Trend-cycle filters comparison for real time macroeconomic data

Leading, coincident and lagging indicators have long been used to analyze and assess the current stage at which the economy stands. Official statistical agencies have generally applied linear filters developed by Musgrave (1964) to produce preliminary estimates of the trend-cycle of these indi- cators, but these estimates are subject to revisions as new observations are added to the series. To reduce the revisions size, cascade linear filters developed by Dagum and Luati (2009) have been recently used. However, only asymmetric filters related to the 13-term symmetric one are available, whereas, due to more variability in the data introduced by major financial and global changes in the economy, different filter lengths are needed to produce smoother estimates. We describe and propose a new procedure to reduce the size of the revisions and make the indicators more timely. These new filters significantly outperform their older counterpart. They offer substantial gains in real-time by providing timely and more accurate information for detecting short-term trend turning points.