Forecasting and Planning with Auxiliary Data During a Pandemic: Evidence from U.K. Google Trends

In this chapter, we develop reliable methods to predict excess demand during the COVID-19 and similar pandemics. Weekly Google Trend data is gathered for nine retail products in three different categories electronics, toiletries, and groceries. Forward Stepwise Selection is used to investigate the relationship between excess demand and multiple other variables at once. Finally, various models based on Regression, Decision Trees, and Neural Networks are developed for demand prediction. The findings are that forecasting during a pandemic was less challenging for electronics than for toiletries and groceries. Consumer demand is affected by the public health environment of a pandemic, and consumers’ panic buying and hoarding had a serious impact on demand for those products for which they had most immediate need, such as toiletries and groceries.