Time Series Analysis Of GDP and Market Indices
The paper is concerned with time series analysis of GDP growth and returns of securities market indices. The main goal was to identify cyclical patterns in the examined series and to demonstrate correlation among the individual series. First part of the paper presents deals with fitting the selected models to real data. Chapter 3 is devoted to correlation analysis. The analysed time series are quarterly GDP growth in the USA, in Germany and in the Czech Republic and monthly and quarterly returns of S&P500, DAX and PX50. All models applied in Chapter 2 are based on autoregressive processes. The analysis starts with linear ARMA model, followed by non-linear extensions thereof: the ARCH family based on conditional heteroscedasticity and the Threshold AutoRegressive Model (TAR). Non-linear models appeared to be more appropriate than simple ARMA models. Traces of cyclical patterns were identified in several time series, however, only US GDP growth and S&P500 seemed to be closely related in terms of cyclical behaviour. The correlation analysis confirmed two frequent hypotheses. First, returns on different capital markets in the world are strongly correlated reducing thus investors' possibilities to diversify risk. Second, the development of the Czech GDP growth and of PX50 responds (with certain delay) to the development of German GDP growth, whereas no correlation can be found between the Czech GDP growth and PX50 returns.
Volume (Year): 10 (2003)
Issue (Month): 18 ()
|Contact details of provider:|| Web page: http://ces.utia.cas.cz|
More information through EDIRC
When requesting a correction, please mention this item's handle: RePEc:czx:journl:v:10:y:2003:i:18:id:119. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Jozef Barunik)
If references are entirely missing, you can add them using this form.