Temporal disaggregation and the adjustment of quarterly national accounts for seasonal and calendar effects

The statistical treatment of seasonality and calendar effects in the estimation of quarterly national accounts raises a number of issues that bear important consequences for the assessment of current economic conditions. In many European countries, the quarterly national accounts are constructed by national statistical institutes by disaggregating the original annual measurements using related monthly indicators. In this article we propose and evaluate an alternative approach that hinges upon the estimation of a bivariate basic structural time series model at the monthly frequency, accounting for the presence of seasonality and calendar components. Its main virtue is to enable the adjustment and temporal disaggregation to be carried out simultaneously. The proposed methodology also complies with the recommendations made by the Eurostat - European Central Bank task force on the seasonal adjustment of quarterly national accounts. The overall conclusion is that the identification and consequently the separation of seasonal and calendar effects from aggregate data is highly controversial.