Quantifying the impact of a natural disaster on retail stores' customer behavior: the case of Ecuador

Data analytics are essential to retailers when making decisions during disruptive events such as natural disasters. In recent years, research on the effects of natural disasters through digital data has gained relevance, but there is little information regarding how the customer behavior changes after such an event in developing countries. In this work, we get hold of 23 millions anonymized transaction records that occurred in Ecuador in a three months period in 2016, the year when an earthquake of magnitude 7.98 on Ritcher' scale, took place in that country. We analyze the differences between an area directly affected by the earthquake and a neighbor province, by aggregating the transactions at customer level. We find significant differences in the month when the disaster occurred: an increment of up to 45.47% in the average units sold per customer and up to 53.33% in the average expenditure per customer. Moreover, our results show evidence that more than 270 kilometers away from the epicenter, in a slightly-affected city, changes remain relevant: an increment of up to 25.82% in the average units sold per customer and up to 23.77% in the averaae expenditure per customer.