Further Descriptive and Predictive Evidence on Alternative Time-Series Model for Quarterly Earnings

Several studies have provided evidence suggesting the usefulness of 3 alternative, parsimonious time-series models of the ARIMA class for quarterly earnings data. The models are: 1. Foster (F), 2. Griffin-Watts (GW), and 3. Brown-Rozeff (BR). The validity of the models for quarterly earnings, using...

Full description

Saved in:
Bibliographic Details
Published in:Journal of accounting research 1983-04, Vol.21 (1), p.317
Main Authors: Lorek, Kenneth S, Icerman, Joe D, Abdulkader, Abdullah A
Format: Article
Language:English
Subjects:
Earnings per share
Quarterly
Time series
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Several studies have provided evidence suggesting the usefulness of 3 alternative, parsimonious time-series models of the ARIMA class for quarterly earnings data. The models are: 1. Foster (F), 2. Griffin-Watts (GW), and 3. Brown-Rozeff (BR). The validity of the models for quarterly earnings, using statistical goodness of fit tests, is reassessed. Alternative structures are searched for after controlling for industry membership. The data consisted of 56 observations for model identification and 8 for forecast evaluation. The predictive ability of the models were obtained for one-, 4-, and 8-step-ahead forecasts to provide evidence of predictive ability across short, medium, and long horizons. Analysis indicates the F model is slightly inferior to the other models at the one-step-ahead level. However, there were no significant differences at the 4- and 8-step levels.
ISSN:0021-8456
1475-679X