Modeling time series to forecast households consumption of apple and orange in Iran

Document Type : Original Article

Abstract

Forecasting behavior of economic variables is an essential element of national planning, which is conducted frequently by time series techniques. However, the type of the model selected influences precision of forecasting results. Yet, there are many studies that have chosen an empirical model based on an ad hoc procedure rather than following a rigorous systematic approach. The main objective of this study is to present a systematic approach of specifying appropriate time series model and empirically compare the precision of forecast of the two procedures. To this end we model per capita consumption of apple and orange fruits over the period 1975-2006 and compare the MAPE of forecast errors of the models specified through the suggested approach and the ad hoc one. Results indicate that following such a systematic approach can considerably improve the precision of forecast. The correctly specified models were used to predict per capita consumption of apple and orange fruits for period 2007-2016 by the Iranian households.
JEL Classification: C22, C32, C51, C53, D12, Q11