Course Objective

To gain insights in the time series analysis, modelling and prediction
based on state space models with a focus on theory, methods and
computations. To obtain experience in empirical modelling and
computational implementation, various computer programs need to be
written.

Course Content

This course focuses on theory, methodology and computing aspects of time
series analysis, modelling, and prediction based on a general class of
state space models. First, the econometric methodology is explored under
linear Gaussian assumptions. Second, departures from these assumptions
are considered. In particular, we study dynamic models with unobserved
components, signal extraction of dynamic latent variables, parameter
estimation via maximum likelihood, and forecasting. We derive Kalman
filter methods, their nonlinear extensions such as particle filter
methods, and related Monte Carlo simulation methods. All derivations
rely on basic principles in statistics and econometrics. The models and
methods are illustrated for the analysis of macroeconomic, financial and
marketing time series data.

Teaching Methods

Main lectures (4 hours); tutorials (2 hours) and computer lab (2 hours)

Method of Assessment

Written exam (0.8), homework and written assignments (0.2).

Recommended background knowledge

Statistics, Econometrics, Programming, Introduction to Time Series