Aggregation and seasonal adjustment : empirical results for EMU quarterly national accounts

This paper investigates the differences between directly and indirectly seasonally adjusted aggregates. This difference is derived analytically for linear seasonal adjustment methods. GDP data for five European countries and three classes of seasonal adjustment methods are used to show empirically the differences between both approaches. For this purpose cointegration methods and cross-spectral analysis are applied. The analysis shows that there are no differences in the long-run components of directly and indirectly adjusted aggregates, whereas differences in the short-run components depend strongly on the seasonal adjustment methods applied in the indirect and direct approaches.