DGPE PhD Course: State Space Models

11-13 October 2016. Lecturer: Siem Jan Koopman, VU University Amsterdam, Tinbergen Institute and CREATES

2016.09.13 | Solveig Nygaard Sørensen

Date Tue 11 Oct Thu 13 Oct
Time 09:00    17:00
Location Fuglesangs Allé 4, 8210 Aarhus V

The course is organised around the book “Time Series Analysis by State Space Methods” of Durbin and Koopman (Second Edition, 2012, OUP). A set of articles will complement the material for the course. Each day covers a range of topics that are grouped around three themes, they are:

  • Local level model, unobserved components models, statistical properties, reduced form ARMA representations. introducing the Kalman filter and signal extraction methods, linear Gaussian state space models.
  • General derivation of Kalman filter, missing observations, forecasting, maximum likelihood estimation, initialisation, multivariate extensions, dynamic factor models, collapsing methods, quasi-maximum likelihood methods.
  • Nonlinear and non-Gaussian models in economics and finance, introduction to simulation-based state space methods for estimation, signal extraction and forecasting, including importance sampling and particle filtering, observation-driven alternatives, score-driven time-varying parameter models.

The aim of the course is that students get a good overview of state space time series analysis together with a solid understanding of some key derivations of the main results and a hands-on training for the implementation of various methods on the computer. Data sets will be provided.


  • Registration via webshop (closed)

The course is free of charge for DGPE members from AU, KU, CBS, SDU, and Nordic universities. The course fee is EUR 200 for other participants.


Lecturer: Prof. Dr. Siem Jan Koopman: s.j.koopman@vu.nl

Teaching Assistant: Martin Thyrsgaard: thyrsgaard@econ.au.dk

DGPE Courses