Description
The goal of the course is to introduce non-linear time series models. Particularly, we examine mixture models (TAR, STAR, and MAR), conditional heteroscedasticity models (ARCH, GARCH), and their combinations. These models are widely used in empirical macroeconometrics and especially in empirical finance, because they are able to incorporate typical features encountered there. These features include asymmetry, aberrant observations, and plausible existence of regimes within which returns and volatility display different dynamic behaviour.
We also discuss model diagnostics and the problems encountered when testing statistical hypotheses in non-linear models.
Level
Master´s Degree – Econometrics, Y1 Optional courses
Scope
6 ECTS
Prerequisites
Basic knowledge on statistical properties of linear time series models (ARMA models) and on statistical inference.
Lectures (24 hours)
Fridays at 9-12 (Economicum, seminar room 1)
Exams
Final exam, 13th of May 2013 at 10 – 12, Economicum, seminar room 3-4
[2nd exam, 31st of May 2013 at 10 – 12, Economicum, seminar room 1]
Textbook
Non-linear Time Series Models in Empirical Finance by P.H. Franses and D. van Dijk, Cambridge University Press.
My recommendation is that you buy or borrow this book before the course begins.