This thesis describes a procedure based on Principal Component Analysis in conjunction with Kalman _ltering and smoothing which is able to give consistent estimates of the parameters for dynamic factor models. It is currently in use among others by central banks around the world to estimate economic indicators before they get published and has the main advantage of being able to handle publication lags of the input variables as well as mixed frequency data. The procedure was _rst applied by Giannone et al. (2008) on the GDP of the United States. The consistency of the method was proven by Giannone et al. (2011). This master thesis gives a detailed proof of consistency for the procedure based on the proof of Giannone et al. (2011) including the background needed to understand how the procedure works. We will outline the derivation of principal components and give a detailed derivation of the Kalman _lter and smoother, and show a computationally e_cient method for Kalman smoothing. Finally, an empirical analysis is performed where we test and verify the procedure and its advantages, using mixed frequency data to predict the GDP growth of Germany.     «

This thesis describes a procedure based on Principal Component Analysis in conjunction with Kalman _ltering and smoothing which is able to give consistent estimates of the parameters for dynamic factor models. It is currently in use among others by central banks around the world to estimate economic indicators before they get published and has the main advantage of being able to handle publication lags of the input variables as well as mixed frequency data. The procedure was _rst applied by...     »