A Comparison Between Direct and Indirect Seasonal Adjustment of the Chilean GDP 1986-2009 with X-12-ARIMA

It is well known among practitioners that the seasonal adjustment applied to economic time series could involve several decisions to be made by the econometrician. In this paper, I assess which aggregation strategy delivers the best results for the case of the Chilean GDP 1986-2009 quarterly dataset (base year: 2003). This is done by performing an aggregate-by-disaggregate analysis under different schemes, as the fixed base year dataset allows this fair comparison. The analysis is based exclusively on seasonal adjustment diagnostics contained in X-12-ARIMA program. A detailed description of the program and its quality assessment are also provided. The results show that it is preferred, in terms of stability, to use the first block of supply-side disaggregation as well as the direct mode.