Due to its great usefulness, time series analysis is regarded as one of the most important fields in statistics. This Master's thesis focuses on multivariate time series and the vector time series models they follow. A more detailed examination of vector autoregressive moving average models is provided, including methods for their identification and use in forecasting. Furthermore, the concept of cointegration is introduced. In this context, the vector error correction model is presented, along with various testing procedures and approaches to estimating cointegrating vectors. The thesis also includes empirical modeling of multivariate time series using the R programming language. Finally, a comparative analysis of forecasts from a model that accounts for cointegration and a model that does not is conducted.
|
