Online shopping is so 2019.
This is the claim analysts and pundits who point to the consumer’s return to the physical store as evidence of their waning appetite for digital purchases. Specifically, their claim is that the rate at which consumers shop online has returned to what one would have expected without the pandemic – that consumers’ increased use of digital channels to shop has been a blow. hard. In 2022, shopping is now business as usual in 2019.
Undoubtedly, the highest usage of e-commerce channels was recorded in the second quarter of 2020, when e-commerce shares accounted for 16% of sales. Since then we have seen these levels fluctuate, decreasing to 14% in Q1 2022 and increasing more recently to a seasonally adjusted level of 15% in Q3 2022. We have also seen an increase in brick usage . and mortar retailers from 2021.
However, it is incorrect to say that the growth of online shopping reflects longer-term pre-pandemic trends rather than a boost from consumer reliance on digital channels created by the pandemic.
The table below compares US Census data, and projections of e-commerce sales based on 2010 e-commerce share data to Q4 2019 (the last quarter before the start of the pandemic) with e-commerce sales reported by the census.
What we find is that the share of sales via e-commerce is about a percentage point higher today than it would have been without the increase in the use of digital channels by consumers between 2020 and 2022.
Source: PYMNTS.com, US Census Bureau data. Estimated quarterly US retail sales (seasonally adjusted): total and e-commerce.
While one percentage point may seem insignificant, it represents $81.6 billion in e-commerce sales over the past four quarters, an 8% increase in e-commerce sales attributed to increased channel usage. to make purchases between 2020 and 2022. According to census data, total retail sales over the last four quarters were $7 trillion, and $1 trillion of those retail sales were made via e-commerce techniques.
These data and the conclusions drawn from them are consistent across a variety of statistical models and estimation procedures. The one percentage point difference between actual shares of e-commerce sales and what they would have been based on projected data is the same when making estimates based on seasonally adjusted and unadjusted data. seasonally adjusted.
Similarly, these results also hold when using simple linear regression (OLS) models or regression models with robust standard errors.
More importantly, these results are also valid even if one uses more advanced statistical techniques such as Arima models specially designed to predict future data over a given period.
Even though analysts are revising their forecasts for brick-and-mortar store closures down, it’s easy to forget that brick-and-mortar retail has been in serious decline since 2010 with massive store closures every year – including 9,100 in 2019. consumers may be heading to stores more now than they were during the pandemic, but they’re still using e-commerce more than they would have if the pandemic hadn’t happened.
What all of this implies is that conclusions drawn about e-commerce gains that now go upside down fail to analyze the data in detail, and instead jump to a conclusion unsupported by both data and a common sense look at how consumers buy. PYMNTS first exposed the story of inaccurate reporting of census data on e-commerce earnings in May 2022. What this data shows is that what we were saying then about pandemic-spurred e-commerce earnings is, in fact, what happened.