ANALISIS TIME-SERIES PERILAKU LABA QUARTALAN PERUSAHAAN PUBLIK DI BURSA EFEK JAKARTA
DOI:
https://doi.org/10.34208/jba.v4i3.556Keywords:
ARIMA, Random walk model, Time-seriesAbstract
The main objective of this paper is to analyze the behavior of earnings quartely. This research is based on inconsistency of the result previous research. I derive two variables of earnings, origonal earnings and earnings deflated to total asset. I employ Box-Jankins methodology to identify ARIMA models of 32 firms over 1994:1 to 2001:4. The result shows that the earnings quarterly follow random walk for both the two variables of earnings. Parameters of ARIMA models that identified are relevant to explain the behavior of two variables of earnings quarterly. However, the parameters of random walk and identified models (ARIMA) are not different. The result also provides evidence that there is not difference forecasting ability between random walk and identified models for earnings original.The forecasting ability of ARIMA models, however, is more accurate than that random walk model for deflated earnings.