Ersel, Derya Hacettepe Üniversitesi - Fen Fakültesi - İstatistik Bölümü, Turkey , Atılgan, Yasemin Kayhan Hacettepe Üniversitesi - Fen Fakültesi - İstatistik Bölümü, Turkey , Günay, Süleyman Hacettepe Üniversitesi - Fen Fakültesi - İstatistik Bölümü, Turkey

Bayesian Analysis of Stochastic Volatility Models and an Application

35944

Time series analysis is generally used to analyze financial assets. Recently, researchers have been studied on time series models with changing variance over time. Two well known models in this area are ARCH and GARCH models where variance is defined as a deterministic function of time. An alternative to ARCH/GARCH is SV model where variance is determined as a stochastic function of time. The SV model provides more flexible modelling of financial time series than ARCH/GARCH models. Since the structure of the likelihood function of SV model is very complicated, it is very hard to estimate the model parameters via the classical approaches. By using Bayesian analysis and MCMC techniques, this problem can be solved. In this study, Bayesian analysis of SV models will be explained and an application of this analysis to the financial time series data (Jan 1999/Apr 2009 monthly Euro/TL andDollar/TL exchange rates) will be exhibited.

62

Stochastic volatility , MCMC methods , Gibbs sampling , Bayesian analysis.

Süleyman Demirel University Faculty of Arts and Science Journal of Science

72