search query: @journal_id 1112 / total: 6
reference: 4 / 6
Author: | Crato, N. |
Title: | Forecasting Business and Economic Time Series with Overdifferenced Models |
Journal: | Estudos de Gestao
1993 : -94 WINT., 1:2, P.77-82 |
Index terms: | FORECASTING TECHNIQUES |
Language: | eng |
Abstract: | Box-Jenkins and ARIMA models have been widely used tools for modelling and forecasting in economic and business time series. A critical point with these models remains in the possibility of overdifferencing the data in order to achieve stationarity. Contrarily to popular belief, overdifferencing can have serious adverse forecasting consequences. A spe- ctral estimation method is proposed as a decision making tool for estimating the degree of differencing. An applica- tion with financial data is provided. |
SCIMA