Refurbished data-driven selection of X-11 seasonal filters

The classical X-11 seasonal adjustment method for monthly and quarterly data contains an automatic selection routine for seasonal filters. Although this routine has a long-standing tradition, it has not been adopted in a recent X-11 modification tailored to the specifics of sub-monthly time series. We seek to fill this gap by suggesting a generic refurbishment of the legacy selection concept based upon the moving seasonality ratio and concurrent seasonal filters. The key idea is to consider multiple derivation principles for such filters, to develop a preliminary selection rule for each of these principles in line with the classical concept, and to obtain the final selection rule through a proper intersection of the preliminary rules. This approach stabilises the final filter choice and, as a by-product, naturally generates the warranted thresholds controlling the potential recalculation of the moving seasonality ratio from suitably shortened detrended observations. We consider one specific intersection based upon threshold quartiles to illustrate our proposed blueprint with three real-time macroeconomic data examples for the German economy. The case of daily electricity consumption particularly highlights the general need for additional intermediate steps in the calculation of the moving seasonality ratio when the data contain multiple seasonal patterns.