Asymmetric Linear Filters for Seasonal Adjustment: Applications to the COVID-19 Impact in Monthly Time Series

• Activity Number: 199 - Novel Time Series Approaches for Official Statistics in the Time of COVID-19
• Type: Topic-Contributed
• Date/Time: Tuesday, August 10, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Alain Quartier-la-tente* and Dominique Ladiray
• Companies: ENSAE and Independant
• Keywords: Seasonal adjustment; asymmetric filters; turning points; Covid-19

Abstract

The COVID-19 crisis highlights that business cycle analysis, and in particular the early detection of turning points, is a major topic in the analysis of economic outlook. Moving averages, or linear filters, are ubiquitous in business cycle extraction and seasonal adjustment methods. For example, the X-12ARIMA seasonal adjustment method uses Henderson moving averages and composite moving averages to estimate the main components of a time series, while TRAMO-SEATS uses Wiener-Kolmogorov filters. Symmetric filters are applied to the center of the series, but when it comes to estimate the most recent points, all of these methods must rely on asymmetric filters which introduce phase-shifts and delays in the detection of turning points. Construction of good asymmetric filters, in terms of fidelity, revisions, smoothness and timeliness, is still an open topic. This presentation describes and compares the recent approaches on trend-cycle extraction and asymmetric filters, and applies them to the COVID-19 economic crisis. All the methods are implemented in the R package rjdfilters (https://github.com/palatej/rjdfilters) and the results can be easily reproduced.