Forecasting The Amount Of Oil and Non-Oil and Gas Exports In Indonesia Using The Box-Jenkins Method

Abstract

Indonesia is one of the countries in Southeast Asia that has diversity and abundance of natural resources, both oil and gas and non-oil and gas. Export is an international trade activity that is profitable for economic growth in Indonesia. The purpose of this study is to determine the best results and models for oil and gas and non-oil and gas export data. The Box-Jenkins method (ARIMA) is a forecasting method that does not require or ignores independent variables. The stages used in the box-Jenkins method are identification of the model to see whether the model is stationary or not, for a model that is already stationary, a parameter estimation test will be carried out by looking at the significance value of the parameter, then a diagnostic check will be carried out using residual white noise and residual tests. normal distribution. After that, the initial prediction model is obtained and to get the best ARIMA model, it can be done by looking at the smallest MAPE value of each model. The best model obtained in this study to predict the amount of oil and gas and non-oil and gas exports in Indonesia uses the Box-Jenkins or ARIMA method, namely model (5,1,3) with the smallest MAPE of 8.142%. Forecasting results for the number of oil and gas and non-oil and gas exports in Indonesia in January - December 2023, namely: January 22194.9 million dollars, February 22795.6 million dollars, March 24796.0 million dollars, April 24106 million dollars, May 22281.6 million dollars, June 23186.6 million dollars, July 25184.3 million dollars, August 24325.8 million dollars, September 22456.3 million dollars, October 23565.6 million dollars, November 25602.2 million and December 24540.3 million dollars. Actual data or actual data for January – April 2023, namely January 22,323.8 million dollars, February 21,321.3 million dollars, March 23,416 million dollars, April 19,290.5 million dollars. It can be concluded that the results of the forecasting value of the box-Jenkins method are close to the actual value.