The forecasting of the exports and imports of paper and paper products of Turkey using Box-Jenkins method

10.31195/ejejfs.502397 Eurasian Journal of Forest Science 2019 7(1): 54-53 http://dergipark.gov.tr/ejejfs The forecasting of the exports and imports of paper and paper products in Turkey using Box-Jenkins method Nadir Ersen1*, İlker Akyüz2, Bahadır Çağrı Bayram3 , Ormancılık Bölümü, Artvin Meslek Yüksekokulu, Karadeniz Teknik Üniversitesi, 08000, Artvin, Türkiye. Orman Endüstri Mühendisliği Anabilim Dalı, Orman Fakültesi, Karadeniz Teknik Üniversitesi, 61100, Trabzon, Türkiye. 3 Orman Endüstri Mühendisliği Anabilim Dalı, Orman Fakültesi, Kastamonu Üniversitesi, 37100, Kastamonu, Türkiye. Corresponding author: *1 2 Abstract In this study, it was aimed to determine the most suitable time series models with Box-Jenkins method, which was the most widely used in prediction studies. Export and import values were predicted by 2020 with the most suitable models. The data used in this study were obtained from the Turkey Statistical Institute. Data were monthly data covering from January 2003 to December 2014. Sum of Squared Errors (SSE) and Mean Squared Error (MSE) criteria were taken into consideration when selecting the best Box-Jenkins models. Also, in order to test the success of forecasting of the models, Root mean Error Square (RMSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) were used. As a result of the analyzes, it was determined that the most suitable models for export and import data were ARIMA (2,1,0) (0,0,1)12 and ARIMA(3,1,2)(1,0,1)12. It was predicted that the rate of exports meeting imports in paper and paper products of Turkey will be approximately 0.86 in 2020. Key words: Paper and paper products, Box-Jenkins, Exports and Imports INTRODUCTION The paper sector is the industry branch that use wood as raw material and produce pulp, paper, paperboard and other cellulose-based products (Atalay 2012). The paper industry consists of two parts: "paper pulp production and bleaching technology" and "paper and paperboard manufacturing technology" (Gavcar et al. 1999). Paper and paper products sector, which providing significant rate input to chemical products and mining sectors, occupies an important place in terms of number of enterprises and production capacity in Turkey. According to 2014 data, there are 3114 enterprises and 62839 people are employed in this sector (Bayraktar 2014; TSI 2015a; Akyüz et al. 2017). However, the production capacity of paper enterprises in our country is quite low compared to European Union countries, their competitiveness is low in international trade, and they take the pulp used in paper production from outside (Gedik et al. 2005; Akyüz and Yıldırım 2006; Akyüz and Yıldırım 2009; Tutku et al. 2018). 54 Eurasian Journal of Forest Science The forecasting of the exports and imports of paper by Ersen et al. 2019 Turkey's production value of paper and paper products in 2008 was 8.182 million Turkish liras, while the production value of this sector in 2014 was realized as 23.990 million Turkish liras (Bayraktar 2014; TSI 2015b). While total pulp exports in the world were approximately $45.6 billion in 2014, paper and paperboard exports in the same year were approximately $174 billion. Turkey was 45th and 28th in the pulp and paper-paperboard exports, respectively. Considering the import figures, Turkey was ranked as the 18th country that imports most of pulp in the world according to 2014 data (TRADEMAP 2015). In this study, the optimal time series model was determined by Box-Jenkins method and the exports and imports values of paper and paper products in Turkey have been estimated by 2020 with the most suitable model. Literature review Co and Boosarawongse (2007) have tried to estimate the rice export values of Thailand using BoxJenkins, Holt-Winters and Artificial Neural Networks. In a study carried out by Emang et al. 2010, the Seasonal Autoregressive Integrated Moving Average model was used to determine the estimated demand for chipboard in Malaysia. Also, this model is compared with seasonal Holt-Winters and ARAR algorithms. They suggested that the Seasonal Autoregressive Integrated Moving Average (SARIMA) model is better than other methods. Tajdini et al. (2014) used double exponential smoothing, Holt-Winters exponential smoothing and Autoregressive Integrated Moving Average (ARIMA) models to estimate the consumption of wood based panels (chipboard, plywood, veneer) in Iran. The sale of plastic production using ARIMA method was estimated. For this project, the sales data of plastic factory production in Bandung was used. ARIMA (3,0,2) was found to be the best model for PP Trilene and PP Tintapro products (Siregar et al. 2017). In this study, it was aimed that is to forecast monthly Headline Consumer Price Index (HCPI) using the Box-Jenkins ARIMA methodology (Jackson et al. 2018). In a study carried out by Mishra et al. 2018, it was tried to determine estimated the area, production and yield of Sunn hemp in India using Autoregressive Integrated Moving Average (ARIMA) model. It was found that the most appropriate models for the area, production and yield of Sunn hemp were ARIMA(1,1,2), ARIMA(1,1,4) and ARIMA(1,1,5), respectively. MATERIAL AND METHOD In this study, the exports and imports data of the paper and paper products sector were examined. The monthly data covering the periods of January 2003-December 2014 were used to be examined in more detail by considering the trend and seasonal the components. The data were obtained from the Turkey Statistics Institution. The data was taken as $1000. Method Box-Jenkins Method The Box-Jenkins method was developed by George E.P Box and Gwilym M. Jenkins (1970) and is based on the principle of stinginess. This method, which is one of the single variable models, is called as ARIMA models. Autoregressive Integrated Moving Average (ARIMA) models are the most used method in time series analysis due to the simplicity and adaptability (Chen et al. 2014). These models are divided into two. These: non-seasonal and seasonal ARIMA models. The general expression of the non-seasonal ARIMA(p,d,q) model is as follows (Hyndman and Athanasopoulos 2017): 55 Eurasian Journal of Forest Science The forecasting of the exports and imports of paper by Ersen et al. 2019 𝑤𝑡 =𝑐 + ∅1 𝑤𝑡−1 + ∅2 𝑤𝑡−2 + ⋯ + ∅𝑝 𝑤𝑡−𝑝 + 𝜃1 𝜀𝑡−1 + 𝜃2 𝜀𝑡−2 … + 𝜃𝑞 𝜀𝑡−𝑞 + 𝜀𝑡 (1) where 𝑤𝑡 is the differentiated series, c is a constant, p is the order of autoregressive models, q is the order of moving average models, ∅1 , ∅2 , … ∅𝑝 are the autoregressive parameters, 𝜃1 , 𝜃2 , … . 𝜃𝑝 are the moving average parameters, 𝜀𝑡 is the error term at time t. The difference is realized by the wt = yt-yt-1-yt-2-…-yt-d formula; where d is the number of differences taken (Hyndman and Athanasopoulos 2017; Chatfield 2000). The seasonal ARIMA (SARIMA) models are similar to the ARIMA models. In general, Seasonal Autoregressive Integrated Moving Average (SARIMA) model is shown as SARIMA (p, d, q) (P, D, Q); where p is autoregressive (...truncated)