 and 3 as shown by the negative sign of coefficient LSTMT2 and LSTMT3, respectively.

The coefficients of ltplt1 and LTPLT2 are negative in group age 1, and 2, respectively, which shows the biased technological change toward tapping labor is input-saving for both periods. However, in group age 3, the biased technological change toward tapping labor is input-using as indicated by positive coefficient of LTPLT3.

Equipment and number of trees show opposite directions of biased technological change for all periods. While the equipment in group age 1 is biased technological change of input-using, as indicated by positive coefficient of LEQUT1, there is input-saving of biased technological change toward number of trees with negative coefficient of LNTRT1. The same opposite directions for both inputs in group ages 2 and 3 are shown, with the coefficient of LEQUT2 negative and the coefficient of LNTRT2 positive in group age 2. The coefficient of LEQUT3 is positive in group age 3, but the coefficient of LNTRT3 is negative in the same period.

Table 2 shows the comparison of primal rate of technological change across types of farmers and across time. The primal rate of technological change is the rate of change in technology that determines both components bt and btt as the technological change, bit as the biased-technological change.

In general, the traditional farmers have high rates of technological change in group age 1, and 2 with range from an average of 13.9% to 15.1% in group age 1 and on average of 0.4% to 2.4% in group age 2. On the other hand, the farmers in the government estates have high average rates of technological change in group age 3 with a range of values from 4.4% to 4.7%. The lowest rate of technological change in group age 3 is the rate for farmers of traditional farming systems, indicating that the local variety used by traditional farmers are not good in the longer period because of the rapid decrease in potential yield compared to other varieties.

The technological change in group age 1 for all type of farmers ranges from 13.78% to 15.07%, -0.05% to 2.49 for group age 2, and 2.28% to 4.63% in group age 3. Because the technological change in group ages 2 and 3 is less than in group age 1, the technological changes in group age 1-2 and 1-3 are negative. While the technological changes in group age 2-3 are all positive except for pure traditional farmers.

Farmers of NES - Bajubang has the most set back of technological change with value of -14.28%, followed by partial traditional farmers by -14.04%, NES -Durian Luncuk farmers by -13.83%, PRPTE farmers by -13.48%, and pure traditional farmers by -11.64%. It indicates that there is a decreasing progress of technology in group age 1-2 and 1-3 because of low rate of progress in group age 2-3, respectively. In group age 1, farmers are continuously maintaining their fields and practicing the technology they got from the early stage of development under the coordination of government estates. Nevertheless, farmers cannot maintain their field properly during group age 2, causing a decreasing rate of technological change. At this period, the traditional and PRPTE farmers also have spill-over gain from the government estates farmer by learning and imitating what the farmers there are practicing.

Realizing that the productivity of their rubber trees are continuously declining in group age 2, all types of farmers are increasing input use and practicing new technology in group age 3. There is little improvement in technology at this period. The highest value of technological change in group age 2-3 is 4.62% by farmers of NES - Bajubang, and the lowest is -0.21% by pure traditional farmers. There is increasing productivity of partial and PRPTE farmers in the late period due to several special programs for productivity improvement of traditional and PRPTE farmers such as STCPP - ADB and other programs. The pure traditional farmers, however, cannot have much benefit from the technological progress.

The overall technological changes are negative in all periods except the group age 2-3. The magnitude of change is -13.60% in group age 1-2, increases to 3.80% in group age 2-3, and decreases to -10.40% in group age 1-3.

There is no significantly different in mean technological change between the farmers of NES - Bajubang and NES - Durian Luncuk in all age groups. However, there are all significantly different in the level of mean technological change among the others at 10% to 1% level in group age 1-2, 1, 5% to 1% in group age 2-3, and all at 1% in group age 1-3. It means that the mean technological progress between both government estates is indifferent, but they differ to that of other type of farmers. Besides that there are highly significantly different of mean technological change among traditional and PRPTE farmers in all age groups.

CONCLUSION

The study consists of two steps. Step one is the construction of the lifetime matrix of the perennial crop and Step two to derive technological change. All data on output and inputs are classified into three groups based on the potential yield of rubber for each variety, as follows: (1) group age 1 is the tree age range of 7 to 13 years with potential yield ranging from 0.5 to 1.9 ton per hectare, (2) group age 2, 14 to 17 years with peak potential yield of 2 ton per hectare, and (3) group age 3, 18 to 30 years with the decreasing potential yield 1.8 to 0.3kg/tree. The lifetime matrix presents all the three lifetime data of output and inputs of rubber farming so that it will provide the missing data in the early period of plantation and the future data till the end age of the tree.

The farmers are experiencing positive technological change in group age 1with values ranging from 13.78% to 15.07%1, and 2.28% to 4.63% in group age 3. However, the level of technological change in group age 2 is much lower compared to that in group ages 1 and 3. Farmers of NES-Durian Luncuk are experiencing negative technological change, which is -0.05%. The highest technological change is that of partial traditional in group age 1, that of the pure traditional in group age 2, and the NES-Bajubang farmers in group age 3.

The lower yield of the traditional farmers is mainly caused by improper maintenance of their field, especially underutilizing the necessary inputs such as fertilizers, herbicides, tapping labor, and equipment. There is an over-use of inputs such as tapping week and stimulant. Besides that, the level of technology is much lower in time of the peak season which results in lower yields.

LITERATURE CITED

AIGNER, C.D, and S.F. CHU. 1968. On Estimation the Industry Production Function. American Economic Review, 58(4): pp. 826-839.

ANTLE, J.M. and T. McGUCKIN. 1993. Technology Innovation, Agricultural Productivity and Environment Quality. In G.A. Carlson, D. Zilberman, and J.A. Miranowsky. Pp. 175-219.

CAPALBO, S.M. 1988. A Comparison of Econometric Models of U.S. Agricultural Productivity and Aggregate Technology", in J.M. Antle and S.M. Capalbo: Agricultural Productivity: Measurement and Explanation, Resource for The Future. Washington D.C.: pp, 159-188.

CHAND, R. 1994. Economics of Perennial Crops: Some Methodological Issues. Indian Journal of Agricultural Economics, 49(2): pp. 246-249.

FØRSUND, F.R. and L. HJALMARSSON. 1979. Frontier Production Functions and Technical Progress: A Study of General Milk Processing in Swedish Dairy Plants. Econometrica, 74(4): pp. 883-900.