Reducing revisions in real time trend-cycle estimation.

Recently, reproducing kernel Hilbert spaces have been introduced to provide a common approach for studying several nonparametric estimators used for smoothing time series data (Dagum and Bianconcini, 2008 and 2011). Based on this methodology, Bianconcini and Quenneville (2010) focused on the properties of the Henderson reproducing kernels when the filters are adapted at the end of the sample period, and with particular emphasis on the influence of the kernel order and bandwidth parameter. In this paper, we design a family of trend filters applied for real time estimation that are optimal in terms of reducing revisions when new observations are added to the series, and that are characterized by a fast detection of true turning points.