High-dimensional conditionally Gaussian state space models with missing data

We develop an efficient sampling approach for handling complex missing data patterns and a large number of observations in conditionally Gaussian state space models. Two important examples are dynamic factor models with unbalanced datasets Bayesian VARs variables multiple frequencies. A key observation underlying the proposed is that joint distribution conditional on observed Gaussian. Furthermore, inverse covariance or precision matrix this sparse, special structure can be exploited to substantially speed up computations. illustrate methodology using two empirical applications. The first application combines quarterly, monthly weekly VAR produce GDP estimates. In second application, we extract latent factors from involving over hundred via model stochastic volatility.