AN ECONOMETRIC MODEL OF THE MALAYSIAN COCOA MARKETAmna Awad Abdel Hameed, Shri Dewi d/o Applanaidu , Fatimah Mohamed Arshad and Mad Nasir

Shamsudin1

AbstractThis study aims to describe the important factors affecting Malaysian cocoa industry through formulating amarket model representing cocoa production, import, domestic consumption, export demand and prices inaddition to a stock equation as a closing identity. The model is estimated using two-stage least squares utilizingannual data extending over the period 1978-2008. The results suggest the important factors affecting cocoaproduction are cocoa beans and palm oil relative prices, interest rate and government rural developmentexpenditure. The key determinants of domestic consumption are cocoa bean and palm oil prices, the Malaysianeconomic activity and lagged domestic consumption. Furthermore, the export demand turned out to be elastic toall the variables including cocoa world price, real effective exchange rate, world income and the export demandlagged one year. The cocoa stock, cocoa world price and lagged domestic price are the main important variablesinfluencing the domestic price of cocoa.Key words : Simultaneous Equations , Malaysian Cocoa , Counterfactual Analysis

Field of Research: Commodity Market, Aggregate Supply and Demand Analysis

I INTRODUCTIONThe Malaysian cocoa sector has undergone dramatic changes during the last few decades as the development ofthe cocoa industry in Malaysia follows various national policies, inter alia, National Agricultural Policy (NAP)and Industrial Master Plan (IMP). To attain national economic development on the road to maximizing incomethrough optimum utilization of resources, these national policies focus on crop diversification and value added agro-based industrialization (Azhar, 2007). Previously, with low production costs and efficient marketing structure,cocoa production was a profitable venture in Malaysia and the sector maintained an upward development in thearea and, subsequently, an increase in the production of cocoa to reach their topmost levels of over 414 thousandhectares and 247 thousand tonnes in 1989 and 1990, respectively. However, since then declining world prices, higherlabour costs, loss of production due to pests and diseases (which is the main reason of economic decay in mostproducing countries) along with a switch in the relative competitiveness of other crops (particularly oil palm) havetransposed the previous trend, when production grew at a rapid rate. In 2009, the total area under cocoa stood inthe vicinity of only about 21 thousand hectares with an estimated total production of about 28 thousand tonnes(Figure 1). The distribution of the area among plantation categories also changed. Until 1989, most of the cocoa areawas under Estate and the rest under smallholdings. During the ten years ending 1989 the minimum share of Estate was49% with an average of 59%. This ratio, however, was reversed later with a bigger share to small holdings (Figure 2).Geographically, most of the planting areas are situated in Sabah, while most of the grindings and manufacturingare centred in the peninsula.

1 Research Fellow, Institut Kajian Dasar Pertanian dan Makanan, graduate student,Faculty of Economics andManagement;; Director; Institut Kajian Dasar Pertanian dan Makanan and Dean, Faculty of Agriculture, respectively,Universiti Putra Malaysia.

2GrindingsImp

orts

ExportsProduction

Area

0

100,000

200,000

300,000

400,000

500,000

600,000

1980198

2198

4198

6198

8199

0199

2199

4199

6199

8200

0200

2200

4200

6200

8

Quan

tity (T

ons)

0

100,000

200,000

300,000

400,000

500,000

Area

(Ha)

Source: Malaysian Cocoa Board (2009)Figure 1: Cocoa Area (Ha) and Cocoa Production, Imports , Exports and Grindings (Tonnes) in Malaysia,1980-2009

E state

S mall-holdings

0

50,000

100,000

150,000

200,000

250,000

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

Area

(Ha)

Source: Malaysian Cocoa Board (2009).Figure 2: Distribution of Cocoa Planted Area in Malaysia by Category (Ha), 1980-2008

Penins ularMalays ia

S abah

S arawak

0

50

100

150

200

250

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

PLan

ted

A rea

(000

Ha)

Source: Malaysian Cocoa Board (2009)Figure 3: Distribution of Cocoa Planted Area in Malaysia (Ha), 1980-2008

3The Malaysian cocoa grindings and downstream industry has, however, expanded dramatically. Currently,Malaysia, with a total grindings of 324 thousand tonnes in 2008 is occupying the fifth position among the largestcocoa grinders in the world and the largest in the South East Asia. To fill the gap between the local production ofcocoa beans and the rising demand by the grinding industry, Malaysia increased its imports of this commodity.In 2008, Malaysia exported 8 thousand tonnes of cocoa beans, 18 thousand tonnes cocoa paste, 104 tonnes ofcocoa butter, 141 thousand tonnes of cocoa powder and 27 thousand tonnes of chocolate (Figure4). Cocoa butterwas the highest contributor to the total export earnings of cocoa beans and its products in 2008 as it accounted for63 per cent of them (Figure 5). This is attributable to the premium price received for it as a result of the specialcharacteristic of the high melting point of the Malaysian cocoa butter, which is advantageous for chocolateproducts in warm countries.

Source: Malaysian Cocoa Board (2009)Figure 4: Export Quantities of Cocoa Bean and Cocoa Products of Malaysia, 1981-2008

0

500

1,000

1,500

2,000

2,500

3,000

3,500

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

Expo

t Valu

e(RM Mn

)

Cocoa Beans Cocoa Paste Cocoa ButterCocoa Pow der Chocolate Total

Source: Malaysian Cocoa Board (2009).Figure 5: Export Values of Cocoa Bean and Cocoa Products of Malaysia, 1981-2008

4Malaysian cocoa industry plays an important role in the global market of cocoa products. In 2007, it accounted for15% of the world exports of each of cocoa butter and cocoa powder and cake compared to 8% and 6% sharesrespectively in 1990. Its share in the global exports of cocoa paste increased from 1% in 1990 to 5% in 2007.However, it decreased its share in the global exports of cocoa beans from 9% in 1990 to 1% in 2007 to benefitfrom the value added resulting from manufacturing its downstream products. Currently, 70 per cent of the locallyproduced dried cocoa beans are consumed by local grinders and manufactures. Currently, there are 11 localgrindings and more than 40 chocolate and other cocoa products manufacturers.Owing to the growing importance of cocoa to the Malaysian economy and the global cocoa industry, this studyattempts to estimate the supply and demand model of the Malaysian cocoa to investigate the relationships betweenthe different market variables. Furthermore, as mentioned previously in this study, the Malaysian cocoa industryhas undergone drastic change during the last two decades, mainly the reduction of land devoted for cocoa whichhas been justified by the decline in the relative profitability with respect to other crops specially palm oil.Recently, with increasing cocoa beans and anticipated growth in the demand for cocoa products specially byChina and India, there are calls on increasing the cocoa planting areas. This study therefore, also attempts toassess the effect of increasing the area under cocoa.The contribution of this paper to the existing literature on the Malaysian cocoa industry is twofold. First, it employsdifferent model specification. Secondly, this study uses a relatively more recent data than those used in previousstudies on the Malaysian cocoa.The remainder of the paper is organized as follows: Section II briefly reviews the literature on previous economicstudies on cocoa and the methodologies used for examining its market models, Section III outlines the empiricalmethodology and Section IV reports and discusses the results while a summary and some conclusions are presentedin Section V.

II LITERATURE REVIEWThe basic structure of models proposed in the analysis of agricultural commodity markets are formed from thecomponents of the market model approach developed by Labys (1973) which suggests that for a particularcommodity, four equations supply, demand, price and stock (commonly used as an identity to reveal the marketclearing condition) are used simultaneously. To elucidate more complex structures of the market behaviour, thesebasic market models can be tailored and reformulated. Incorporation of more variables enables the market model tobe extended. In this section, relevant previous econometric studies of perennial crops including cocoa arehighlighted with special reference to cocoa in order to understand the market model especially in the system wherethe market model works.The relatively simple generalized theoretical model of Labys (1973) has widely been applied to most of theagricultural commodities. In Malaysia, it has been applied to analyze and model the palm oil (e.g. Shamsudin et al.,1988; Shamsudin and Arshad, 1993; Shamsudin et al., 1994; Lubis, 1994 and Talib and Darawi, 2002, Ernawati,2004, Amna Awad, 2005), rubber (e.g. Yusoff , 1988a) and cocoa market (Rosdi, 1991, Yusoff et al., 1998 andKox, 2000). The flowing paragraphs will briefly review some of the recent studies on the Malaysian cocoaindustry.

A study by Rosdi (1991) investigated the main factors that determine the cocoa prices Econometric cocoa models forthe world and Malaysian markets were developed. The models were estimated using annual time series data. Eachmodel consists of supply, demand and price equations, with stock as the identity. His results show that domesticcocoa prices are determined by prices prevailing in the world market. Domestic stock change is not significant. Theworld market itself, stock and consumption are the main factors that influence the behaviour of cocoa prices. Worldconsumption and export demand are significantly influenced by the production index of the industrial nations andprice of cocoa. On the supply side, cocoa production is determined by cocoa price lagged by the gestation period.This implies that investment decision on cocoa three to five years earlier is an important factor that determinescocoa supply.

5Shamudin et al.(1993) constructed a model of the Malaysian cocoa market following the approach by Labys(1973). The model consists of production, demand and price equations in addition to stock as an identity. Theyconducted their study utilizing data over the period 1965-1987. The findings of their study indicate that theMalaysian cocoa price is dependent on the world cocoa price. The study revealed that the industrial index of theindustrialized nations is the key determinant for the demand of cocoa and consequently, its price. Moreover, theprice elasticity of supply with respect to cocoa has been found to be low (consistent with a priori expectations) andis inline with the previous studies of perennial tree crops.Yusoff et al., (1998) conducted a study to determine the factors affecting the supply and demand for the Malaysiancocoa beans using annual data over the period 1973-1992. His supply sub-model followed the work of Alias et al.,(1987) that diversified the supply into matured area and production equations. They also differentiated demand intodomestic demand, specified as derived demand, and export demand. The results of their study suggest that cocoahectarage is affected by the local prices of dried cocoa beans, the price of copra, the cost of labour, and interestrates. Production is found to be dependant on hectarage of matured cocoa, quantity of labour used, amount offertilizer applied and the level of technology used. In addition, their results show that domestic demand for cocoa isdependant on local prices of dried coco beans and palm oil as well as price of local cocoa- based beverage (MILO)and the local demand of the previous year. It is also revealed that among the factors affecting the foreign demandfor Malaysian cocoa are: the exchange rate of the Malaysian ringgit against the US dollar, level of world cocoagrinding and one year lagged foreign demand.Nasir (1998) conducted a study to explicate the economic implications of an export levy on the Malaysian cocoaindustry. He used a model consisting of five behavioural equations to describe the production, exports, domesticdemand, imports and export price; along with two identities to define the domestic price and stock level of cocoabeans. The results of his study indicate that the imposition of an export levy would lower producer prices and raiseexport prices and consequently production and exports would decline while domestic utilization and imports, onthe other hand, would increase.A study by Mohammad Haji Alias et al., (2001) has investigated the impact of government rural developmentexpenditure, a proxy for government policy, particularly, public investment on the supply response of three majorperennial export crops. Time series annual data for the period 1975-1997 have been used. The Engle-Granger twostep procedures for co-integration analysis have been applied. The response of cocoa producers to governmentexpenditure is elastic in the long run, while yield has remained fairly stable.

III METHODOLOGY1.Model SpecificationThe basic market model which was proposed by Labys (1973) has been used to develop the framework in thisstudy. The market model can be summarized into supply (Qt), demand (Ct) and price (Pt) sub models in additionto supply equal to demand (Qt = Ct ) as market equilibrium condition , is as followsQt = q (Qt-1, Pt-i, Zt, Ut) (1)Ct = d (Dt-1, Pt, Pst, At, Tt) (2)Pt = p (Pt-1, It) (3)Qt = Ct (4)where;Qt = commodity supplyCt = commodity demandPt = commodity pricePt-i= prices with lag distribution

6It = inventory or stockZt = policy variables influencing supplyPst = prices of substitute commoditiesAt = economic activity level or incomeTt = technical factorsi =1,2,3..According to Labys (1973) and Pollak (1984), it is assumed that in the system of equations, prices adjust to clearthe market. The supply of the commodity depends on the lagged supply, lagged price and policy variables.Demand is dependent on lagged demand, own price, prices of one or more substitute commodities, level ofeconomic activity and technical factors. Lagged price and changes in inventory can also be used to explain theprice. Since the supply process normally uses the general class of distributed lag functions so the lagged pricevariables are included. The market model is closed using an identity which equates quantity supplied minusquantity demanded.i.Cocoa Beans ProductionThe specification used in this study for estimating the supply response for Malaysian palm oil production is basedon the model developed by Shamsudin et al. (1993) Shamsudin(1993 and 1998) . It also follows Mohammad HajiAlias (2001) who included government expenditure and Yusof et al. (1998). by including the interest rate ofagricultural sector lagged two years. The supply equation is specified as a function of the lagged production, priceof cocoa, price of cocoa at lag two, price of competitive crop (palm oil) lagged two years, and the harvested area.So, the supply equation is specified as followsPRCt= f (PRCt-1, CPM t-2, POPt-2, HAt, GOVDE t-2, IRt-2, U1t) (5)wherePRCt = Malaysian production of cocoa at time t.PRCt-1 = Malaysian production of cocoa lagged one year .CPM t-n, = Local Price of cocoa beans lagged two years.POPt-2= Local Price of palm oil lagged two years.HAt = Cocoa beans harvested area at time t.GOVDE t-2, = Government agricultural and rural development expenditure lagged 2 years(RM million)IRt-2 = Interest rateU1t = Disturbance term.ii.Cocoa Beans DemandAccording to Yusoff et al.( 1998) and Nasir (1993 and 1998) the demand for cocoa beans is a derived demand asit used as input for producing final cocoa products such as cocoa butter, cocoa powder and chocolates etc. Whenprice determining factors change over time, firms in general do not respond immediately, but they rather, delaytheir responses to changes affecting the demand. Thus, they spread their responses over some period of time. Thenature of such response would vary from commodity to another, the differentiating factor being the durability orperishability of the commodity of interest (Labys, 1973). Due to lagged or incomplete information andadjustment cost, the relaxation of the equilibrium hypothesis in the general function has usually been achieved byspecifying a dynamic model within the framework of a general distributed model. The demand for cocoa beansdoes not often adjust immediately to the changes due to various institutional and technological rigidities.Therefore, lagged price variables were included, then, some of them were dropped from the demand equationsduring the general to specific exercise.a. Domestic DemandDomestic demand is assumed to depend on the domestic price of coca and domestic economy activity as follow:DDt= f (IPIM t , CPMt,DDt-1,U2t,) (6)whereDDt= Domestic consumption at time tIPIM t= Malaysian industrial production index at time tCPMt= Domestic Price of cocoa beans at time t.

7POPt= Local Price of palm oil at time tDDt-1,= Domestic consumption lagged one year.U2t = disturbance termb.Export DemandThe specification of export demand of cocoa beans is slightly different from the domestic demand in the variablesi.e. instead of local price we used world price of coca beans and instead of the Malaysian Industrial Production weused the world GDP as a proxy for the world economic activity . Incorporating a separate exchange rate variablein the foreign demand models was found to be very important in previous studies(Schuh, 1974; Chambers and Just(1979). Thus, export demand is modelled as a function of world GDP, real effective exchange rate, and worldprice of cocoa as follows:DEXt = f(WGDPt, REERt, WCPt, DEXt-1, U3t)(7)whereDEXt =Export demand for cocoa beans.WGDPt, = world GDP at time t.REERt, = Real effective exchange rate at time t.WCPt = world Price of cocoa beans at time t.DEXt-1= Export demand for cocoa beans lagged one yearU3t = disturbance term.c.Import DemandAs mentioned before, Malaysia imports cocoa beans to meet the rising demand by the grinding industry.The specification of the import demand of cocoa beans is similar to that of the domestic demand withslight differences in the variables i.e. instead of local price we used world price of coca beans. Thus, theimport demand function can be specified as follows:DIMPt = f2 (WPCt, IPIM t, DIMP t-1,U4t) (8)WhereWCPt = world Price of cocoa beans at time t.DIMPt = Import demand for cocoa beans.DIMPt-1 = Import demand for cocoa beans lagged one year.IPIM t= Malaysian industrial production index at time tU4t = disturbance term.iii. PriceStockholding of cocoa is assumed in the market place for the international trading. The price of cocoa isdetermined by the supply and demand. The price in the market is a result of partial adjustment process. The priceequation can be specified as price of domestic cocoa is regarded as a function of stock of cocoa, domesticconsumption, lagged price of cocoa, and world price of cocoa.PCt =f (DDt, WCPt, SCt, U4t) (9)whereCPt = Price of cocoa beans at time t.CPt-1= Price of cocoa beans at previous year.DDt= Domestic consumption at time tWCPt= World price of cocoa beans at time tSCt= Stock of cocoa in period t,U3t = disturbance termThe model is closed by Malaysian palm oil ending stock as identity

SCt = SC t-1+ PRCt + DIMPt DDt DEXt (10)Based on the rules for identification, the structural equations satisfy both the order and rank conditions. The sixendogeneous variables, ten exogenous variables and eight predetermined variables are represented in Table 1.Table 1: Definitions and Classification of Variables

8Variables Definitiona. Endogenous Variables1 PRt = Cocoa production (tonnes)2 DMPt = Imports of cocoa (tonnes)3 DDt = Domestic demand of cocoa (tonnes)4 DEXt = Malaysian exports of cocoa ( tonnes)5 CPMt = Domestic price of cocoa (tonnes)6 SCt = Cocoa ending Stock (tonnes)b. Exogenous Variables1 CPMt-2 = Price of cocoa lag 2 years (RM/tonne)2 POPt-2 = Crude palm oil price lag two years (RM/tonne)3 GOVDEt-2 = Government agricultural and rural development expenditure lag 2 years (RM million)4 IRt-2 = Interest rate lag two years (%)5 HAt = Harvested area (tonnes)6 WCPt = World price of cocoa beans7 POWPt = Palm oil world price (USD/tonne)8 IPIM t = Malaysian industrial production index at time t (2000=100)9 REERt = Real Effective Exchange rate (RM/USD)10 WGDPt = World GDP at time t (Million USD)c. Predetermined Variables1 PR t-1 = Malaysian cocoa production lag ged one year (tonnes)2 DMPt-1 = Import demand of cocoa lagged one year (tonnes)3 DDt-1 = Cocoa beans domestic consumption lag 1 year ( tonnes)4 DEX t-1 = Export lag one year (tonnes)5 CPM t-1 = Domestic price of cocoa lagged one year (RM/tonne)6 SCt-1 = Stock one period lag (tonnes)

2.Functional FormFor the appropriate specification the theory does not provide any specific suggestion on the best functional formand the most pertinent measures of variables involved in the analysis. An appropriate model was defined as onewhich produce unbiased (or at least consistent) and efficient elasticity estimates (Thursby and Thursby, 1984). Inthe current study the linear form has been rejected against the log-linear form. In log-linear form the equivalentspecifications are2:

ttttttt irhapopcpmprcprc 12542322110 ttttt dmpIPIMwcpdmp 213210

ttttt ddIPIMcpmdd 313210 tttttt dexwcpreerwgdpdex 4143210 . ttttt wcpddsccpm 5312110

Expected signsFollowing the above discussion the a priori expected signs of the regression coefficients are as follows:

0,0,0,0 4321 ,5< 0; 0,0,0 321 ; 0,0,0 321 ; 1>,2 >0, 3> 0,4>0; 0,0,0 3,21

Estimation Methods and Data Sources andThe specified econometric model is estimated using the Two Stage Least Squares method (2SLS). the systemis solved using 2SLS estimates. Macro-economic variables are generally non-stationary in natureand such non-stationary variables cause serious problems for typical inference procedures from OLS regressions.Hsiao (1997a, 1997b) investigated whether similar problems arise in the context of 2SLS regressions. His workssuggest that nothing needs to be changed in applying conventional 2SLS estimator formula to estimate the unknown

2 The lower case denotes the natural log of the variables

9parameters and formulate Wald type test statistics. One gets the same point estimates and asymptotic covariancematrix. The resulting Wald type test statistic remains asymptotically chi-square distributed. In other words, non-stationarity and cointegration do not call for different estimation methods or statistical inference procedures. Onecan just follow the advice of Cowles Commission in constructing and testing structural equation models . . . All oneneeds to do in structural model building is to follow the conventional wisdom (Hsiao, 1997a, 1997b).According to Sekhar (2003), the essence of structural equation model is an explanation of the movementofendogenous variables in terms of the exogenous variables. Following his s recommendation, all the variablesappearing in each of the equations are tested for stationarity using ADF and PP tests but the presence of a long-term equilibrium relationshipin this study is tested using the Bounds test instead of Johansen cointegration methodbecause, unlike his case, the variables in this study are not integrated of the same order.To gauge the impact of increasing the cocoa bean area a simulation excercise is coducted to find out what if thedeclining trend in the cocoa area since the early 1990s is replaced with an inrease of the are at an assumed annualgrowth of 10%.Data on the world cocoa price, the price indices of Malaysia and advanced counties as well as the real effectiveexchange rate are obtained from the International Financial Statistics of the International Monetary Fund. Thedata on cocoa production, area, imports and grinding (as proxy for domestic demand) are available at theMalaysian Cocoa Board website and FAOSTAT.

IV RESULTS AND DISCUSSIONEstimated EquationsThe econometric analysis of the data produced the results displayed in Table 2 below. The results, overall,indicate that the model described in methodology section is the proper specification to examine the marketvariables response. Moreover, the models are satisfactory in terms of their explanatory powers.The supply equation appears to fit the data better according to the diagnostics and it shows a high explanatorypower. Furthermore, all the coefficients carry the correct signs and they are all statistically significant excludingthe price of palm oil at lag 2. The price of palm oil variable is found to be negative, which indicates an existenceof a negative relationship between and cocoa production. That is a plausible result since an increase in the priceof palm oil leads to an increase in its profitability and consequently, the investment in its production at theexpense of cocoa or , in other words, the shift away from cocoa, which is highly probable as shown in thehistorical data of cocoa since 1980s (Figure 1). Due to the increase in palm oil prices a significant area of land hasbeen converted to palm oil in the early 1990s resulting in a continuous decline of cocoa area. The harvested areaof cocoa turned out to be an important determinant in supply equation due to the direct association of cocoaproduction and harvested land. The price of cocoa beans lagged two years is another factor affecting its supply,which indicates that the farmers take decision about cocoa production two years ahead. The previous yearproduction also emerged as a major determinant factor for the current one. Furthermore, the Governmentagricultural and rural development expenditure lagged 2 years has been found to be significant at 10% level. Thisresult is in line with Mohammad Haji Alias (2001) who included government expenditure in his paper as one ofthe supply determinants factors of palm oil, rubber and cocoa and found it to be significant. Finally, the interestrate variable lagged 2 years has the expected negative sign but it is statistically insignificant .Regarding the import demand, it is modelled incorporating the world price of cocoa and Malaysias index ofindustrial production as well as the previous year imports of cocoa beans. The results indicate that domesticconsumption is highly sensitive to all of them. All the determinants carry the a priori expected signs and arestatistically significant.The domestic demand specification incorporates the local price of cocoa and Malaysias index of industrialproduction in addition to the consumption of the last year. All the coefficients carry the a priori expected signsand are statistically significant. The results indicate that domestic consumption is highly sensitive to the industrialproduction index of Malaysia, which is a reasonable result as most of the domestic consumption of coca bean inMalaysia is to feed the grinding industry, which is directly affected by the development of the country industrial

10

production activity. Similar to the commodity demand in all the studies presented above, the role of the economicactivity level was found to be highly significant. The coefficient of one lagged domestic demand is highlysignificant and its value (0.66) indicates that the demand adjusts to its required level at a relatively moderate rate.The demand of cocoa beans is negatively related to its own price, which is consistent with theory, and therelationship between them is statistically significant. However, the own price elasticity of cocoa beans falls in theinelastic range. Previous studies on the demand for the agricultural primary commodities have shown that theprice elasticities, both in the short and long run were generally low and inelastic. Similar results of the studieswere obtained for cocoa by Bateman (1965), Behrman (1968), Melo (1973), Hwa (1981) Claessens (1984)Shamsudin et al., (1993) and Yusoff (2001). These findings were consistent with most of the other studies onother primary commodities which were presented earlier such as on palm oil (Yusoff, 1988b and Shamsudin et al.,1988), and on rubber (Yusoff, 1978 and Mohammad Haji Alias, 1988).The export demand estimation results show that the world GDP parameter carries the expected positive sign andis statistically significant. World income elasticity for cocoa export demand from the model compares favourablywith the results obtained by Shamsudin et al., (1993) who found 1.7640. The export demand for cocoa is alsosignificantly related to real effective exchange rates. Another important determinant of the export demand is theworld price of cocoa which is found to be negative, statistically significant. The results also indicate that exportdemand for Malaysian cocoa is highly sensitive to the world price of cocoa.

Table 2: Estimated Structural EquationsSupplyProductionprt= -1.6781*+0.000050cpmt-2* 1.9461popt-2 + 2.3354hat *** - 0.0486ir t-2 +0.1317govde t-2*+0.7046 prt-1***(-1.98) (1.83) (-1.02) (-5.30) (-0.75) (1.74) (10.47)

R2 = 0.9889 h=1.16Importsdmpt = 33.866*** 2 6.3627wcpt*** +3.3607l IPIM t ** +0 .17000dimpt-1 ***(3.23) (-4.03) (2.33) (5.61)R2 = 0.7128 h = -1.600DemandDomestic Demandddt= 0.0146 - 0.2457cpmt* +1.702 IPIM t*** + . 0.664ddt-1***(0.01) (-1.88) (10.41) (-2.13)

R2 = 0.8935 h =1.8615Export Demanddex=28130.4 8.7538 wcpt*** +8.045reert* +5.5217wgdpt*+0.9849dext-1 ***(-0.62) (-4.84) (1.76) (1.91) (10.11)R2 = 0.9251 h = -1.600Domestic Pricecpm t = 1.0105 - 0.0719 tsc ***+0.0188 tdc +0.7536wcpt***+0.3604 cpmt-1***

(-1.43) (-4.91) (1.5825) (7.44) (3.592)R2 = 0.8944 h = 1.200IdentityStock: SCt = SC t-1 + PRt + DMPt DDt DEXtNote: Number in parentheses are t-values.*** Significant at 1 percent level** Significant at 5 percent level* Significant at 10 percent levelAll the estimated parameters in the price equation, carry the expected signs and, apart from the parameter of cocoabeans stock, they all statistically significant. Although the coefficient of the domestic consumption variable hasthe expected positive sign, it is statistically insignificant. The results also indicate that the domestic price of cocoais determined by the lagged local price and the world price of cocoa. The coefficients on lagged price and world

11

price of cocoa are found to be positive and statistically highly significant. One percent increase in average worldprice of cocoa would increase the domestic price by about 0.8 percent. However, the elasticity with respect to theone year lagged local price is very low which indicates that it is more elastic to the world price than to its previousyears level.Effects of Increasing the Harvested AreaTo gauge the impact of increasing the cocoa bean area a simulation excercise is coducted to find out what if thedeclining trend in the cocoa area since the early 1990s is replaced with an inrease of the are at an assumed annualgrowth of 10%. The simulated values of all the endogeneous variables then were compared to the baselinesolutions. The counterfactual results are given in Table 3.Table 3: Simulation Average Value (1980-2008) for all the Endogenous Variables and Baseline Compared to 10%

Increase in Cocoa Bean Harvested AreaVariables Baseline Increased Harvested

AreaPercentageChange

PR 115,156 129,518 12.5DD 118257 149510 26.4DMP 153303 159873 4.3CPM 4310.8 3902.8 -9.5DEX 67191.3 67569.1 0.6

On the whole, the directions of response are consistent with theory. The increase of 10 percent in the harvestedarea of cocoa leads to about 12% increase in the estimated growth in cocoa bean production. The increase ofproduction is expected to increase the domestic and , consequently, the world stock ofcocoa beans will increasepushing its world price down. The domestic price will be affected by the world price which turned out in theestimation results to be the most equential factor affecting the domestic one . The results show that under thesimulated scenario the prices are expected to go down 9.5% lower than the baseline. The decline in the domesticprice is expected to boost the domestic demand 26.4 higher than the baseline Both import and export demandquantities are expected to increase but to lesser magnitude than the domestic demand as they are not effected bythe domestic price.The increase is in the former ones is a result of the movement of the global import and exportdemand which will be pushed up as a result of decreasing world price.

V CONCLUSIONSThe main objective of this paper is to examine the market variables of the Malaysian cocoa. The 2SLS techniqueis employed to investigate the domestic supply, domestic and foreign demand, as well as domestic price responsesto the main determinants. All the models are consistent with theory as all the variables carried the correct signs. Theresults of the diagnostics test indicate that the model described in methodology section is the proper specificationto examine the market variables response. Moreover, the models are satisfactory in terms of their explanatorypowers. The findings indicate that the cocoa production is mainly determined by its planted area and its previousyear production (significant at 1%). The domestic price of cocoa lagged two years and the Governmentagricultural and rural development expenditure lagged 2 years are also important factors at 5% significance level.The impact of palm oil price lagged 2 years, however, appeared to be insignificant. The Malaysian cocoa beansexports are influenced by cocoa world price, world GDP, the real effective exchange rate of the Malaysian ringgitalong with its previous years level. The domestic demand appears to be dependent on its own price and theeconomic activity of Malaysia. The previous year domestic price as well as the world price of cocoa emerged tobe key factors influencing its current price. The price is also, affected, to less extent, by the demand whereas stocklevel is found to be an insignificant factor. The estimated model appears to explain the cocoa market in terms ofits major structural elements. Thus, it is able to simulate the impact increasing instead of decreasing the harvestedCocoa area. The counterfactual analysis of changing the actual scenario of decreasing the harvested area duringthe last two decades by assuming another scenario with a sustained 10 percent increase in the harvested areareveals a direct effect of a 12.5% in Malaysian production of cocoa beans. The indirect effects via increasing thedomestic stock and global stock , causing the global and domestic prices to sink down. The later is expected toundergo a 9.5% decrease. The decrease in the domestic price will enhance the domestic demand i.e. grindings by

Formatted Table

Formatted Table

12

26.4 % while the drop in the world price will boost the cocoa beans trade. The Malaysian imports and exportswill increase by far less amount of 0.6% and 4.3, respectively, which is in favour to the industry because theincrease in production is required to feed the downstream industry and increase the exports of the cocoa productsrather than cocoa beans. Although increasing the area under cocoa seem to produce beneficial effect on theindustry the land limitaion factor in Malaysia confines the expansion in the cocoa production areato certain limit.Thus the increase in area should be accompanied with enhancing productivity. Further research is required toaddress the impact of increasing productivity which has been stagnant for long time.

ReferencesAbdul Rahman Lubis (1994) "Malaysian Market Model for Palm Oil: Some Policy Simulations," PhD Thesis(unpublished), Universiti Putra Malaysia, Serdang, Selangor.

Amna Awad Abdel Hameed (2005). An Econometric Study of Palm Oil Import Demand in the Middle East andNorth African Countries, PhD Thesis (unpublished), Universiti Putra Malaysia, Serdang, Selangor.

Azhar, I. (2007). The Ways towards Sustainability of Cocoa Industry in Malaysia, presentation at the ICCO RoundTable on A Sustainable World Cocoa Economy, International Conference Centre Accra , Ghana, 3-6 October2007.

Bateman, J.R. (1965). Aggregate and Regional Supply Functions for Ghanaian Cocoa, 1946-1962. Journal ofFarm Economics. 47 (2): 384-401.

Behrman, M.J. (1968). Monopolistic Cocoa Pricing American Journal of Agricultural Economics. L:702-719.Chambers, R. and Just R. (1979). A Critique of Exchange Rate Treatment in Agricultural Trade Models.

American Journal of Agricultural Economics, 61, 249-57.

Ernawati (2004). Trade Liberalization Impact on The Indonesian Palm Oil Industry," PhD Thesis (unpublished),Universiti Putra Malaysia, Serdang, Selangor

FAOSTAT online database of the Food and Agriculture Organization of the United Nations @http://faostat.fao.org/default.aspx , accessed on 30th August 2009.

Hsiao, C. (1997a). Cointegration and Dynamic Simultaneous Equations Models.Econometrica, 65(3), 647-670.________(1997b). Statistical Properties of the Two Stage Least Squares Estimator Under Cointegration.Review of Economic Studies, 64, 385-398.

Hwa, E.C. (1979) Price Determination in Several International Primary Commodity Markets: A StructuralAnalysis. IMF Staff Papers. 26:157-188.

IFS online data-base of the International Monetary Fund @ http://www.imfstatistics.org/imf/logon.aspx, accessedon 30th August 2009.

Kox , Henk M.L. (2000). The Market for Cocoa Powder: Modeling and Forecasting the Market for Cocoaand Chocolate, paper prepared for the Ministry of Foreign Affairs, The Netherlands.

Malaysian Cocoa Board (2009). Malaysian Cocoa Board website@ http://www.koko.gov.my/l accessed on 12September 2009.

Melo, H.D. (1972). An Analysis of World Cocoa Economy in 1980. Ph. D. Dissertation, North Carolina StateUniversity.

13

Mohammad Haji Alias (1988). Pembinaan dan Pemilihan Model respon Penawaran dan Pengeluar Getah AsliJurnal Ekonomi Malaysia. 18: 3-25.

Mohammad Haji Alias, Anizah Md Ali dan Maria Abdul Rahman. (2001). The Impact of GovernmentPolicy on the Supply Response of Malaysian Palm Oil, Rubber and Cocoa Producers. UtaraManagement Review. 2(1):41- 64.

Mohammad Haji Alias, Mansor Jusoh and Zulkifli Senteri, (1987). A Model of Malaysian AgriculturalSector: Preliminary Results . Paper presented in the Fourth economic Meeting (MIER), Kuala Lumpur.

Rosdi M. L. (1991). An econometric Analysis of the Malaysian Cocoa Prices: A Struc-tural Approach, Masters Thesis , UPM.Schuh, G. E. (1974). The Exchange Rate and U.S. Agriculture. American Journal of Agricultural Economics,

56, 1-13.Sekhar, C.S.C.(2003) Price formation in world wheat markets implications for policy, Journal of Policy Modeling ,

25 (2003)85-106Shamsudin, M. N., Rosdi, M L. and Ann, C. T. (1993). Malaysian Cocoa Market Model. In Arshad,F.M., Shamsudin, M.N. and Othman, M.S. (Eds) Malaysian Agricultural Commodity Forecastingand Policy Modelling. UPM: Center for Agricultural Policy Studies.

Shamsudin, M.N. (1993), The Effect of Import Liberalisation of Cocoa Beans on Malaysian CocoaIndustry. Borneo Review 4: 97-111.

Shamsudin, M.N. (1998), The Effect of an Export Levy on the Malaysian Cocoa Industry, Pertanika J. Soc. Sci.& Hum. 6(1): 23-29 (1998)

Yusoff, Mohammed (1988a). Malaysian Natural Rubber Market Model. Pertanika, II (3): 441-449.Yusoff, Mohammed (1988b). Production and Trade Model for the Malaysian Palm Oil Industry. Asean

Economic Bulletin: 169-177.Yusoff, Remali, Chee, M. and Senteri Zulkefli (1998). Supply and Demand Model of the Malaysian

Cocoa Industry , Borneo Review, 9 (1) pp: 67-86.

14

The estimated model is simulated over a 20-year period, without and with the export levy. For this exercise, a dutyof 20% of export price of cocoa beans is assumed. The average simulated values of the endogenous variables,without and with export duty, are presented in Table 2. The imposition of an export levy means that producerswill receive a smaller share of world prices. It is found that the imposition of an export levy of 20% lowersproducer prices by an average of 14.4%, and raises the export price by 3.9%. This means that producers have tobear the burden of 70.5%, and importing countries bear the burden of 29.5% of the 20% levy of export prices.Thus the outcome of the levy falls mainly on the producers rather than the foreign buyers. This is expected sincethe domestic price is basically determined by the world market, which in turn is influenced by the global factorssuch as supply, demand and stock levels. As the export price becomes relatively higher with the imposition ofexport duty, a resultant decrease in exports is expected. Exports of cocoa beans decrease by an average of 0.6%.Production also declines, at 6.4% due to lower domestic prices. The domestic utilization and imports, however,increase by an average of 0.8%, respectively, since the domestic price of cocoa beans is lower with the impositionof export duty. The increase in imports is due to the complementary relationship between imported and localbeans for blending to obtain the desired flavour and colour of cocoa products.CONCLUSIONIn light of the prevailing low cocoa bean prices, the imposition of an export duty as an instrument to discourageexports, and simultaneously encourage domestic downstream activities, should not be recommended since it willfurther lower the domestic prices and hence production. Furthermore, due to the inelastic export demand, theeffect on exports is only marginal. While it is recognized that the domestic grinding and downstream activitiesshould be encouraged, other strategies should be sought to overcome the problem of shortages of the local supplyof cocoa beans. One such strategy is production efficiency through labour-saving and land-augmentingtechnology. Like the palm oil and rubber industries, the cocoa industry is facing labour scarcity problems, i.e. theland/labour ratio is high. Labour productivity as such can be raised significantly by adopting modern technologieswhich will simultaneously improve land productivity. The progress in tree mechanization has hitherto beenrelatively slow. Universiti Putra Malaysia has conducted studies on mechanized cocoa pod breaking, cocoa seedseparation, fermentation and an integrated roasting system. With increasing shortage and rising cost of labour,there is a need to intensify research in this area. Collaboration between research institutes and other agriculturalengineering organizations abroad, as well as with local universities, is desirable in order to develop a suitabletechnology framework for more mechanized intensive operations in the cocoa industry. To date, there is still awide disparity between estates and smallholders' yield. The poor performance of the smallholder sector isattributed to non-optimum production and marketing practices and financial constraints. Farm managementpractices are still much below the recommended ones. Thus agronomic management to improve the smallholderfarming system must be emphasized to improve yield. An inventory to determine technology transfer andadoption among smallholders and factors that influence technology adoption should also be undertaken.Harvesting, postharvest, fermentation and drying practices can influence cocoa yield.Harvesting must be done at right maturity. The use of sharp knives to harvest the fruits and uniform harvestintervals should be practised. Care should be taken to ensure that the beans are not damaged during pod splitting,and the infected and healthy beans are separated. If pod storage is practised, the pods should be kept as dry andaerated as possible by placing them in the shade for not more than nine days. Cocoa beans which have undergonespreading and pod storage before fermentation would have a higher recovery rate (around 40%) compared to thosewhich are fermented immediately after harvest (33-35%). Drying under the sun or at 50-60C ifusing an artificial dryer will ensure that the resultant beans are plump and the shells are easily loosened during thewinnowing process. Proper drying also decreases the percentage of double and clump beans. During secondaryprocessing, the level and efficiency of roasting, grinding and winnowing processes must be controlled to minimizethe percentage of waste (nib in shell, dust).The transfer of technology model in Malaysia still follows the traditional approach which was introduced in the1950s. In this approach, technology is created at the research institutes and transferred to the target audiences.Such an approach does not consider the differences among the farmers in a specific locality in terms of farm size,education level, age, manpower, capital and the like. 1 ot only does this result in only the 'resource rich' farmersadopting the recommended technology but in a generally low level of adoption. While efforts in agriculturalproduction research and extension are vigorous, it appears that marketing and marketing extension have not beenfully emphasized and developed in the past. In the chain of agricultural development, marketing in general (and

15

marketing extension in particular) appears to be the weakest link. This lack of emphasis on marketing extensionhas been due to greater attention being given to other aspects of farming, lack of suitably trained personnel and thecomplexity of the marketing system. Thus it is very timely that marketing extension should be emphasized notonly to producers but also to marketing intermediaries to obtain better quality of beans and returns and henceproduction. The efforts of the Malaysian Cocoa Board and the Department of Agriculture to jointly embark on aprogramme to rehabilitate the smallholding sector are very timely. Together with the on-going rehabilitationprogramme undertaken by the estates, the productivity is anticipated to increase and will partially offset theproduction loss through reduction in hectarage.