Topics:
- Overview
- ARMA processes
- Estimation
- Prediction
- Structural analysis
- Modeling nonstationary time series
- Modeling time-varying parameters
- GARCH
Target audience: Advanced students and PhD students in econometrics, statistics, VWL, BWL, mathematics or computer science.
Prerequisites: Profound knowledge in matrix-algebra and econometrics (econometrics I) or statistics (linear models). Basic knowledge in univariate time series analysis is not demanded but of advantage.
Record of achievement: This course consists of two parts. The first part of „Multivariate Time Series Analysis“ is equivalent to the lecture „Multivariate Zeitreihen/Multivariate Time Series“ (3 ECTS-Credits) for statisticians; the second part can be recognised as „Ausgewählte Gebiete der theoretischen Statistik B/Selected Topics in theoretic Statistics B“ (3 ECTS-Credits). For each part there will be a separate exam that takes one hour. That is, you only have to attend the first part (see below) of the course if you want to obtain credit points for the course “Multivariate Zeitreihen”. Apart from that, the whole course is equivalent to "Time-Series Econometrics" and counts as a class for Ph.D. candidates in economics. Both exams constitute the exam you have to pass in order to obtain the "Schein" for "Time-Series Econometrics".
Altogether, there are 5 problem sets in the tutorial. Problem sheets 1, 2 (theoretical properties of VAR(p) processes), and 4 (estimation of VAR(p) processes) belong to the first part „Multivariate Zeitreihen“. Problem sheets 3 (granger-causality, impulse-response-analysis, innovations accounting) and 5 (integrated and cointegrated processes) are relevant for the second part.
- Trainer/in: Fabian Spanhel