Buys – Ballot Estimates for time series decomposition

Abstract

An estimation procedure based on the Buys – Ballot (1847) table for time series decomposition is given in this paper. We give two alternative methods called the Chain Base Estimation and Fixed Base Estimation methods. Simulated examples are used to illustrate the methods, while comparing them with the least squares approach. U.S. quarterly beer production is re-analysed and the descriptive model obtained is shown to outperform the ARIMA model of Wei (1989) in terms of forecasts. KEY WORDS: Trend, Seasonality, Cycles, Decomposition, Periodicity, Buys-Ballot Estimate. Global Journal of Mathematical Sciences Vol.3(2) 2004: 83-98