Time Series Methods for Maintaining Official Statistics in the Face of COVID-19

• Activity Number: 411 - Time Series Methods for Maintaining Official Statistics in the Face of COVID-19
• Type: Topic-Contributed
• Date/Time: Thursday, August 12, 2021 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Kathleen M McDonald-Johnson* and Steve Matthews* and Craig McLaren* and Jay Mousa* and Rodrigo Mariscal Paredes* and Yingfu Xie*
• Companies: U.S. Census Bureau and Statistics Canada and Office for National Statistics and U.S. Bureau of Labor Statistics and Secretaría de Hacienda y Crédito Público and Statistics Sweden
• Keywords: seasonal adjustment; pandemic; outlier; intervention analysis; regression; modeling

Abstract

The pandemic resulting from the SARS-CoV-2 coronavirus illness (COVID-19) has affected every corner of the world, resulting in turmoil as formerly predictable business activity changed suddenly, for example, more activity at supermarkets and less activity at restaurants. Such changes in turn affect longstanding behavior of time series. Like the global recession of the late 2000s, the pandemic effects were far-reaching and ongoing, meaning that modeling tools were limited during initial estimation. But unlike the recession, the recovery period has been less predictable, and the full impact of the pandemic on these series might not be apparent for some time. In this panel, representatives from a varied group of agencies will discuss their approaches to maintaining time series outputs in real time, presenting examples and concerns related to the short-term and long-term efforts to identify and account for the pandemic’s effects on the time series.