Η κυματοειδής μορφή της μεταβληκότητας των μηνιαίων αποδόσεων των χρηματιστηριακών μετοχών και η χρονική ολοκλήρωση προσομοιωμένων δεδομένων
Financial markets display pronounced volatility clustering. Over the last decade financial economists have began to seriously model the temporal dependencies in return volatilities. This paper investigates the volatility clustering using econometric models based on the methodology of temporal aggregation for GARCH processes. We initially derive low frequency volatility models first from low frequency data and then from high frequency data using a Monte Carlo simulation. We compare the different models and finally we try to find out whose model’s parameters are more efficient. The one’s that were produced directly from low frequency data or the other’s that was produced through temporal aggregation from high frequency data? The paper is divided in three parts. In the first part we summarize the theoretical background on which this empirical survey is based. In the second part we make use of real data and more precisely, data of the Dow Jones Industrial 65 composite price index and the Standard and Poor΄s 500 composite price index. In the third part we produce, through a Monte Carlo simulation, low frequency volatility models from high frequency data using the methodology of temporal aggregation. The initial values for our simulation are taken from the analysis of the real data and our effort is to replicate the behaviour of Dow Jones Index’s volatility.