Domestic Factors That Affect the Price of Styrene Butadiene Latex in Indonesia

This paper investigates or examines the causal links between price of styrene butadiene latex (SBL) and other domestic variables such as supply, demand, Herfindahl Hirschman Index (HHI), price of styrene, and price of butadiene during the period of 1995–2017 on the monthly basis.. Results from Augmented Dickey Fuller (ADF) or unit root test showed that the stationary state of the whole variables is of first difference (I). The Granger pairwise causality test revealed a bidirectional causality of SBL with BTD(Butadiene), SBL with STY(Styrene), SUPP(Supply) with DMD(Demand), DMD with STY, DMD with BTD and STY with SUPP and vice versa. Johansen test found out a co-integration of SBL prices. However, this study proved no significant links between the price of SBL and supply of SBL (SUPP) in the long and short run. This paper highlights the fact that SBL price has a significant direct impact in the long run on the styrene (STY) and butadiene (BTD) prices, as well as HHI. Similarly, these aspects also have a significant direct impact proved by the positive coefficient in the long run on the SBL pricing. The findings further showed that the price of SBL has a connection and significant direct impact on the negative arrow for the short and long run equilibrium, in line with the price and demands for SB Latex, pressing and oil classification.

The derivation of error correction model (VECM) vector is based on the theorem by (Usman & Aliyu, 2009) (Johansen, 1990). Suppose {Z} is a degree of VAR to-p and Zt = {Y: X}, where Y is the variable of Endogenous and X is the variable of Exogenous. It can be expressed as follows: (1) where: ry.
= Gaussian Error Term. = vector of variables are stationa One vector Zt has error correction representation if it can be expressed as follows: (2) where: (3) There are two ways to solve the regression equation: First, provides a unified maximum likelihood procedure in which α and β obtained from the decomposition of the matrix Π. Both (Engle & Granger, 1987) proposed a two-step estimation using cointegration regression, so that β Zt-1, residual estimation (estimated residue) is included in the regression equation above. This study will adopt the Johansen procedure.
One restriction that will be incorporated into the model estimates the VEC on the long-term coefficient (β). The theoretical model shows that β is not full rank matrix. Due to the small sample size, inclusion of all the variables lagged first difference in each VEC equation, which will reduce significantly the degree of freedom. Therefore, to overcome these problems, the procedure to follow is as suggested by (Usman & Aliyu, 2009) [Aliyu, 2014]. The model will be divided into several blocks, and then the variables included in the model are based on theoretical models. Based on the division, six blocks are then formed in the VEC SBL, VEC SUPP, VEC DMD, VEC HHI, VEC OIL, VEC STY and VEC BTD., so the Error Correction Model equation model can be written as follows: (5) where: L = Operating Lag (LZ = Zt-1), αno is a vector (nx 1) intercept, am xn is the coefficient matrix (mxn), єn xt is the coefficient of the error correction (error correction term), and Δ is the first difference order used to reduce stationary variables.
Unlike other procedures, methods of dynamic equations proposed by Johansen integrate long-term and shortterm relationships in one unit at a time. One is to look through the causal relationship between the dependent variables, between the dependent variables and independent variables, and between the independent variablesat the optimum lag (to see the short-term connections) and through relations of causality additional error correction (error correction channel), to see the long-term and short-term relationships.

Unit Root Tests
Augmented Dickey-Fuller Test is used to examine the unit root (Unit Root Test) in terms of the stationary states of variable price of SB latex (SBL), SB Latex production capacity in Indonesia (SUPP), the level of demand SB Latex by the user (DMD), a market segment producer SB Latex in Indonesia (HHI), the price of the main raw materials styrene (STY) and butadiene (BTD).To include trend and intercept, with provisions, the tested hypothesis is Ho: β1 = 0 (indicating the presence of unit root test and no stationary state) and H1: β1 ≠ 0 (indicating the absence of unit root and the presence of stationary state). Here, β1 is the value of the ADF. If the absolute value of the ADF is greater than the critical value of HO hypothesis,and if it states that the data containing unit root are rejected, it means

A C C E P T E D M A N U S C R I P T
the time-series data are not stationary. Likewise, if the absolute value of the ADF is smaller than the critical value of HO hypothesis, stating the data are accepted, it can be concluded that the time-series data are stationary unit root or data. Unit root test results are as follows: To make sure that the data are stationary, the unit root test in the first difference of data needs to be carried out. Test results using the ADF test shown in Table 1 indicate that all endogenous variables or variables of the study have been stationary at the 5% significance level. This means that all the variables are above stationary in the first difference,thus it can be said that variables are integrated on the degree of one or I.

VAR Stability Test
The next step is to conduct the VAR stability test or VAR stability condition check. If all the roots of the polynomial function are inside the unit circle or if the absolute value is smaller than one, the VAR model is considered stable. VAR stability test results are as follow. From the data in Table 2, SBL, SUPP and DMD, STY, BTD, and HHI, the whole root and modulus are 0.42208 (less than one/1). None of its roots lies outside the unit circle, so that the VAR model is proven to be in a stable condition.

3.3.Optimum Lag Test
An important step that must be done in the VAR-VECM modelis the determination of the optimal number of lag used in the model. Testing the optimal lag length requires some information that is provided through the Akaike Information Criterion (AIC), Scharwz Criterion (SC) and Hannan-Quinn Criterion (HQ). The results are asfollows:  Table 3 shows that the value of the SIC at the smallest lag is 1. Therefore, for the next test, using 1 as the Lag Optimum value, the selection criteria use traditional Scharwz Information Criterion (SIC).Obtaining the SC, if goes well, brings in the selection of optimal lag. This serves as the basic instructions of the next test.

3.4.Granger Causality Test
Granger causality test aims to determine the causal relationships between variables. The core of this test can indicate whether a variable has a two-way or one-way, or no connection. This test is seen as the influence of the past on the present condition (Wen, 2007)]. The results of Granger causality test are as follows: Testing in pairs (pairwise) in Table 4 was carried out by using the Granger Causality Test.The level of significance (α) 5% showed significant results in endogenous variable (SBL, SUPP, DMD, HHI, STY and BTD). The total test of six Endogenous variables yielded fifteen pairs of variables that influence each other. Six pairs of variables affect each other or have bilateral bidirectional causality (SBL with STY, SBL with BTD, SUPP with DMD, DMD with STY, DMD with BTD, and STY with SUPP). Four pairs of variables do not affect each other, Not having two-way connection or independence (SUPP with HHI, DMD with HHI,STY with HHI and SUPP with BTD). Meanwhile, the remaining five pairs of variables have opposite properties, namely one direction or directional interplay,. Therefore, a total of sixty percent of endogenous variables affect each other. (Marno Verbeek, 2008) argued that the existence of a cointegration relationship in a system of equations implies that there is an error correction model that illustrates the consistency of short-term dynamics with the long-term relationship. In other words, cointegration presented a long-term equilibrium relationship. Cointegration test in this study uses Johansen approach by comparing the trace statistic with the critical value used, which is 5%. If the trace statistic is greater than the critical value, it means that there is a cointegration in the equation. Cointegration test results are as follows: Based on the results of cointegration tests shown in Table 5, SB Latex pricing in Indonesia is determined by variable interplay, including production capacity (SUPP), SBL products by customer demand (DMD), the market share of the SBL (HHI) and the main raw material usage. Styrene Butadiene (STY and BTD) indicates no cointegration in the rank 3 to trace. With the cointegration, estimation results further using VECM models.

VECM Estimation Results
Cointegration test in the previous VECM system proves that there is a cointegration between variables involving SBL, SUPP, DMD, HHI, STY and BTD.Therefore, the SB Latex pricing responsiveness analysis of various microeconomic variables in this study is combined with the VECM models. The VECM models provide estimations of the two main outputs (Jones, Leiby, & Paik, 2004) that measure the cointegration or long-term equilibrium relationship with the short-term relationship, as well as error correction or velocity measuring these variables in moving towards the long-term

A C C E P T E D M A N U S C R I P T
equilibrium. In this study, the significance of a variable to another variable assessed the significance level of 5% and 10%. The VECM test results are as follows:  At the long-term relationship, SB Latex prices show a significant positive correlation with the raw material prices, which are the prices of styrene and butadiene.This is due to the reliance of SB Latex industry on both main raw materials.There are moreproducts with more use of styrene butadiene, (Styrene/ Butadiene approaching 1.5 to 2) that are seen with the coefficient of styrene than those seen with the coefficient of butadiene.Therefore, whenever the prices of raw material increase, it would also directly trigger the price increase of Latex. SB Latex industry structure in Indonesia is a quite competitive developing oligopoly. This industry started in 1991 and there was only one producer, namely PT BASF (monopoly). However,during 1995-2010 there had been three producers, namely PT BASF, PT Dow Chemical, and PT Latexia (oligopoly).In April 2010, PT BASF Indonesia acquired PT Latexia Indonesia (duopoly). Therefore, the market share (HHI) has a positive impact on the SB Latex price.If the market structure is controlled by only one or two companies, the company(s) is able to set the price.Higher index tends to increase the market share of SB Latex prices. Further, the error correction model for styrene butadiene latex is represented by Δ (SBL), which is able to describethe short-term relationship and longterm changes as follows:

Impulse Response Function
Impulse Response Function (IRF) is useful to show the waya variable respond to a shock about the variable itself and other endogenous variables. To conduct the IRF of Indonesian SB Latex prices in the VECM models, standard Cholesky decomposition is used. Cholesky decomposition aims to generate impulse response that depends crucially on the order (ordering) of variables in the system (Barbi, 2011). In this study, the analysis period of the SB Latex pricing as a result of theinternal variables of SB Latex industry or microeconomic variables is 36 months (three years) ahead. More results for IRF analysis can be seen as follows:

Forecast Error Variance Decomposition (FEVD)
The period of time used in projecting FEVD is 36 months (three years). The results are shown in Figure 2 below: Short-term observations (36 months) through the IRF and FEVD on the SB Latex prices show that the price variable of butadiene (BTD) is the most influential variable on SB latex pricing compared to other variables, such as SB Latex production capacity or supply, SB Latex demand, market segment index (HHI), and styrene prices. This is evident in accordance to the real occurrencein the SB Latex industry today and fifteen years back. This is due to the reliance on butadiene imports. In contrast, styrene products have been produced in Indonesia since 1995 through PT Styrindo Mono Indonesia. Therefore, almost all demands were met, mainly for styrene SB latex polystyrene industry and acrylonitrile styerene industry.In fact, the demands for butadiene had increased with an average growth of 10% from 2000-2010, even thoughthe butadiene has not been produced in Indonesia. This triggers the price to be more expensive than the price of butadiene styrene today(CIC, 2012)

4.DISCUSSIONS
Based on the findings of the current study, variables associated with the long-term and short-term effects on the long-term price of SB latex will have implications on the managerial SB latex as follows: For company management, the price of raw materials (styrene and butadiene) is the main variable affecting the total manufacturing cost, which in turn will affect the company profitability.As a result,the management seeks to reduce costs through the efficiency and policies of various divisions within the company, in this case SB Latex industry in Indonesia.
Butadiene production projects carried out by PT Chandra Asri Petrochemical Centre (CAPC) with a total investment of $145 million set off in January 2012,while production started in early 2014. This will lower the price of butadiene in Indonesia and reduce the dependence on imports, in addition to the reduction on SB Latex industrial material costs. Butadiene is used also in styrene butadiene rubber (SBR) industry and Acrylonitrile Butadiene Styrene (ABS) industry.
Competition SB Latex produce in Indonesia now only between PT BASF Indonesia with PT Dow Chemical since Ciba Speciality Chemical Corporation was acquisited by BASF The Chemical Company at 2008, so that PT Latexia Indonesia (Ciba Speciality Chemical) was eliminated to PT BASF Indonesia. Affecting to market share SB Latex in Indonesia which PT BASF Indonesia as market leader and HHI increasing almost 70 %. (IHS, 2018) The price of raw materials (Styrene and Butadiene) is the main variable affecting the total manufacturing cost that will ultimately affect the company profitable, with this phenomenon. Management seek to reduce cost through efficiency programs of the various divisions within SB Latex industry in Indonesia (PT BASF Indonesia and PT Dow Chemical).
Industry SB Latex is an intermediate of the derivative tree industry petrochemical industry, which generally requires a steady supply, the main raw material, namely Styrene and Butadiene, where the development of the main raw material prices have a tendency to rise from year to year due to higher crude oil prices , so that should be pursued and developed process technology to reduce the use of Styrene and Butadiene, but does not reduce the quality of SB Latex in industrial applications, namely Paper Coating and Carpet. BASF The Chemical Company April 2010 have applied dextrem process to the entire plant in the worldwide through methods add starch to the polymerization process in order to reduce the use of Butadiene, and this has an impact on the selling price of SB Latex in the market.

A C C E P T E D M A N U S C R I P T
Innovation technology in industrial SB latex process is a necessity that must be done in a sustainable manner through research and development, especially SB Latex products are shaped emulsion polymer (viscous white liquid), making it easier do modifications in process technology than for solid polymer (condensation polymer). A modified form is the first, namely the reduction of the use of reagent, the second is the reduction in the use of the main raw material, namely Styrene and Butadiene, the third is the substitution of other raw materials at cheaper prices, the fourth one is saving and substitution of energy for utilities, the fifth is the reduction of the polymerization reaction time (cycle time process). Cycle Time Process in the manufacturing SB Latex has developed gradually, starting in the early years of 1950-1970 manufacture of SB Latex takes nearly 24 hours, the period from 1971 to 1980 is shortened to 16 hours, the period from 1981 to 1990 was reduced to 12 hours, the period 1991 -2000 can be reduced again to eight hours, the period from 2001 to 2010 can be accelerated to 6 hours. Developing cycle time can not alone but always simultaneously accompanied by other process. Modification Process Technology above will have direct impact on pricing SB Latex, in order to obtain more competitive prices.
Previous study which The prices of Petrochemical Industry products in Indonesia cannot avoid the effects of globalization on the prices of Petrochemical Industry products in the global market. In more detail and in more detail petrochemical products are an inseparable part of the upstream industry, while the SB Latex Petrochemical Industry is a derivative industry from the upstream industry or often called an intermediate industry which is a supplier of industrial paper coating, carpet backing, paint, material construction and textiles. So if there is a change in SB Latex prices on the global market (SB Latex prices USA, Japan, China and Western Europe) then this will have an impact on SB Latex prices in Indonesia (Huang & Chen, n.d,2013.)

Conclusions and Suggestions. 5.1. Conclusions
Domestic factors that affect SB latex prices in Indonesia are the main raw material (styrene and butadiene) prices and the market share of demands for latex and SB latex in the paper coating industry and carpet industry in Indonesia.
The short-term aspects affecting the SB Latex pricing in Indonesia is the price of SB Latex itself and demands for SB Latex by the paper coating industry and carpet industry.Meanwhile, the long-term aspects that affect the SB Latex pricing in Indonesia are the prices of main raw (styrene and butadiene) materials and the market share of SB Latex industry in Indonesia (HHI).
In terms of industrial market share of SB Latex (HHI), the prices of styrene and butadiene could contribute significantly to the changes of SB Latex prices in the long run. This contrasts to the short-term relationship between the intense manufacturer competition and the SB Latex price.Tough competition between manufacturers of SB Latex nation wide results in the price declines, although demands for SB Latex increases.

Suggestions
This study has alimitation that it only focuses on one derivative petrochemical product, which is Styrene Butadiene Latex (SBL).Therefore,further studies need to be carried out focusing onother petrochemical derived products (downstream and intermediate).