Special Topic in Statistics - Advanced Time Series Methods

Description

In many applications, in particular in finance and economics, observed data series often exhibit a behaviour which cannot be modelled with linear time series models (i.e. ARMA processes). Thus alternative models allowing for a nonlinear behaviour are called for and are successfully used. For instance, Robert Engle was awarded the Novel Prize in 2003 for introducing the so-called (G)ARCH model.

In this course we will first review some materials on linear time series models, such as GARCH, Markov-switching as well as threshold autoregressive time series models. We study their common probabilistic and statistical concepts and theory (Markov chains with uncountable state space, stochastic recurrence equations, ergodicity and mixing). Finally, we will derive and apply estimators for the model parameters.


For a full list of Honours courses, please refer to the School of Mathematics and Statistics Honours Booklet Mathematics and Statistics Honours Booklet