Research Article Special Issue

Transcription

1 Research Article Special Issue DESIGNING PRICE STABILITY MODEL OF RED CAYENNE PEPPER PRICE IN WONOGIRI DISTRICT, CENTRE JAVA USING ARCH/GARCH METHOD F. Dianawati, R. W. Purnomo Department of Industrial Engineering, Universitas Indonesia Depok, West Java, Indonesia Published online: 16 April 2018 ABSTRACT Food and agricultural sector become the biggest sector contributing to inflation in Indonesia. Especially in Wonogiri district, red cayenne pepper was the biggest sector contributing to inflation on A national statistic proved that in recent five years red cayenne pepper has the highest average level of fluctuation among all commodities. Some factors, like supply chain, price disparity, production quantity, crop failure, and oil price become the possible factor causes high volatility level in red cayenne pepper price. Therefore, this research tries to find the key factor causing fluctuation on red cayenne pepper by using ARCH/GARCH method. The method could accommodate the present of heteroscedasticity in time series data. In the end of the research, it is statistically found that the second level of supply chain becomes the biggest part contributing to inflation with 3,35 of coefficient in fluctuation forecasting model of red cayenne pepper price. This model could become a reference to the government to determine the appropriate policy in maintaining the price stability of red cayenne pepper. Keywords: ARCH/GARCH, Forecasting, Red Cayenne Pepper, Volatility Author Correspondence, author@gmail.com doi: Journal of Fundamental and Applied Sciences is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. Libraries Resource Directory. We are listed under Research Associations category.

2 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), INTRODUCTION Price fluctuation problem, especially in food sector, occur almost every year in Indonesia. The grow of fluctuation measured by volatility level, a standard to measure the rate of price change over a period to the next period. Therefore, higher price change means higher volatility level. Volatility is one of factors causing higher inflation in a country or district [1]. Related to price change as the basis of volatility, commodity price is very flexible since they have high competitive auction market. Commodity price is quickly change as respond to economy-wide shocks to demand. Therefore, commodity price become the strong indicator of future inflation [2]. In 2016, inflation of foodstuff, as food commodity, is quite high, that is 5,7%. This condition occurs because foodstuff has high volatility level. Volatility in food commodity, especially on strategic food commodity, become such crucial problem in every country. In Indonesia, these strategic foods include cayenne pepper, chicken, onion, garlic, and beef [3]. Recent five years, these types of commodity always include in 10 lists. of food with highest volatility level. In Wonogiri, a district of Central Java, red cayenne pepper become commodity with the highest volatility level and having highest contribution on inflation. Its contribution reached 0,35% of 2,08% inflation in Wonogiri in 2016[4]. Price change in red cayenne pepper occur almost in every market. In change in red cayenne pepper occur almost every in market. In high price period, the price increased up to198% from price in low price period. In the middle 2016, price range are under Rp ,00($1.48) and increased up to Rp ,00($5.93) in the end of the year as showed in Fig. 1. [5]. Many factors could bring the high volatility issues in food and agriculture commodity. Previous research, as did by Francesco, Pierangelo, and Manuel [6], explain that energychanges has an impact in agriculture commodity movements. Meanwhile, stock price uncertainty also affected commodity price movements in Another research, as did by J. Lübbers, P.N. Posch [7], show that commodity price movement is related to the main factors, these are inventory level, hedging pressure, and demand and supply level. The similar finding, as did by Satoto [1] in his research related to fluctuation in onion, show that fluctuation occur caused by inventory movement, meaning inventory level in every district. Supply chain also have deep relationship to this problem of volatility since all level contribute on price making. Therefore, deep analysis of activity on each level of supply chain should be done to count its contribution to fluctuation along the year. In general, supply chain level of red cayenne pepper in Wonogiri district are showed by Fig. 2.

3 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), Fig.1. Price of Red Cayenne Pepper on 2016 in Wonogiri District As many factors found related to commodity price movement, it is necessary to do further research to find the key. Based on this issue, further research to design price stability model in Wonogiri did by using secondary data during ARCH/GARCH method chosen because its advantage to accommodate the present of veryhigh variation in price, which usually called heteroscedasticity. This research began by finding indicated factors related to fluctuation, designing the stability model, and ended by giving alternativee solution based on model. The result of this research can be used by government as consideration in creating policy to stabilize red cayennee pepper price along the years. It is not a fixed point to choose only one action, such as choose to solve the strongest factor. Action could be done in some ways according to condition occur on the field along the years, these because the ending result is a mathematical model to stabilize price change of red cayenne pepper price in Wonogiri. It is necessary to be noted that stabilization here is not directly for price. Red cayenne pepper price could be stabile (low volatility) only if the price change from a period to the next period is very low. It is mean that price difference in a period to another is not significant along the year. In order to find the correct step in creating the ARCH/GARCH model, it is necessary to do literature review related to time series data. As literature review, stationarity and autocorrelation testing are very important to ensure the time series data could be used further, as forecasting. Hence, heteroscedasticity testing is further step as the basis of using ARCH/GARCH method in the research to creat forecasting model. Each step is determined as follow. Stationarity is the nature of data characterized by constant average and variance over time and also the covariance value between two times periods depends only on the distance or lag difference between the two times periods, not the actual time at which itscovariance being calculated [9]. In mathematical equation, data Y is stationer during k period only if:

4 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), Fig.2. Supply Chain of Red Cayenne Pepper on 2016 in Wonogiri District Average: E(Yt) = μ (1) Variance: var(yt) = E(Y t - μ) 2 = σ 2 (2) Covariance: γk = E[(Y t - μ) (Yt+1 - μ)] (3) Dependency on a set data is a sign that the data has autocorrelation. Therefore, the value covariance is no longer 0. Covariance data containing autocorrelation shows in equation 4 as follow. Cov (ui, uj xi, xj) = E(ui, uj) 0 i j (4) A set time series data has autocorrelation only if there is dependency over period of time, it is mean data of time t depend on data of time t1. In example, for regression of two variables in equation (5). Yt = β1 + β2xt + ut (5) Only if equation (5) satisfies equation (4), it said time series data has error factor called white noise. The error factor may arise as present of autocorrelation and pure random error factor as illustrated in equation (6). Heteroscedasticity testing is further step as the basis of using ARCH/GARCH method in the research to creat forecasting model. Heteroscedasticity, as an unexpected condition in designing a model, cause the decision to become invalid due to an error in the model. In simple regression model in equation (5), valid standard error could be calculated by formula in equation (7). However, if heteroscedasticity occurs then a valid equation will change to equation (8) [9]. Var (β1) = σ ( ) (7) Var (β1) = ( ) σ ( ) (8)

5 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), When the heteroscedasticity appears, according to [8], the best method to be used is ARCH / GARCH. Prior to achieving ARCH / GARCH modeling, next step is creat forecasting model following Box-Jenkins procedure to determine the period of the data deviation and then analyzed whether the time series data follows the autoregression, moving average, or both of them. The ARCH / GARCH model equations are given in equations (9) and (10). GARCH is a development of ARCH, which takes into account only the residual periods of the previous period. The variance in ARCH is as follow. σ = αo + α u (9) On GARCH, which takes into account the residual of the previous period, the variance equation will be as follow. σ = αo + α u + λ 1 σ (10) III. METODOLOGI As determined at the end of the introduction, the data used are time series data from 2012 to Data gathered are taken from government report (DinasKoperasi, Usaha Kecil- Menengah, danperdagangan(department of trade); DinasPertanian(aglicultural department)) and direct interview to farmer, seller, and other supply chain component. These data are price data at consumer level per month (rupiah per kilogram), production (rupiah per month), crop failure (rupiah per month), farmer's price per month (rupiah per kilogram), value of activities of the second level of supply chain (rupiah per kilogram), value of activities of the second level of supply chain (rupiah per kilogram), seller activity (rupiah per kilogram), and oil price (diesel, rupiah per liter). Data calculation is done to determine the difference in price/return of each dependent and independent variables. Number of return are used to show price change of all factor along the years. The result of calculation (return) is used as input data in softwareeviews 9. This software is also used to creat model of price stability by ARCH / GARCH method. In accordance with the literature, the first step to creat the model is stationary testing. We used Augmented Dickey Fuller method in stationary testing, which the data would be stationer if the absolute t-statistic exceeded the absolute of Mac Kinnon critical number. The result of stationary testing can be seen in Table 1. The result show that t-statistic of all variables exceed the Mac Kinnon critical number at level (without differentiation).

6 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), Table 1. Result of Stationarity Testing Mac Kinnon Variable t-statistic Critical Number p-value (1%) Return of pepper price -7,396,949-3,550,396 0,0000 Return of production -1,336,209-3,574,446 0,0000 Return of crop failure -4,562,915-3,574,446 0,0006 Return of price in farmer level -6,508,964-3,555,023 0,0000 Return of activity in first level of supply chain -7,712,633-3,550,396 0,0000 Return of activity in second level of supply chain -8,027,046-3,550,396 0,0000 Return of activity of seller -1,164,865-3,548,208 0,0000 Return of oil price -7,838,057-3,548,208 0,0000 Fig.3. Correlogram of Model Testing After that, autocorrelation and heteroscedasticity tests were performed. The result of both testing showed that the data contain autocorrelation and heteroscedasticity problems. Therefore, the ARCH / GARCH forecasting method is required. Previously, it is necessary to determine the best ARMA model (autoregressive moving average) using Box-Jenkins

7 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), procedure. By correlogram showed in Fig. 3., it is known that irregular data occur in the second lag so that the model candidates are ARMA (2.0), ARMA (0,2), and ARMA (2,2). ARMA is illustrated by value of PACF and ACF in correlogram which more than Barttlet score (±0.255) for 59 observations. Then, each candidate is tested in order to choose the best model with highest adjusted R-square and lowest AIC and SC. AIC and SC is representation of error. The result of the testing in Table 2 show that ARMA (0.2) is the best model. Therefore, ARMA(0,2) used to create the GARCH (1,1) equation which results as show in Table 3. The result of creating model in Table 3, with a degree of freedom of 5%, show that almost all of independent variables are significant to the dependent variable. The third level of supply chain is not significant to the dependent variable due to its p-value that exceeded degree of freedom of 5%. This model has a trust rating, expressed as 76% of adjusted R-square. This model does not contain heteroscedasticity, proved by p-value exceeded 0.05 in testing with ARCH-LM. It can be concluded that the model has been free from the problem of heteroscedasticity. Table 2. The Result of Candidate Model Testing Model Adjust ed R- square AIC SC ARMA(2,0) 0, , 19, ARMA(0,2) 0, ,6 19,03

8 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), Table 3. Result of GARCH(1,1) Variables Coefficient Prob. Pepper price Production Crop failure Price in farmer level Activity in second level of supply chain Activity in third level of supply chain Activity of seller Adjusted R-square 0, Durbin Watson stat Prob. Chi Square ARCH LM IV RESULT AND DISCUSSION The model of price stability in Table 2, show that almost all independent variables are significant to the dependent variable. It can be seen from the probability value to accept Ho (not significant) is very small, under the degree of freedom (ɑ) The third level of supply chain is the only independent variable that is not significant to the result with p-value of 0.35, exceeding ɑ This means that the value change of the third level of supply chain activity will not have a significant effect on the red cayenne pepper price movements. It should be underlined that this equation is the equation of movements in commodity prices. This means that the stability will be achieved if only the price change of red cayenne pepper (Y) is very small or in a certain range expected. In example, if an assumption states that the price will be considered stable if the movement is no more than Rp 2, then in the price movements equation (Y), the result of the sum of variables with their coefficients, is in the range (-Rp 2.000,00) to Rp 2.000,00. Based on the model, the same change in allindependent variables will cause price changes according to their respective coefficients. In example, if there is a production increase of 1% then there will be a decrease in volatility of cayenne price by 0,02% [11]. The model resulted show configuration of movement making which created by supply chain component and effected by supply and demand movement along the year.

9 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), The configuration is flexible to use in many condition and could be the basis of each action should be done to take price movement in range. In equation, production is significant to the price of red pepper price movements evidenced by p-value of below ɑ The production coefficient on the model is (-0,00015), it indicates that negative changes in production will cause a positive price change. A decrease of 1% in production will lead to a rise in price of cayenne pepper by 0.015%. At present, the production of red cayenne pepper in Wonogiri district has not been able to fulfill 100% of the society's needs (assuming 80% of residents consume 0.25 ounces of red cayenne pepper per week). These conditions cause the whole production of cayenne pepper will be sold to fulfil the market needs. By the same level of supply quantity throughout the year, production reduction will lead price to increase. This is in line with the statement described by the equilibrium curve Related to production, there is a crop failure which is significant to red cayenne pepper price movements. Crop failure on the model has a coefficient of , which means an increase in crop failure causes the increase of red cayenne pepper price. Any increase in crop failure by 1%, will contribute to the increase of volatility rate of cayenne pepper price by 0.01%. Quoted price parallel to crop failure because crop failure is considered as a loss, as well as reducing production. Loss of production caused by crop failure also gives disadvantages to farmer since they have given same effort, including fertilizer and employee, for the loss. Therefore, to back up all of disadvantages, they will charge the crop failure on the sale of cayenne. Thus, as the increase of crop failure, the price at the farmer level will also increase. In the end, the final price offered to consumers also increased. The next component of supply chain is collector as the second level component after farmer. They buy cayenne from farmer and take the cayenne directly from the field. In creating new price, they also considering fare of distribution as a part of activities. In distribution chain (supply chain), the changes made by this second level of supply chain have the highest impact on the volatility of the red cayenne pepper price. It has 3,34 of coefficient, which is the highest coefficient of the model. This means that every 1% change in second level of supply chain activity will impact an increase of volatility level of cayenne pepper price by 3 times. This happen in the real market because the change made by the second level of supply chain will affect to activities did by next level of supply chain as a whole. The second level supplying the third level in Wonogiri did not come from only one region. This condition causes the amount of activities is also different. The impact of this difference does not directly affect other second level of supply chain in rising prices. However, the impact will bring

10 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), impact at the third level of supply chain. If some third levels are supplied by different second level with different purchase price, the third level will tend to follow the highest or lowest price in order to stay profitable and win the market. So, one time the third levels followed the lowest price, they can keep the price stable. Therefore, it doesn t have significant effect in red cayenne pepper price movement. After the third level, the cayenne pepper moves to the end level before end consumer. This level is seller.in the model, the seller s activity has a coefficient of 0.55 to the change of the price of red pepper. A 1% increase in the value of seller activity could lead the increase in volatility of cayenne pepper prices by approximately 55%. The volatility of prices at this level occurs because pricing is sated through an agreement among sellers and they take the highest price. Seller are the last component who directly deal to the end-consumer. In addition, not only influenced by the difference in the base price in each seller, the deal of selling price determination also influenced by other condition like price changes of other goods in the market. There is a condition in which the goods on the market have increased simultaneously at a time when there is a change of an important component. This kind of condition is called market shock. One component brings the market shock is oil price movements. Related to oil price, oil price movements have significant effect in price commodity movements. Oil change in positive range could affect commodity price change in positive range too. This happen because the change in oil price not only increase distribution cost but also gives shock to the market. Previous research done by Bank of Indonesia said that shock market occurs because the owners of goods increase the price not only for distribution but also as speculation if in the future the purchase price or their costs start to increase. Although the second level of supply chain is variable with the highest influence, but the solution could not be determined only from that the component. Solutions could be very flexible according to condition on that period. Remember, the model is configuration of movement price. Therefore, as the basis configuration, it is possible to change the variable number since the fixed number of this configuration (mathematical model) is the coefficient number only. As an example, in the low season of production and high crop failure, the best solution could be reached not only by controlling first level of supply chain activity, but also by increasing production or decreasing crop failure. Thus, because all six variables give effect with different level according to coefficient in the model. Focusing on the second level of supply chain which is the main variables in the model, control of its activity become the ultimate priority.

11 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), Previously, pricing is determined purely by market without government interference. Control needs to be done so that the price offered by all the first level of supply chains to the second has the same amount. In this case, government needs to control the price directly as a way for the government to participate in pricing. This alternative can be achieved by adding institutions between first level of supply chains and the second. The function of this institution is to accommodate all the cayenne pepper offered by the first level of supply chain and then sold at the same price to all middlemen. In the end, all seller will have the same purchase price. If there is no change of condition in the market, the price will be stable all the time. Hence, the government can do control and maintain directly on the price and supply of red cayenne pepper in Wonogiri regency. CONCLUSSION The factors causing fluctuation come from supply and distribution through the supply chain components according to GARCH (1,1) model for the stability of red cayenne pepper price in Wonogiri. The best alternative solution can be applied by considering the conditions occur at that period. However, control at the second level of supply chain is necessary. It can also be used as an effort for local governments to participate in determining and controlling the price of red cayenne pepper in Wonogiri. Further research, like an assessment of the causes of each independent variable and their interrelationships with others will greatly enrich the workable solution. Acknowledgment The author would like to thank the financial support provided by Universitas Indonesia through the PITTA 2017 funding scheme under Grant no. 857/UN2.R3.1/HKP.05.00/2017 managed by the Directorate for Research and Public Services (DRPM) Universits Indonesia REFERENCES 1. HaptoSatoto: Analisisfaktor- faktor yang mempengaruhi fluktuasiharga bawangmerah danperama lannya (studikasus Pasar Induk Kramat Jati DKI Jakarta, 2007) 2. Furlong, Fred; Ingenito, Robert, Commodity prices and inflation Economic Review - Federal Reserve Bank of San Francisco; San Francisco, Peraturan Presiden Republik Indonesia Nomor 71 Tahun 2015 Tentang Penetapandan Penyimpanan Barang Kebutuhan Pokokdan Barang Penting BPS, DinasKoperasi, Usaha Kecil Menengah, danperdagangankabupatenwonogiri, 2016

12 F. Dianawati et al. J Fundam Appl Sci. 2018, 10(5S), F. De. Nicola, P. De. Pace, and M.A. Hernandez: Co-movement of major energy, agricultural, and food commodity price returns: A time-series assessment. Vol , p Lübbers, J., &Posch, P. N. (2016). Commodities common factor_ An empirical assessment of the markets drivers. Journal of Commodity Markets, 4(1), Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, 31(3), D. Gujarati, D. Powter Dasar-dasarEkonometrika 5 th Edition. Jakarta: SalembaEmpat 9. Hegerty, S. W. (2016). Commodity-price volatility and macroeconomic spillovers: Evidence from nine emerging markets. North American Journal of Economics and Finance, 35, AriniHarjanto How to cite this article: Dianawati F, Purnomo R W. Designing price stability model of red cayenne pepper price in wonogiri district, centre java using arch/garch method. J. Fundam. Appl. Sci., 2018, 10(5S),