Optimal mean-volatility model for BET
Optimal mean-volatility model for BET
Author(s): Kinga SiposSubject(s): Economy
Published by: Alma Mater & Universitatea »Babes Bolyai« Cluj - Facultatea de St. Economice si Gestiunea Afacerilor
Keywords: Bucharest Exchange Trade Index; ARMA&GARCH meanvolatility model; model selection; model validation
Summary/Abstract: The values of the Bucharest Exchange Trading index recorded in the period 10.09.1997-28.09.2012 constitute the subject of this study. Our aim was to identify a model able to explain the mean and volatility behaviour of the log-return series. The series shows some first order autocorrelation and carries the signs of the specific GARCH effects (high second order autocorrelations, volatility clustering and fat-tailed distribution). Using the Akaike, respectively Bayesian information criteria and the likelihood ratio test, we find that the most parsimonious model fitting our scope is a combined ARMA(5,4) & GARCH(2,1) model, which is validated by the results of the postestimation analysis, regarding the lack of second order autocorrelations in case of the standardized innovations, the stationarity of the model and its applicability for forecasts.
Journal: Review of Economic Studies and Research Virgil Madgearu
- Issue Year: V/2012
- Issue No: 2
- Page Range: 183-205
- Page Count: 23
- Language: English
- Content File-PDF
