Missing observations in stochastic difference equation with arma errors

Abstract

This paper considers the estimation of stochastic difference equation with missing observations. Time series data is available at different levels of aggregations. The most common is the annual-quarterly model, where for the first part of the time series data, some of the endogenous and/or exogenous variables are available on an annual basis and the second part on a quarterly basis. A method of estimation, based on the Kalman Filter, enables us to obtain the exact maximum likelihood estimates of the model.