Yumuşak geçişli genelleştirilmiş kendiyle bağlaşımlı koşullu farklı yayılım modelinde sözde ençok olabilirlik tahmincisinin yanaşık özellikleri

This thesis examines asymptotic properties of the quasi maximum likelihood (QML) estimator in a specific nonlinear generalized autoregressive conditionally heteroskedastic (GARCH) process. The conditional mean is set to zero. The nonlinearity is established via smooth transition mechanism in the conditional variance, where the distribution function of a logistic distribution is used for the smooth transition function. Strong consistency and asymptotic normality of the QML estimator is proved in this smooth transition GARCH(1,1) model. For most of the analysis, we follow the work done in Meitz and Saikkonen (2008c), where asymptotic properties of the QML estimator are studied in nonlinear AR-GARCH models.