Abstract

This paper introduces a latent GDP indicator at a weekly frequency approximating week-on-week growth rates of the GDP in Germany. To this end, I use a mixed-frequency dynamic factor model with quarterly, monthly and weekly indicators to extract a weekly factor approximating the latent weekly GDP growth rates. In addition, the employed methodology enables to create a weekly GDP series in levels, which is also of paramount interest for practitioners interested in analysing the current state of the economy. In a recursive nowcasting exercise for the German GDP growth I show that (i) the proposed index may provide a timely and data driven assessment of the GDP and (ii) may be beneficial to the nowcasting performance; (iii) while its forecast performance weakens beyond the current quarter.