This paper investigates the feasibility of using earlier provisional data to improve the now- and forecasting accuracy of final and official statistics. We propose the use of a multivariate structural time series model which includes common trends and seasonal components to combine official...