Visualizing the impact of COVID-19 on retail sales

Retail is one of the sectors that suffered the most with the pandemic, and it may never recover

Daniela Fernandes
Towards Data Science

--

Source: Graph created by the author using data from Census Bureau Monthly Retail Sales Report

The global pandemic due to Covid-19 has now stretched beyond the first half of 2020, with several countries still reporting thousands of cases every day. Initial actions designed to contain the spread of the virus and flatten the curve to save lives also had the unfortunate effect of limiting economic activity drastically, as most people stayed home to comply with the lockdown measures. Sectors that depended heavily on people’s traffic were sent into UCI, with an unprecedented number of businesses either working at a limited capacity or ceasing operations altogether. It is unclear if some industries will ever recover from the impact.

One such industry is retail, according to the data retrieved from the Census Bureau, in two months adjusted retail sales went from 527.3 billion in February to 412.8 billion in April. It represents an overall fall in sales of approximately 22%, and it also translated into many businesses closing their doors, perhaps forever.

The data

The Census Bureau routinely publishes data on retail sales across the country that can be downloaded directly from the official site in excel format. The excel file contains data for all months since 1992, with every sheet named after the year. From there, I extracted the adjusted data (If you are interested in the process, you are welcome to visit the Github project and see the transformations and the code, for the plotly plots use the NBViewer).

I chose adjusted data because the estimates shown are adjusted for seasonal variations and holiday and trading-day differences, meaning that the effects shown in the graphs will correspond to COVID-19 alone. One of the trade-offs that come with using adjusted data is that some of the detail from the non-adjusted is lost as you can see below.

Source: Photo by the author. Data from Census Bureau Monthly Retail Sales Report
Source: Photo by the author. Data from Census Bureau Monthly Retail Sales Report

There are less NAICs in the adjusted slice, which means that we will be working with the aggregates that compose each code. In total, adjusted data has thirty sectors, which is more than enough to get a general picture of how each fared.

The pre-pandemic reality

Source: Graph created by the author using Data from Census Bureau Monthly Retail Sales Report

In the graphic above, most of the thirty areas identified in the adjusted data reported to have growth in sales in January, with only some exceptions. Five sectors were presenting declines below the 2% margin, and only Fuel dealers seemed to exhibit signs of ‘struggle’, with a glaring drop of ~7.5% respect to December of 2019.

Source: Graph created by the author using Data from Census Bureau Monthly Retail Sales Report

February brought a bit of struggle for almost all sectors in retails, with an overall decrease of 0.44% in the industry and most sectors lost a relatively small chunk of sales before disaster struck in March.

The lockdown

Source: Graph created by the author using Data from Census Bureau Monthly Retail Sales Report

While some area thrived under the lockdown measures like Grocery Stores, Pharmacies and Nonstore retailers; others suffered a disproportionate impact, hurting more than its peers by a wide margin. One such case is Shoe Stores that lost a whopping ~50% in March alone.

The situation only worsened in April, when distancing measures entered into effect. Only two sectors experienced growth in April as the industry as a whole bled a record 14.71% of its sales away.

Source: Graph created by the author using Data from Census Bureau Monthly Retail Sales Report

Among the most affected by the lockdown were clothing stores, furniture and appliances stores, food services and drinking places, sporting and hobbies stores and gasoline stations.

Going by intuition alone, these results are not surprising, as these stores rely heavily on traffic to secure most of their sales. They also share the quality of being non-necessities, which means that the public could drop these expenses in favour of diverting funds to buying necessary items in case of a crisis.

The Beginning of the recovery

With the start of May, some states lifted the restrictions and allowed people to go back to their daily routines. These actions translated into sales going up for most sectors, especially the most hit by the pandemic. Shoe Stores alone recovered over 200% in sales, with Clothing stores recovering 176.88%.

Source: Graph created by the author using Data from Census Bureau Monthly Retail Sales Report

These numbers, however, still fall short on what the sectors used to sell before the lockdown induced by COVID-19, as the graphic below can prove.

Source: Graph created by the author using Data from Census Bureau Monthly Retail Sales Report

Conclusion

Small steps towards recovery from the initial shock from coronavirus are in progress. It is pleasing to see signs of further improvement in the preview of the monthly report of June 2020, that signals an overall increase of retail sales of approximately 7.5% as reported by the Wall Street Journal.

However, with the number of cases in the United States increasing daily, the probability of lockdown measures coming back raises and the gains made in May and in June could be lost.

[1] Census Bureau, Monthly Retail Trade Report, (2020)

[2] H. Torry, U.S. Retail Sales Rose 7.5% in June as Stores Reopened, (2020), Wall Street Journal

[3] Worldometer, United States Coronavirus, (2020).

--

--

Venezuelan economist passionate about all things data. Currently working as a data scientist/engineer at Graphite