Διερεύνηση των χαρακτηριστικών των χρονολογικών σειρών αποδόσεων των μετοχών

View/ Open
Subject
Ανάλυση χρονολογικών σειρών ; Μετοχές ; Στατιστική -- Οικονομετρικές μέθοδοι ; Δείκτες μετοχών ; Stocks ; Time-series analysis ; Statistics -- Econometric modelsAbstract
Time series data are increasingly used by researchers, in their attempt to study some of the existing relationships among the parameters of the economic and especially the financial environment. As a consequence, the identification of their properties has attracted the interest of the academic community, thus resulting to the enrichment of the relevant literature. Within this framework, the present study, aiming to identify the properties of the time series financial data of the developing countries. Specifically, it has analyzed the stock index prices of four Balkan countries, namely Slovenia, Croatia, Bulgaria and Romania, for the period 03/01/1994-29/05/2009. Through the application of modern statistical methodologies, we finally concluded that our data arc characterized by non-linear dependency, lack of normality and heteroscedasticity. To deal with these distortions, several ARCH/GARCH models were employed. Finally, the GARCH (1, 2) model was shown to be the best to describe the evolution of the conditional variances of the stock returns of the countries under examination.