Time Series Clustering with Multiscale Bootstrap Resampling with Application to the Analysis of the COVID-19 Economic Impact

Abstract

The outbreak of COVID-19 had disruptive effects on China, and in this moment, and is spreading all over the world. The spread of the virus had impacted strongly on global markets and, at the same time, have threatened economic growth. Time-series clustering helps detect the relevant macroeconomic time series clusters related to the different economies and how the different economies have performed before and after the terrible shock. In this work, we consider an approach of clustering macroeconomic time series using a multiscale bootstrap approach considered on a helpful rolling analysis to measure the shock's impact on the significant clusters we can identify before and after the shock. We observe that were before the crisis, and the economy seems to be characterized by different clusters; after the shocks, the spread of the outbreak spread worldwide with disruptive effects globally. The cluster analysis is helpful to identify the propagation effects between the countries.