We develop a new model representation for high-dimensional dynamic multi-factor models. It allows the Kalman filter and related smoothing methods to produce optimal estimates in a computationally efficient way in the presence of missing data. We discuss the model in detail together with the implementation of methods for signal extraction and parameter estimation. The computational gains of the new devices are presented based on simulated data-sets with varying numbers of missing entries

Additional Metadata
Keywords Kalman filter, high-dimensional vector series, maximum likelihood
Publisher Tinbergen Institute
Persistent URL hdl.handle.net/1765/14942
Jungbacker, B, Koopman, S.J, & van der Wel, M. (2009). Dynamic Factor Analysis in The Presence of Missing Data (No. TI 09-010/4). Discussion paper / Tinbergen Institute. Tinbergen Institute. Retrieved from http://hdl.handle.net/1765/14942