World Applied Sciences Journal 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011 ISSN 1818-4952 © IDOSI Publications, 2011

A Time Series Analysis of the Relationship Between Total Area Planted, Palm Oil Price and Production of Malaysian Palm Oil 1

1

Fadli Fizari Abu Hassan Asari, 1Nur Hayati Abd Rahman, 1Errie Azwan Abdul Razak, 1 Bashir Ahmad Shabir Ahmad, 1Nurul Fahana Aini Harun and 2Kamaruzaman Jusoff

Faculty of Business Management, Universiti Teknologi MARA, 23000 Dungun, Terengganu, Malaysia 2 Faculty of Forestry, Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysia

Abstract: Malaysia is a very well-known country in the oils and fats sector especially palm oil because it is the world’s biggest producer and exporter of the commodity. This study was conducted to analyze the relationship between total area planted and palm oil price with production of palm oil in terms of magnitude and direction using the time series analysis method. Johansen cointegration technique, error correction model and Granger causality tests were used to estimate those relationships. The findings showed that the total area planted and palm oil price have negative relationship towards production of Malaysian palm oil. On the other hand, there is no causality relationship between total area planted and production of Malaysian palm oil in the short run. However, there is a unidirectional causality relationship between palm oil price and production of palm oil in Malaysia. For future recommendation, it is suggested that other researchers will supplement this research by integrating other factors that might affect the production of palm oil such as climate change and geographical area. Key words: Total area planted

Palm oil price

Production of palm oil

INTRODUCTION

Time series analysis

STATA

Palm Oil Industry in Malaysia (1960-1980): Prior to the independence of Malaysia, the economic development of Malaya was highly dependent on the agricultural sector, particularly on the plantation of timber and rubber [1]. After 1957, the government had conducted various programs that to help diversify the agricultural industry. One of the initiatives conducted was to develop an exchange program with the West African economies as well as other private plantations companies. The research and development activities were conducted particularly in the area of palm oil industry. Later, it formed the Oil Palm Genetics Laboratory. Moreover, the government also established Kolej Serdang, which became the Universiti Pertanian Malaysia (UPM) in the 1970s to train agricultural and agro-industrial engineers and agrobusiness graduates to conduct research in the field [6]. In the late 1970s, producers began to integrate the refineries of palm oil with plantations. In order to improve the profitability of the refinery sector, the government had decided to impose huge export taxes on crude palm oil. As a result, the price of crude palm oil reduced drastically and it then resulted in an increase in the profits.

The oil palm tree is also called as Elaeis Guineensis Jacq. It originated from West Africa. In 1870’s, British had introduced this plant to Malaya (Malaysia) as one of the agricultural plants for the development of the agricultural industry [1]. This plantation started to became commercialized after the first commercial plantation was established in Tennamaran Estate, Selangor [2]. As of today, palm oil had contributed the most in the Malaysian agricultural sector as compared to other plantations such as timber and rubber [3]. This industry alone had created employment opportunities for up to 570,000 people in 2009. Besides, the products of Malaysian palm oil are currently being consumed by people from 150 countries [3]. It illustrates the importance of the palm oil industry in Malaysia. In relation to that, the global demand for palm oil is expected to increase in future from 45.5 million tonnes in 2010 to 63 million tonnes in 2015 [4]. Even though the price of palm oil is the cheapest among other edible oils, the palm oil’s price has reached RM 3,000 per metric ton since 2008 [5].

Corresponding Author: Nur Hayati Abd Rahman, Faculty of Business Management, Universiti Teknologi MARA Terengganu, 23000 Dungun, Terengganu, Malaysia. Tel: +609-8400400 E-mail: [email protected].

34

World Appl. Sci. J., 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011

Studies on Malaysian Palm Oil Market: A structural model was developed to describe the Malaysian palm oil industry from 1970 to 1999, taking into account the total palm oil area, oil palm yield, domestic consumption, exports and imports [12]. It was found that the previous years of technology and total palm oil significantly influenced the current total area of palm oil. In 1960, there were 60,000 hectares of oil palm plantation and it is expected to increase up to 5.1 million hectares in 2020 [13]. As the plantation areas are increasing rapidly, the total palm oil production is expected to increase from 91,793 million tonnes in 1960 to 18.81 million tonnes in 2020. Even though the expansion of the palm oil plantation helps to increase the total palm oil production, it does not reflect the technological and scientific advances made. Hence, the expansion of the planted area is not the main reason that contributed to the increase in the palm oil production in Malaysia. Indirectly, it means that there is negative relationship between total area planted and production of palm oil in Malaysia. Maturity of area planted, total area planted, replanting and yield influence the supply of palm oil in Malaysia [14]. It is reported that there were 3.67 million hectares of land which have been planted with oil palms in 2002. It is expected that the production will keep increasing in the future due to the increase in the planted areas that have entered the maturity stage [12]. The price of crude palm oil started to reduce in 2005 due to an increase in its production and a reduction in the price of its substitute goods; namely soya bean oil [15]. Hence, there is negative relationship between the price of palm oil and its production in Malaysia. The relationship between the supply of Malaysian palm oil and its determinants has been widely examined by using Johansen cointegration technique and error correction model [16]. The result shows that the palm oil production is responsive to its relative price, government’s support and interest rate, both in the long run and short run. From previous literatures, it is found that the price of palm oil is quite volatile. Given that the cost of production will increase as the supply of palm oil increases, it is necessary to stabilize the price of palm oil in order to maximize profits. Generally, there are various reasons that can be attributed to the decline in the price of palm oil, such as the maturity of area planted, total area planted, replanting, palm oil price and yield. Indirectly, the instability of the palm oil’s price will affect the production of palm oil [8]. Thus, the objective of this paper is to analyze the influence of total area planted and the price of palm oil on the production of palm oil in Malaysia.

Later, as the production of crude palm oil increased, the government aggressively promoted this sector by encouraging producers to diversify the production of crude palm oil into various end uses. It then can be used to export palm oil products to other countries. During that time, Western Europe was the main importer of the Malaysian crude palm oil’s products [7]. Palm Oil Industry in Malaysia (1981-2000): Western Europe was among the major importers of the Malaysian palm oil products from 1960 to 1980. However, in early 1980s, there was a campaign set by the American Soybean Association (ASA) that spread the news that the consumption of palm oil might give bad health implications. Due to that scenario, Malaysia needed to find new markets to export the palm oil products. At the same time, it was highly crucial to conduct scientific research that can refute the campaign made by ASA. Therefore, Malaysia had started to invest in scientific research and export the palm oil products to other international markets such as China, Vietnam, Pakistan, Egypt, UK, Mexico and Netherlands. It then created an increase in demand for palm oil products. The world palm oil production in 1990 doubled to 11.0 million tonnes from a mere 5.0 million tonnes in 1980 [7]. The following decade, the production doubled to 21.8 million tonnes by the year 2000. About half of the world palm oil production (10.8 million tonnes) was accounted for by Malaysia. Thus, Malaysia is considered as the second largest producer of palm oil products [8]. Besides Malaysia, other palm oil producing countries also recorded favourable growth in production during this period. Palm Oil Industry in Malaysia (2001-onwards): Apart from timber and rubber, palm oil is the biggest contributor to the Malaysian economy in the agricultural sector [9]. Almost 67 percent of the agricultural land in Malaysia is used for palm oil plantation [3]. In December 2006, the Malaysian government had created the world’s largest oil palm plantation by initiating a merger between Sime Darby Berhad, Golden Hope Plantations Berhad and Kumpulan Guthrie Berhad. Once the merging process was completed, this big entity was called Sime Darby Berhad [10, 11]. This merger resulted in an increase in the global production of palm oil by 5% in 2006. In terms of its contribution to the national economy, this merger had resulted in an increase in the level of employment as well as exports. As reported by MPOC, the export values of palm oil had increased tremendously from 2000 to 2010 [3]. 35

World Appl. Sci. J., 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011

Methodology: Annual data from 1972 to 2008 were used where the data from 1972 to 1979 were gathered from Department of Statistics Malaysia, while the remaining were gathered from library of Malaysian Palm Oil Board (MPOB). The general long-run model can be written as follows: productiont = f(areat, pricet) (1)

Augmented Dickey Fuller (ADF) test should be adopted. This test is conducted by “augmenting” the preceding three equations by adding the lagged values of the dependent variable (?Yt). It is based on the regression equation by including a constant and time trend. ∆Yt =

Where, productiont represents the production of palm oil in Malaysia; areat represents the total area planted; while pricet represents the price of palm oil within a given period of time. The subscript, ‘t’ in each variable represents time trend. These three variables are then transformed into the linear logarithm form; i.e. lnproduction (log of production of Malaysian palm oil), lnarea (log of total area planted) and lnprice (log of palm oil price). The multivariate relationship represented by equation 2 was investigated by the researcher: In(productiont) =

+

1

In(areat) =

2

0+

t

+ Yt −1 +

m

∑ i =1

i ∆Tt −i

+

t

(3)

Where Yt = Variables of interest in the logarithm forms at time trend t, Yt-i expresses the first differences with k lags, is the white noise residual of zero mean and constant variance. The coefficients { 0, , µt, 1,…, k} are parameters being estimated. The null and the alternative hypothesis for the existence of unit root in variable Yt is; H0: H1:

= 0 (Yt is non stationary or contains a unit root) 0 (Yt is stationary or non unit root)

Phillips-Perron Test: This type of unit root test was developed by Phiillips and Perron (1988) [16]. The main difference between Phillips-Perron (PP) test and ADF unit root test is on the perspective of how the serial correlation and heteroskedaticity problems in the errors are handled. In particular, ADF uses a parametric autoregression to approximate the ARMA structure of the errors in the test regression. In opposite, PP tests ignore any serial correlation in the test regression. Phillips and Perron (1988) use nonparametric statistical methods to take care of the serial correlation in the error terms without adding lagged difference terms [20]. Besides that, the PP test deals with potential serial correlation in the errors by employing a correction factor that estimates the long run variance of the error process with a variant of the Newey–West formula.

In(pricet) + µt (2)

Since there are two independent variables in the equation, there can be more than one co-integration vector. In this context, the variables in equation (2) may feature as part of several equilibrium relationships governing the joint evaluation of the variables. The model is derived into log model since the result can be used to determine the elasticity of each variable. Four stages of analyses which are involved in this paper are, stationarity test or unit root-tests, cointegration test, stability test and Granger causality. Unit Root Test: Over the past decade, the unit root tests in autoregressive time series models have received considerable attention in the econometric literature [17]. The unit root tests are the test for the order of integration. If a series is proven to be a non-stationary series, it needs to be transformed into stationary series by using a method of differentiation. This is because if a nonstationary series is incorporated in a time series analysis, the results might be spurious. The most popular unit root tests are Augmented Dickey-Fuller (ADF) and PhillipsPerron (PP).

Cointegration Tests: A critical element in the specification of VEC models is the determination of the lag length of the VEC [21]. Various lag lengths selection criteria are used by different researchers such as final prediction error (FPE), Akaike Information Criterion (AIC), Schwarz Criterion (SC) and Hannan-Quinn Information Criterion (HQIC) [22]. In this paper, the researcher is considering results from all the criteria listed above. After completing the stationarity test for each variable, the level of cointegration between the examined variables need to be investigated. The purpose of this test is to investigate whether the stochastic trends in the examined variables which are supposed to contain unit roots have long term relationship or not. There are two

Augmented Dickey-fuller Test: The classical regression model requires that all variables should be stationary in order to avoid spurious regression [18]. If the error term is not correlated, the presence of a unit root or nonstationary can be tested by using Dickey Fuller (DF) test [15]. However, if the error term is correlated, the 36

World Appl. Sci. J., 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011

main approaches for the cointegration test; namely Engle and Granger (1987) and Johansen and Juselius (1990). Between these two, Johansen and Juselius approach is more powerful than the Engle Granger’s approach. Hence, the researcher adopted the Johansen and Juselius approach in examining the cointegration between the examined variables. Two test statistics can be used to identify the number of cointegrating vectors (r) namely maximum eigenvalue and trace statistics [23]. If r is 0, it means there is no cointegrating relationship between those variables. However, if r is one, it means there is one cointegrating relationship between the examined variables. The r can be more than one as well, which indirectly means there are more than one cointegrating equations [24]. The analysis is based on the following equations: Y1 = A1 Yt–1 + A2 Yt–1 +...+ Ap Yt–p +

t–p

over-fit the causality test underlying model with additional d-max lags – so that the VAR order is now p = k + d, where k is the optimal lag order. This modified version of the Granger causality test is done by estimating a two-equation system: k +d max

k +d max

∑

productiont = 1+ i Area1−1 +

∑

=i 1 =i 1 k +d max k +d max

∑

∑

Areat = 1+ i Area1−1 + =i 1 =i 1

i Productiont −1 + t

i Productiont −1 + t

(5)

Where d – max is the maximal order of integration of the series in the system and µt and v t are error correction terms that assumed to be white noise. The Wald test was then applied to the first k coefficient matrices using the standard X2 -statistics. The null hypothesis set for equation (5.3) is i = 0Ai k and for equation (5.4) is Øi = 0Ai k. From equation (5.3), area “granger-causes” production if its null hypothesis is rejected and from equation (5.4), production “Granger-causes” area if it’s null hypothesis is rejected. Unidirectional causality will occurs between two variables if either null hypothesis of equation (5.3) or (5.4) is rejected. Bidirectional causality existed if both null hypotheses are rejected and no causality existed if both null hypotheses are not rejected. The lag selection is a crucial step in non-causality test especially when theory and statistical result indicate a small number of lags in the VAR component. In order to choose the optimal lag length (k), the Akaike Information criterion (AIC) and Final Prediction Error (FPE) are implemented.

(4)

Where Yt is a k-vector of non stationary 1(1) variables, A with i=1,…, is a lag operator and t is a white noise residual that has zero mean and constant variance. The lag order can be determined by using Akaike’s Information Criterion (AIC). The power of r that illustrates the number of cointegrating vectors can be tested by using two likelihood ratio test statistics. Vector Error Correction Model (VECM): Once the variable was found to be co-integrated, a corresponding error correction model needs to be formed in order to estimate the short run relationship between the examined variables. Through the error term, the VECM establishes an additional channel for Granger causality to emerge, a channel that is ignored by the standard Granger and Sims tests employed in the earlier works.

RESULTS AND DISCUSSION Unit Root Test: ADF and PP tests have been conducted in order to examine the stationarity status for each series of variable. The results were given in Table 1. The MacKinnon approximate p-value for Z(t) for both tests is lesser than 1%, 5% and 10% for all variables except for total area planted. The p-value for total area planted for ADF test is significant at 10% significance level, while PP test is significant at 5% level of significance. Since 5% and 10% are weak significance levels, it suggests that the null hypothesis of unit root cannot be rejected and all the series under study are non-stationary at their level form. At first difference, the MacKinnon approximate p-value of Z(t) for all variables is equal to 0.000. It means that the result is significant at 1% significance level and stationary at first difference.

Granger Causality: Granger causality analysis (GCA) is a method to investigate whether a variable can granger cause the other variables or not (Granger, 1969). This method is based on multiple regression analysis. Although cointegration indicates the presence or absence of Granger causality, it does not indicate the direction of causality between the variables. Thus, the causality test helps us to verify whether change in any series can be explained by the other two series [23]. This study used the more recent [25] non-causality test to establish the directional of causation between the two variables. The procedure essentially suggest the determination of the d-max, namely, the maximal order of integration of the series in the model and to intentionally 37

World Appl. Sci. J., 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011 Table 1: Results of Unit Root Tests P-Value -----------------------------------------------------------------ADF PP

Form

Variables

Level

Production Area Price

0.8211 0.0633 0.1311

0.8701 0.0172 0.1713

First Difference

Production Area Price

0.0000 0.0000 0.0000

0.0000 0.0000 0.0000

Table 2: Lag Order Selection Criteria Lag 0 1 2 3 4

LL

LR

-1272.33 -1138.8 -1125.26 -1119.47 -1114.27

df

267.08 27.073* 11.579 10.407

9 9 9 9

p

FPE

AIC

7.4e+29 3.9e+26 3.0e+26* 3.8e+26 5.2e+26

0.000 0.001 0.238 0.319

HQIC

77.293 69.7452 69.4703* 69.6649 69.895

77.3388 69.9283 69.7907* 70.1226 70.49

SBIC 77.4291 70.2894* 70.4226 71.0253 71.6636

Notes: *indicates lag order selected by the criterion Table 3: Results of Johansen Tests for Cointegration Maximum Rank (r) Parms 0 1 2 3

12 17 20 21

LL

Eigenvalue

47.885362 60.315658 67.159025 68.712431

0.50850 0.32365 0.08494

Trace Statistic

5% Critical Value

41.6541 16.7935 3.1068*

29.68 15.41 3.76

Table 4: VECM Estimation Beta

Coef.

Std. Err.

Z

P>IZI

95% Conf. Interval

lnproduction Lnarea Lnprice _cons

1 -0.7039237 -2.152894 10.31768

0.2061479 0.5040023

-3.41 -4.27

0.001 0.000

-1.107966 -3.140721

-0.2998813 -1.165068

Table 5: Pair-wise Granger Causality Tests Null Hypothesis lnarea does not Granger cause lnproduction lnproduction does not Granger cause lnarea Lnprice does not Granger cause lnproduction Lnproduction does not Granger cause lnprice

Determination of Lag-Length: Based on the results shown in Table 2, the researcher had decided to use two lags for the Johansen Cointegration test and Granger Causality test. This is because majority of the selection criteria (LR, FPE, AIC and HQIC) indicated that two lags were sufficient. The lag order selection is consistent with the practice made by various researchers before conducting Johansen cointegration test and Granger Causality test [21].

Probability > chi2

Decision

0.6994 0.7433 0.1917 0.0064

Do not reject Do not reject Do not reject Reject

implies that two cointegrating relations exists between all variables. Indirectly, it means that the production of palm oil and other explanatory variables in the model move closely to achieve the long run equilibrium. It is necessary to identify the relationship between dependent variables with independent variables. Based on Table 4, the estimation of the relationship between variables can be derived as follows: Lnproduction = 10.3177 - 0.7039 (Lnarea) - 2.1529 (Lnprice) (6)

Johansen Cointegration Test: Table 3 shows the results of the Johansen cointegration test based on maximum eigenvalue and trace statistics. The value of trace statistics is smaller than 5% critical value when r is 2. It

In the long run, one percent increase in the total area planted will decrease the production of palm oil by 0.7%. 38

World Appl. Sci. J., 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011

It is inconsistent with the initial expectation whereby the production of palm oil should increase as the total area planted increases. Even so, the long run negative relationship between these two variables is supported by Jalani due to the limitation of edaphic conditions [13]. However, if the palm oil price increases by one percent, the production of palm oil will decrease by 2.15%. It implies that the production of palm oil is very elastic or sensitive to the changes in its price. This result is consistent with the research made by [15].

Therefore, it is recommended that other researchers will supplement this research by integrating other factors that might affect the production of palm oil such as climate change and geographical area. REFERENCES 1. 2.

Granger Causality: According to Table 5, it is found that lnarea does not granger cause lnproduction; and lnproduction does not granger cause lnarea. This is because the p-values are insignificant at five percent confidence level. In other words, the total area planted and the production of palm oil in Malaysia does not influence each other. This result is consistent with the research done by [26]. Apart from that, it is also proven that lnprice does not granger cause lnproduction since the p-value is 0.1917. However, lnproduction does granger cause lnprice [27]. In this case, price of palm oil does not influence the level of palm oil production in Malaysia. In the opposite case, the production of palm oil in Malaysia can influence the palm oil price.

3.

4. 5. 6. 7. 8.

CONCLUSION This study presents an analysis of the domestic structure of the Malaysian palm oil market. The Malaysian palm oil industry has undoubtedly made significant contributions towards the domestic economy as well as to the development of the world palm oil market. This paper developed a simple theoretical model that integrates the factors that influence the production of palm oil in Malaysia. Using 37 observations of palm oil production, total area planted and palm oil price from 1972 to 2008, the researcher achieved both research objectives. The findings show that the production of palm oil in Malaysia can influence its price level. On the other hand, there is no causality relationship between total area planted and the production of palm oil in Malaysia. As a result, the total area planted and palm oil production does not influence each other in the short run. In the long run, there is negative relationship between the production of palm oil with the total area planted and the palm oil price. The researcher believes that there are other factors that may affect the performance of the palm oil production in Malaysia.

9.

10. 11.

12.

13.

39

MPOC, 2009. Oil Palm Tree. Malaysian Palm Oil Council, Malaysia. Laan, T. and G.P. Lopez, 2008. Government Support for Biodesel in Malaysia. International Institute for Sustainable Development, Geneva, Switzerland. Basiron, Y., 2009. Trends and Potentials of Malaysia’s Plantation Sector. A presentation by Tan Sri Datuk Dr Yusof Basiron at the Perdana Leadership Foundation Seminar 2009 at Sime Darby Convention Center, August 20, 2009. Adnan, H., 2011. Global Demand for Palm Oil Growing Rapidly, The Star, Malaysia. Business Times, 2010. Palm Oil Breaches RM3,000 Since 2008, Malaysian Palm Oil Council. Hartley, C.W.S., 1988. The Oil Palm. Longman Scientific and Technical, Harlow, U.K. Basiron, Y., 2002. Palm Oil and Its Global Supply and Demand Prospects. Malaysian Palm Oil Board, Malaysia, 2(1): 1-10. Abdullah, R. and M.L. Lazim, 2006. Production and Price Forecast for Malaysian Palm Oil. Oil Palm Industry Economic Journal, Faculty of Information Technology and Quantitative Sciences, Universiti Teknologi MARA, Selangor, 6(1): 39-45. Lam, M.K., K.T. Tan, K.T. Lee and R.T. Mohamed, 2008. Malaysian Palm Oil: Surviving the Food Versus Fuel Dispute for a Sustainable Future, Renew Sustain Energy Review. EPU, 2006. Strengthening Agriculture and the Agro-Based Industry. Economic Planning Unit, Malaysia. Sime Darby, 2006. Synergy Drive Forms Merger Integration Committee: Eight Companies Set To Sign Conditional S&P Agreement. Sime Darby Press Release, Malaysia. Talib, B.A. and Z. Darawi, 2002. An Economic Analysis of the Malaysian Palm Oil Market. Faculty of Economics, Universiti Kebangsaan Malaysia, Oil Palm Industry Economics J., 2(1): 19-27. Jalani, B.S., Y.A. Basiron, K.W. Darus and N. Rajanaidu, 2002. Prospects of Elevating National Oil Palm Productivity: A Malaysian Perspective. Malaysian Palm Oil Board, Malaysia, Oil Palm Industry Economic J., 2(2): 1-9.

World Appl. Sci. J., 12 (Special Issue on Bolstering Economic Sustainability): 34-40, 2011

14. Abdullah, R., 2003. Short-Term and Long-Term Projection of Malaysian Palm Oil Production. Malaysian Palm Oil Board, Malaysia, 3(1): 32-36. 15. Kui, C.F., 2006. Annual Palm & Lauric Oils Conference & Exhibition: Price Outlook 2006/2007. Minister of Plantation Industries & Commodities, Kuala Lumpur, Malaysia. 16. Alias, M. and T.C. Tang, 2005. Supply Response of Malaysian Palm Oil Producers: Impact of Interest Rate Variations. Monash University Malaysia, Oil Palm Industry Economic J., 5(2): 11-22. 17. Ahmad, N., Fadli Fizari and S. Mat Zin, 2010. Exchange Rate and the Demand for Export in China. In the proceedings of the Terengganu International Business and Economics Conference 2010, pp: 420-436. 18. Granger, C. and P. Newbold, 1974. Spurious Regression in Econometrics. J. Econometrics, 2: 111-120. 19. Dickey, D.A. and W.A. Fuller, 1979. Distribution of the Estimators for Autoregressive Time Series with a Unit Root. J. the American Statistical Association, 74: 427-437. 20. Phillips, P.C.B. and P. Perron, 1988. Testing for Unit Roots in Time Series Regression. Biometrika, 75: 335-346.

21. Usman, S. and F. Asafo, 2007. Testing Granger Causality with Application to Exchange Rates for Swedish Kronor with GB Pound and US Dollar. Lund University, School of Economics and Management, Department of Statistics. 22. Zaytsev, O., 2010. The Impact of Oil Price Changes on the Macroeconomic Performance of Ukraine. Kyiv School of Economics. 23. Samsu, S.H., A. Mat Derus, A.Y. Ooi and M.F. Ghazali, 2008. Causal Links between Foreign Direct Investment and Exports: Evidence from Malaysia. International J. Business and Management, 3(12): 177-183. 24. Parlow, A., 2010. VEC-Model in Stata. UMW Econ Department. 25. Toda, H. and T. Yamamato, 1995. Statistical Inference in Vector Autoregressions with Possible Integrated Processes. J. Economics, 66(1-2): 225-50. 26. Duasa, J., 2007. Malaysian Foreign Direct Investment and Growth: Does Stability Matter?. J. Economic Cooperation, 28(2): 83-98. 27. Granger, C.W.J., 1969. Investing Causal Relations by Econometric Models and Cross-Spectral Methods. Econometrica, 37: 424-438.

40