This course addresses modelling techniques for time-series data when unit roots are present in the data. An overview of the technical characteristics of time-series data and the concept of non-stationarity is provided; and the econometric techniques of co-integration and error correction modelling are revised in single equations (residual-based co-integration), with emphasis on their empirical application. The main focus of the course is however on the theory and application of multivariate co-integration. The course concludes with a discussion on the application of volatility models.
The course takes place in a computer lab on the main campus of the University of Pretoria. Delegates use Eviews version 10 for practical applications.
After successful completion of this programme, delegates will be able to:
- Understand and apply non-stationary time series analysis
- Understand the concept of stationarity and unit root Testing
- Apply the advanced econometric techniques of cointegration and error correction modelling, specifically in the multivariate context,
- Understand and apply basic volatility modelling.
Delegates complete an open-book evaluation on the last day of the course. A certificate will be awarded upon successful completion of the course.
The course covers the following topics:
1. Overview of residual-based co-integration
- Data generating processes
- Stationary vs. non-stationary time series
- Co-integration in single equations (Engle-Granger)
- Error correction models (ECM)
2. Multivariate co-integration (focus of course)
- Vector autoregressive (VAR) models
- Impulse response functions and variance decompositions
- Johansen co-integration methodology (maximum likelihood estimation)
- Vector error correction models (VECM)
- Block causality and exogeneity tests
- Weak exogeneity tests and model identification
3. Introduction to volatility models
- Properties and theoretical and empirical issues
- ARCH and GARCH models
- Estimation and prediction
Prospective delegates should at least have a relevant honours degree with a focus on time-series econometrics, including a knowledge of the concepts of unit root testing and residualbased (Engle-Granger) cointegration. An understanding of matrix algebra is essential as well as experience as a researcher or analyst in any of the fields of economic application.Proficiency in EViews software is recommended.
|Catalogue and Category:|
|Financial Management and Taxation|
|Who Should attend:|
|Experience as a researcher or analyst in any of the following fields of economic application is required: financial markets, socio- economics and health, development economics, publicfinance and tax policy or international trade and finance.|