Forecasting: Does the Box-Jenkins Method Work Better than Regression?

Abstract

A difficult exercise, forecasting is key to effective management. Models using time series data are frequently used for forecasting likely values of important variables such as supply and demand.The regression method is the most common, although it involves many critical assumptions that are difficult to satisfy in practice. Efforts to reduce the severity of the assumptions and improve our ability to manipulate data have led to generalized regression and the Box-Jenkins Autoregressive Integrated Moving Average (ARIMA) models.Nanda tests the forecasts from two models in each method, using data on monthly milk procurement by Amul Dairy from January 1965 to December 1975. The regression method produced better forecasts than the Box-Jenkins method.