This paper concerns estimating parameters in a high-dimensional dynamic factormodel by the method of maximum likelihood. To accommodate missing data in theanalysis, we propose a new model representation for the dynamic factor model. Itallows the Kalman filter and related smoothing methods to evaluate the likelihoodfunction and to produce optimal factor estimates in a computationally efficient waywhen missing data is present. The implementation details of our methods for signalextraction and maximum likelihood estimation are discussed. The computational gainsof the new devices are presented based on simulated data sets with varying numbersof missing entries.
High-dimensional vector series Kalman filtering and smooting Maximum likelihood Unbalanced panels of time series