Understanding panic buying during COVID-19: A text analytics approach

Abstract

An area of consumer behaviour that caught retailers and supply chains unprepared during the initial stages of the COVID-19 pandemic was the increased prevalence of the purchase of utilitarian goods – referred to in the media as “panic buying.” In this study, we take a novel approach to understanding such panic buying during the pandemic using compensatory control theory (CCT), text analytics, and advanced data modelling. Using a big data set over 14 days from 24,153 Twitter users in Italy, we create dictionaries to capture CCT constructs and note the dates of two government announcements. We measure constructs in the longitudinal data and test the CCT model using generalized linear mixed models for both fixed effects and random variation across individuals and time. The results support CCT, with anxiety driving a lack of perceived control, moderated by effective government announcements, and a lack of perceived control leading to purchasing, negatively moderated by utilitarian qualities. The study demonstrates the benefit of the methods for studying social phenomena and for early warning of potential demand issues via social media.