# Sustainability On Rubber Production In Malaysia Commerce Essay

**Published:** **Last Edited:**

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The determination of sustainability on rubber production in Malaysia was analysed using the Ordinary Least Square method. However, before we could test all the secondary data by using the Double Log Regression, we first need to identify the coefficient using the Unit of Root Test, which primarily tests the characteristics of data using the Augmented Dickey-Fuller (ADF)Test. The analysis consists of series of data covering the year of 1961 to 2006. The problem statement of this paper is determined in order to respond to the growth of rubber production in Malaysia, whereby there were a few factors called production factors that need to be considered including the areas of which the rubbers were planted, labours hired, and the yields gained from the rubber production. From the analysis of production factors, it can be identified whether the Malaysian NR production can be sustained compared to other countries such as Thailand and Indonesia. In accordance to that, the objective of this paper is to determine the relationship between each variable in order to show whether Malaysia can achieve its target to sustain in NR production or not by comparing its production factors to another country, Thailand.

Rubber can be classified into two main categories; natural rubber (NR) and synthetic rubber (SR). The NR is obtained from latex, which is the milky white fluid and easy to find in many plants. However, most of the plants are coming from the Brazilian rubber-tree (HeveaBrasiliensis). In Malaysia, HeveaBrasiliensis covers 2 million hectares areas in Peninsular Malaysia. The main states that are known as the main producers of rubber latex are Johor, Kedah, Perlis and Negeri Sembilan. Other regions include Perak, Sabah, and Sarawak. Even though large areas have been covered but SR is the main threat to NR, which is due to the price of SR still competitive compared to NR. Meaning, if possible occurs fall in price of fuel in the world market will make the SR cheaper than NR. Actually, SR is formed as of unsaturated hydrocarbons.

According to MIDA (2008), Malaysia is ranked third behind Indonesia and Thailand in producing NR in the world. It produces approximately 40 percent of NR annually. Furthermore, there are 500 manufacturers in the Malaysian industry producing related rubber-based products such as gloves, tires, and shoes. For instance, in 2005, Malaysia was known as the largest producer of the rubber gloves in the world. Meaning, the contribution of rubber industry is very important to the nation, especially to the small holders of rubber plants.

## 2. Literature Review

According to Kittipol (2008), Malaysia was declared as the third largest producer of NR after Thailand and Indonesia in 2007. Malaysia produced 1.22 million tons of rubber production compared to Thailand, which was the first largest producer over the world with the volume of 3 million tons followed by Indonesia with 2.79 million tons of rubber production. Thailand has maintained its position as the largest rubber producer in the world due to its consciousness in sustaining and maintaining their position as the main NR producer in the world. Pathanasriskul (2005) elaborated on Rubber Research Institute of Thailand (RRIT) where in the report; Thailand has taken into consideration two ways in order to sustain its position, i.e. up-stream industries and down-stream industries of rubber production. For example, in the up-stream industries of rubber production, the industry does tapping training to their workers and an activity such as farmer group meetings. For down-stream industries, their objective is to increase the domestic consumption of NR within five years from the total number of rubber production worldwide.

The Economic Planning Unit (2009) stated that because of the anticipated yield improvements from 900 kilograms to 1500 kilograms per hectare in 2003 to 2005, the rubber productions were forecasted to increase by 3.8 per cent per annum in the following years. In this case, in order to ensure the sustainable of NR for the long-term period, effort would be continued in producing more the downstream furniture production. Besides that, the EPU asserted that the area of rubber plantation was expected to decline because of the use of land for oil palm plantation and others. According to Carrere (2006), in the case of Cambodia, both rubber and oil palm need larger areas of plantations. Meaning, the decreasing in the area was not a good outlook to sustain in the NR production.

According to Wahid et. al. (2008), in their journal entitled Review ofMalaysian Agricultural Policies with Regards to Sustainability; there was a significant result of contribution of small holders in the agriculture sector. But, they were also the majority group which is suffered due to uneconomic size of land and also being the lowest income group. Furthermore, because of that, the area of agriculture land declines. In contrast, in Menglun Township, Southwest China, the area of rubber plantation was rapidly expanded between 1988 and 2003 (Liu et. al., 2006). The change shows that increasing in land came from forested areas approximately covered 42 per cent(4150 ha) and fallow fields covered 23 per cent(3001 ha).

According to Kaur (2008), labour is very important factors that can give effect in the NR production. In his journal, he said that the availability and ability to control, labour would be the important part of production. It is also important in the social relations of transformation. This statement is also supported by Ke et. al. (2006) which stated that labour intensive productivity is quietly closer to NR production. This is due to their ability and skill can be maintained for a long time period. According to Mesike et. al. (2009), smallholder had a responsibility in increasing and sustaining the rubber production. This can be explained by their job characteristics, which provide a highly labour intensive, although there was a low level of productions operating.

## 3. Data and Methodology

The determination of rubber production in Malaysia was analysed by using the Ordinary Least Square (OLS) method. However, before we could test all data by using Log-Log regression, we must first identify its coefficient using the Unit of Root test, which is a test to determine their characteristics by using ADF tests. The analysis was presented by the time series of data covering the year of 1961 until 2006.

## 3.1 Analysis of Data

Regression Analysis concerns with a study of relationship between a dependent variableand other or more independent variables.

## Y= βâ‚€ + βâ‚X1 + βâ‚‚ X2+ βâ‚ƒ X3 +µ (3.0)

Source: Gujarati and Porter (2010).

The model of Regression sometimes shows the false results or doubtfulvalue, which is involving the series of data. This means, at the condition the result looks better but opposite of further investigation there have not good results.

Ordinary Least Square (OLS) is used most frequently in the regression analysis. It states that b1 and b2 should be chosen in such a way of Residual Sum of Square (RSS) and the µ is as small as possible.

1. Log-log Regression

Single:lnY= βâ‚€ + βâ‚ lnX1 +µ (3.1)

Source: Gujarati and Porter (2010).

This indicates that the changes in dependent variables associated with the changes in an independent variable.

Where in this paper:

lnY = lnProduction = Production of rubber

lnX1=lnArea = Total area of planted hectare age

lnX1= lnLabour = Total number of workers employed during the last pay period

lnX1= lnYield = Yield per hectare

Multiple:lnY= βâ‚€ + βâ‚ lnX1 + βâ‚‚ lnX2 + βâ‚ƒ lnX3 +µ (3.2)

Source: Gujarati and Porter (2010).

This indicates that the changes in dependent variables associated with the changes in one or more independent variables.

Where in this paper;

lnY = lnProduction = Production of rubber

lnX1=lnArea = Total area of planted hectareage

lnX2= lnLabour = Total number of workers employed during the last pay period

lnX3= lnYield = Yield per hectare

According to Heij et. al., (2004), the OLS can be computed as the following steps:

Step 1: Variables choosed.

X1, X3, X4...........Xk where there was constant in X which is usually take the value one.

Step 2: Data collection

The related value of X and the n observation of Y are collected. Than the data of Y would be stored in an n x 1 vector while for data in an explanatory variable would be stored in the n x k matrix X.

Step 3: Estimated computation

It was computed by using regression package by b = (X'X)-1x 'Y (3.3)

Means, rank kshould be had by matrix X. The X is an n x kmatrix requires n>k is the number of parameters, wherenismore than or equal to the number of observations. k is considerably smaller than n, which are almost required by human in a real situation.

## Augmented Dickey-Fuller (ADF)

According to Gujarati and Porter (2010), to test the stationary, the Unit Root Test should be considered. The test could be obtained as follows:

1. Regression is estimated as follows:

## ΔYt = A1 + A2t + A3Yt-1 + µt (3.4)

Letting Yt = represent the stochastic time series (Rubber Production)

Where:

ΔYt = First difference of Yt

t = Trend variable

Yt-1 = One period lagged value of the variable Y

µt = White noise

Source: Gujarati and Porter (2010).

2. Unit Root Hypothesis:

A3 is the null hypothesis, where the Yt-1 coefficient is zero or in other word there is a nonstationary of time series.

H0: The data is stationary in time series

H1: The data is no stationary in time series

3. Dickey-Fuller (DF) test:

To test the a3 which is using the estimated value A3, is zero.If the value of estimated value A3is more than the critical DF values, unit root hypothesis will be rejected. In that case, we conclude that there is a stationary in the time series. In contrast, if the value of estimated value A3 isless than the critical DF values, unit root test cannot be rejected. In this case, we concludethat there is no stationary in the time series.

After doing all the process, we can determine whether there is a problem or not by looking at the Durbin Watson values.

Durbin-WatsonStatistics shows that there is an autocorrelation problem if the result shows less than one or gets near to zero, meaning the value is far from two. In contrast, there is no autocorrelation occur when the result shows near to two.

It can be used to solve the problem of autocorrelation.

Autocorrelation shows that there is an independent of each other of error terms. This can be detected by looking at the Durbin Watson statistic. When this problem occurs, there is an underestimation of the regression coefficient of the standard errors.

According to Anuar (1998) and supported by Gujarati (1995), it also occurs in the time series which is related with the correlation the same variables and not for different variable. The relationship between two variables can be explained as follows:

ut = ρut -1 + νt, -1 < ρ < 1 (3.5)

Where:

ρ = Coefficient of auto covariance

νt= Stochastic distribution

In the classical model, it assumes that the disturbance ui in such correlation does not exist. It meansthat one disturbancetermis not influencedby other disturbance terms and not relating to other observations. It can be expressed as follows:

## E (uiuj) = 0, i≠ jor,

## E (u)= 0, E (ν2) = σu2 and E (uiuj) = 0 (3.6)

Source: Gujarati D.N, (1995)

Multicollinearity shows when there are highly correlated of each independent variable and occurs when there is highly value of R2 and only few variables are significant to explain the dependent variable. This problem can be tested by using Variance Inflation Factors (VIF) test. When the data had equal or more than 10 value of VIF centered, means the data had a multicollinearity problem. In contrast, when the data had less than 10 value of VIF centered, means there is no multicollinearity problem occurs.

## 4. Findings

In this chapter, the findings of the study will be analysed. This includes calculatingthe significance between variables, the relationships, the interpretations, and also the regression problems. The data will be analysed by using the E-view program to run a regression. The finding could be obtained as follows:

## 4.1 Stationary test

Both table 1 and table 2 show the results of the Augmented Dickey-Fuller (ADF), which classify the unit root tests for Malaysia and Thailand. The tests were applied to each variable over the period of 1961-2006 with a time trend at the variables level and at their stationary of different.

The regression analysis is done to examine the correlation between the dependent variable, which is the production of the rubber and the independent variables, which are area, labour and yield. From this regression, it will be determined whether there is a stationary or no stationary in data, which is to detect the spurious results, or of dubious value, which is involving the time series of data.We can reject or accept the Unit Root hypothesis using the DF test.

## Table 1: ADF Unit Root Test result of time series variables for Rubber Production, Area,

## Labour and Yield for Malaysia.

Variables

Level

1st difference

PRODUCTION

-2.167706

-8.936071***

AREA

-3.039886

-6.376800***

LABOUR

-2.653490

-4.728806***

YIELD

-2.278766

-8.592401***

Source: E-views 7.0 Schwarz information

Notes: ***Significant at 1% , **Significant at 5%, *Significant at 10%, critical value of ADF tests are based on one sided p-values.

Based on the ADF test, the DF value of variable (Production, Area Labour, and Yield) are more than any of critical values of proceeding DF values, unit root hypothesis will be rejected. In that situation, we conclude that there is stationary in time series. Furthermore, the result shows that all the variables are stationary significance at 1st difference and also each of them significance at 1%.

## Table 2: ADF Unit Root Test result of time series variables for Rubber Production, Area,

## and Yield for Thailand.

Variables

Level

1st difference

PRODUCTION

-1.347489

-5.209653***

AREA

-1.322762

-6.831533***

YIELD

-1.634102

-4.913268***

Source: E-views 7.0 Schwarz information

Notes: ***Significant at 1% , **Significant at 5%, *Significant at 10%, critical value of ADF tests are based on one sided p-values.

Based on the ADF test, the DF value of all variables (Production, Area and Yield) is more than any of critical values of proceeding DF values. So, these variables of time series also are stationary and the Unit Root Test would be rejected.Furthermore, the result shows that all the variables are stationary significance at 1st difference and also each of them significance at 1per cent.

## 4.2 Log-Log Regression test

## MALAYSIA

## Table 3: The Double Log estimated of Area, Labour and Yield equation over the

## period 1961-2006.

lnProduction= βâ‚€ + βâ‚lnArea +µ

(Model 1)

lnProduction= βâ‚€ + βâ‚lnLabour + µ

(Model 2)

lnProduction= βâ‚€ + βâ‚lnYield +µ

(Model3)

Constant

1.334565

6.902260

-0.512884

Coefficient

0.768991***

0.041085

1.062754***

R-squared

0.209560

0.024284

0.473431

Adjusted R-squared

0.191596

0.001593

0.461464

F-statistic

11.66522

1.070223

39.55984

Akaike info criterion

-0.092194

0.139744

-0.498401

Durbin-Watson stat

0.283515

0.223427

0.430912

Notes: Coefficients of variable is significant *** at 1%, ** at 5% and *at 10%

## lnProduction= βâ‚€ + βâ‚lnArea +µ

lnProduction = 1.334565+ 0.768991Area

Se = (1.684795) (0.225151)

t-Stat = (0.792123) (3.415439)

p Value = (0.4325) (0.0014) R2 = 0.209560

## Interpretations:

## R2 = 0.209560

It means that about 21 percent of the variation in the (log) of production is explained by the (log) of area. There is a percentage of dependent variable that can be explained by the independent variable. In contrast, there is 79 percent of dependent variable cannot be explained by the independent variable.

The low degree of explanation means it is suggesting that the model does not very well fit the data.

The partial slope coefficient of 0.768991 measures the elasticity of the rubber production with respect to the area. Specifically, this number states at 1 percent increase in area leads to a 0.77 percent increases in the rubber production.

## lnProduction= βâ‚€ + βâ‚lnLabour + µ

lnProduction = 6.902260 + 0.041085Labour

Se = (0.185027) (0.039715)

t-Stat = (37.30403) (1.034516)

p Value = (0.0001) (0.3067) R2 = 0.024284

## Interpretations:

## R2 = 0.024284

It means that about 2.4 percent of the variation in the (log) of production is explained by the (log) of labour. There is 2.4 percent of dependent variable can be explained by independent variable. In contrast, there is 97.4 percent of dependent variable cannot be explained by the independent variable.

The low degree of explanation means it is suggesting that the model does not very well fit the data.

The partial slope coefficient of 0.041085 measures the elasticity of rubber production with respect to the labour. Specifically, this number states that 1 percent increase in labour leads to a 0.04 percent increases in rubber production.

## lnProduction= βâ‚€ + βâ‚lnYield +µ

lnProduction = -0.512884 + 1.062754Yield

Se = (1.208740) (0.168968)

t-Stat = (-0.424313) (6.289661)

p Value = (0.6734) (0.0001) R2 = 0.473431

## Interpretations:

## R2 = 0.473431

It means that about 47 per cent of the variation in the (log) of production is explained by the (log) of yield. There is 47 per cent of dependent variable can be explained by independent variable. In contrast, there is 53 per cent of dependent variable cannot be explained by independent variable.

The low degree of explanation means it is suggesting that the model does not fit the data very well.

The partial slope coefficient of 1.062754 measures the elasticity of rubber production with respect to the yield. Specifically, this number states at 1 percent increase in area leads to a 1.06 percent increases in rubber production.

## Table 4: The Double Log estimated of Area, Labour and Yield equation over the

## period 1961-2006.

Particulars (Model 4)

Values

Variables:

lnArea

lnLabour

lnYield

## Coefficient

0.257312

0.111938

1.389763***

Constant

-5.289405

R-squared

0.751515

Adjusted R-squared

0.733333

F-statistic

41.33327

Akaike info criterion

-1.139155

Durbin-Watson stat

1.016990

Notes: Coefficients of variable is significant *** at 1%, ** at 5% and * at 10%

lnProduction = -5.289405+ 0.257312 Area + 0.111938 Labour + 1.389763 Yield

Se = (2.169469) (0.485409) (0.084622) (0.232363)

t-Stat = (-2.438110) (0.530093) (1.322804) (5.980991)

p Value = (0.0192) (0.5989) (0.1932) (0.0001)

R2 = 0.751515

F = 41.33327 (0.000001)

## R2 = 0.751515

It means that about 0.75 percent of the variation in the (log) of production is explained by the (log) of area, labour and yield. There is 75 percent of dependent variable can be explained by independent variable. In contrast, there is 25 percent of dependent variable cannot be explained by independent variable.

The high degree of explanation means it is suggesting that the model fits the data very well.

The partial slope coefficient of 0.257312 measures the elasticity of rubber production with respect to the area. Specifically, this number states that holding the labour and yield constant, at 1 percent increase in area leads to a 0.25 percent increases in rubber production.

The partial slope coefficient of 0.111938 measures the elasticity of rubber production with respect to the labour. Specifically, this number states that holding the area and yield constant, at 1per cent increase in labour leads to a 0.11 percent increases in rubber production.

The partial slope coefficient of 1.389763 measures the elasticity of rubber production with respect to the yield. Specifically, this number states that holding the area and labour constant, at 1 percent increase in yield leads to a 1.39 percent increases in rubber production.

## Intercept = -5.289405

Means, this number states an average value of lnProduction when lnArea, lnLabour and lnYield are zero.

## THAILAND

## Table 5: The Simple Log-Log estimated of Area and Yield equation over the period

## 1961- 2006.

lnProduction= βâ‚€ + βâ‚lnArea +µ

(Model 1)

lnProduction= βâ‚€ + β2lnYield +µ

(Model 2)

Constant

-16.84112

0.989125

Coefficient

2.180200***

0.989125***

R-squared

0.773542

0.918668

Adjusted R-squared

0.768396

0.916820

F-statistic

150.2969

496.9954

Akaike info criterion

1.319162

0.295139

Durbin-Watson stat

0.122048

0.147062

Notes: Coefficients of variable is significant * at 1%, ** at 5% and *** at 10%

## lnProduction= βâ‚€ + βâ‚lnArea +µ

lnProduction = -16.84112 + 2.180200Area

Se = (2.479137) (0.177837)

t-Stat = (-6.793136) (12.25956)

p Value = (0.0001) (0.0001)

R2 = 0.773542

## Interpretations:

## R2 = 0.773542

It means that about 77 percent of the variation in the (log) of production is explained by the (log) of area. There is 77 percent of dependent variable can be explained by independent variable. In contrast, there is 23 percent of dependent variable cannot be explained by independent variable.

The high degree of explanation means it is suggesting that the model fits the data very well.

The partial slope coefficient of 2.180200 measures the elasticity of rubber production with respect to the area. Specifically, this number states at 1 percent increase in area leads to a 2.18 percent increases in rubber production.

## lnProduction= βâ‚€ + βâ‚lnYield +µ

lnProduction = 0.989125 + 1.423787Yield

Se = (0.564473) (0.063866)

t-Stat = (1.752299) (22.29339)

p Value = (0.0867) (0.0001)

R2 = 0.918668

## Interpretations:

## R2 = 0.918668

It means that about 92 percent of the variation in the (log) of production is explained by the (log) of yield. There is 92 percent of dependent variable can be explained by independent variable. In contrast, there is 8 percent of dependent variable cannot be explained by independent variable.

The high degree of explanation means it is suggesting that the model fits the data very well.

The partial slope coefficient of 1.423787 measures the elasticity of rubber production with respect to the yield. Specifically, this number states at 1 percent increase in area leads to a 1.42 percent increases in rubber production.

## Table 6: The Double Log-Log estimated of Area and Yield equation over the

## period 1961-2006.

Particulars (Model 3)

Values

Variables:

lnlnArea

lnlnYield

## Coefficient

1.070982***

0.639566***

Constant

-1.608141

R-squared

0.999989

Adjusted R-squared

0.999988

F-statistic

1891232

Akaike info criterion

-13.74614

Durbin-Watson stat

1.030239

Notes: Coefficients of variable is significant *** at 1%, ** at 5% and * at 10%

lnProduction = -1.608141 + 1.070982Area + 0.639566Yield

Se = (0.003868) (0.001819) (0.000707)

t-Stat = (-415.7971) (588.8488) (904.1659)

p Value = (0.0001) (0.0001) (0.0001)

R2 = 0.999989

F = 1891232 (0.000001)

## R2 = 0.999989

It means that about 99.9 percent of the variation in the (log) of production is explained by the (log) of area, labour and yield. There is 99.9 percent of dependent variable can be explained by independent variable. In contrast, there is 0.1 percent of dependent variable cannot be explained by independent variable.

The high degree of explanation means it is suggesting that the model fits the data very well.

The partial slope coefficient of 1.070982 measures the elasticity of rubber production with respect to the area. Specifically, this number states that, holding the labour and yield constant, at 1per cent increase in area leads to a 1.07 percent increases in rubber production.

The partial slope coefficient of 0.639566 measures the elasticity of rubber production with respect to the yield. Specifically, this number states that, holding the area and labour constant, at 1 per cent increase in yield leads to a 0.64 percent increases in rubber production.

Intercept = -1.608141

Means, this number states an average value of lnProduction when lnArea and lnYield are zero.

## 4.3 Variance Inflation Factors (VIF) Test

## Table5: VIF test of Area, Labour and Yield in Malaysia

Variable

Centered VIF

lnArea

13.57438*

lnLabour

16.99814*

lnYield

3.663947

Notes: * shows there was equal or more than 10 value of centered VIF

This table shows that there is a multicollinearity problem occurs for Area and Labour. This is because the value of data is more than 10 value of centered VIF.

## Table6: VIF test of Area and Yield in Thailand

Variable

Centered VIF

lnlnArea

1.977359

lnlnYield

1.977359

Notes: * shows there was equal or more than 10 value of centered VIF

For this table, there is no multicollinearity problem occurs for both Area and Yield. This is because the value of data is less than 10 value of centered VIF.

## 5. Recommendations and Conclusion

In order to sustain in NR production, Malaysia is encouraged to increase their competency, expertise, and knowledge in rubber production of their rubber industry, (Seminar On Sustainability of Rubber Industry, 2009). In contrast of Thailand, the country will get the benefit if there is an optimum lead from both sides if there are well cooperated between supply and demand (RRIT, 2005). Based on the study, Malaysia needs to train their worker how to work well by providing them expertise and guidance to follow up their work performance and progress. The labour should also provide the motivated intensive such as bonuses and rewards for the hardworking workers. For the area, they should find the right area by considering its temperature and the condition of land. All of these efforts can be used to maintain the sustainability in NR production in order to compete with other producers of NR production, especially Thailand.

In conclusion, the ADF test done with both of Malaysia and Thailand had stationary in their data which does not show the spurious problem. For Simple Double Log estimated of Area, Labour and Yield of Malaysia shows that labour insignificant to explain its relationship with rubber production. All of the data also show there are autocorrelation problems occur because of their values of data are far from value 2 of Durbin-Watson. In contrast for the Thailand, its simple log-log estimated of Area and Yield show that the entire variables are significant to explain their relationship with rubber productions. All of the data also show there are autocorrelation problems occur because of their value of data is far from value 2 of Durbin-Watson. This problem of autocorrelation occurs because of the need of the rubber production to add the other variables to explain accurately relationship between dependent variable and independent variables.

For Multiple Double Log estimated of Area, Labour and Yield of Malaysia shows that area and labour are insignificant to explain their relationship with rubber production. In contrast for the Thailand of Double Log-log estimated of Area and Yield shows that all of variables are significant to explain their relationships with rubber productions. Both Malaysia and Thailand data show that there are autocorrelation problems occur in the multiple regression models. This problem occurs because of the cobweb phenomenon such as disaster, flood or rain effect to the production.

Based on the finding on a single regression model in Malaysia, the result shows only labour is insignificant variable to explain the relationship of sustainable to the NR production. This is because diseconomy of scale will occur in the long run. At that situation, the more the labour will lead to inefficiency in their production because some of the labour will not use all of their abilities in the production process. So, in this case, it will lead to affect the low of NR production. It will bring the same problem in multiple regressions for Malaysia; the result also found that labour and area are insignificant to explain the relationship of sustainable to the NR production. The reason for labour is that they are not competent enough in NR production and for area because of most of the land are unsuitable to plant the rubber such as their temperature and the condition of land.