Using ARIMA Models to Forecasting of Economic Variables of Wheat Crop in Afghanistan

Abstract

Wheat is considered the main food crops in Afghanistan, whether to use it for majority of the population consumption or to use it in some industries and others.
 Problem: Afghanistan suffers from a large gap between production and consumption, so the current research investigates the problem arising from a shortage of wheat production to meet self-sufficiency of the population.
 Methods: The time series analysis can provide short-run forecast for sufficiently large amount of data on the concerned variables very precisely. In univariate time series analysis, the ARIMA models are flexible and widely used. The ARIMA model is the combination of three processes: (i) Autoregressive (AR) process, (ii) Differencing process and (iii) Moving-Average (MA) process. These processes are known in statistical literature as main univariate time series models and are commonly used in many applications. Where, Estimation of future wheat requirement is one of the essential tools that may help decision-makers to determine wheat needs and then developing plans that help reduce the gap between production and consumption. A solid strategy that widely applying of improved seeds and fertilizers, an effective research and extension system for better crop management is necessary to eliminate this gap for self-sufficiency in wheat production, besides providing the necessary financial sums for that. Where most prediction methods are valid for one-year prediction. However, moving prediction methods have been found to measure and predict the future movement of the dependent variable.
 Aims: The current research aims to prediction for Area, Productivity, Production, Consumption and Population over the period (2002-2017), to estimate the values of these variables in the period of (2018-2030).
 Results: The results showed that through the drawing of the historical data for Planted area, Productivity, Production, Consumption and Population of wheat crop it was evident that the series data is not static due to an increasing or a decreasing of general trend, which means the instability of the average, by using Auto-correlation function (ACF) and Partial Correlation Function to detect the stability of the time series, The results showed also, the significance of Autocorrelation coefficient and partial correlation coefficient values, which indicates that the time series is not static.