Temporal disaggregation by dynamic regressions: recent developments in Italian quarterly national accounts

In this paper we discuss the most recent developments of temporal disaggregation techniques carried out at ISTAT. They concern the extension from static to dynamic autoregressive distributed lag ADL regressions and the change to a state-space framework for the statistical treatment of temporal disaggregation. Beyond the development of a unified procedure for both static and dynamic methods from one side and the treatment of the logarithmic transformation from the other, we provide short guidelines for model selection. The inclusion in the regressions of stochastic trends has been also discussed. From the empirical side we evaluate the new dynamic methods by implementing a large scale temporal disaggregation exercise using ISTAT annual value added data jointly with quarterly industrial production by branch of economic activity over the period 1995-2013. The main finding of this application is that ADL models either in levels and logarithms can reduce the errors due to extrapolating disaggregated data in last quarters before the annual benchmarks become available. When the attention moves to the correlations with the high-frequency indicators the ADL disaggregations are also generally in line with those produced by the static Chow-Lin variants, with problematic outcomes limited to few cases.