Understanding Economic Resilience through Consumer Spending Trends

When the COVID-19 pandemic swept across the World, households faced unprecedented financial insecurity through lost jobs, reduced incomes, and dramatic shifts in expenses. Families had to make tough economic decisions—discretionary spending took a backseat, and essentials became a priority. But was the impact the same everywhere and in all sectors of the economy? This study leverages GeoInsights spending data to explore how different communities adapted to economic disruptions. 

Project overview 

Traditional economic indicators, such as employment statistics and GDP growth rates, often fail to capture real-time shifts in consumer behaviour at granular geographic levels (Aladangady et al., 2019). Research indicates that economic impact assessments using surveys and broad economic models lack spatial and sectorial granularity (Flynn et al., 2024), highlighting the need for high-frequency transaction-level data that monitors consumer spending behaviours (Dunn et al., 2020; Chetty et al., 2020). Lower-income areas are also often under-represented, thus making it necessary to blend transaction data with socioeconomic factors (Chronopoulos, Lukas and Wilson, 2020).  

This project integrates spending data with socioeconomic and COVID-19 datasets to examine consumer spending changes across 13 Yorkshire and the Humber industries during key pandemic phases. It seeks to evaluate resilient sectors and evolving consumer priorities, offering real-time insights to policymakers, businesses, and researchers through advanced spatial analysis, data integration and visualisation. 

Data and methods  

This research combines transaction-level spending data from MasterCard with employment statistics and a COVID-19 dataset to assess the adoption of different industries and regions' economic disruptions and how spending trends evolved. The inclusion of open-source datasets from the Office for National Statistics (ONS) and the UK Health Security Agency (UKHSA) was discussed based on the project scope and timeline.  

The project integrated four datasets: 

1. GeoInsights Spending Data – This contains 7 key metrics capturing granular spending insights across industries and regions from January 2020 to June 2024.  
2. Employment Data – Provided workforce statistics as yearly sums from 2020 to 2023 from the ONS and was one representative socio-economic indicator. 
3. COVID-19 Data – Sourced from the UKSHA, this included 14-day average COVID-19 case counts, 14-day average COVID-19 deaths, and daily cumulative vaccination records from June 2020 to December 2023. 
4. COVID-19 Timeline Data – We recorded lockdown phases within the UK using information from the Institute for Government (IFG) and the BBC. 

Limitation of the Data and Challenges in Data Integration: 

1. Indexed Metrics: All the spending metrics were indexed values relative to 2018, not actual amounts, which imposed limitations. While we could compare or aggregate indexes across multiple geographies for an industry, the vice-versa was impossible. 
2. Mismatch in Location Reference: The spending data was based on the Point of Transaction (PoT), whereas the socio-economic and COVID-19 datasets were home-location-based, creating challenges around linking the datasets to capture the correct location information. 
3. Geographic Aggregation Issues: The datasets existed at different spatial levels—spending data at the postcode sector level, employment data at the Middle Layer Super Output Area (MSOA) level, and COVID-19 data at the Lower Tier Local Authority (LTLA) level—creating issues in consistent geo-mapping of these geographic units.  

Solution Approach (Figure 1):  

Addressing the Location Mismatch – We aimed to ensure consistency by mapping home-location-based variables (employment and COVID-19 data) to postcode sectors. 

Resolving Geo-Aggregation Issues—We used QGIS to generate a spatial mapping dataset of MSOAs to postcode sectors across Yorkshire and the Humber. The dataset captured the proportion of each MSOA covered by a postcode sector and thus we redistributed industry employment counts. We extended this proportional mapping to derive the local authority to MSOA mapping to link LTLA level COVID-19 data to their respective MSOAs.   

The datasets were then merged, facilitating consolidated information on spending trends with socio-economic and COVID-19 indicators, forming the foundation for an interactive Power BI dashboard that visualised key insights into consumer behaviour and economic resilience around the pandemic.  

Key findings  

Users can utilise the dashboard to view specific visuals by filtering by time, industry and different geographies to analyse trends dynamically (Figure 2 and 3).  

 

Key insights are visualised through: 

1. KPI Cards – snapshot of key metrics, comparing with previous year. 
2. 2 Time-Series Charts – one displaying spending behaviour and another displaying COVID-19 temporal trends with lockdown phases.  
3. Choropleth Maps – highlight employment distribution across regions 
4. Bar Chart – top 10 postcode sectors by spending quantity with breakdown of full-time and part-time employment.  

The dashboard offers tooltips for quick insights with drill-down, drill-up and expandable date levels for deeper exploration, enabling a detailed assessment of economic resilience. As examples of the dashboard's usage, we analysed one essential (Grocery and Food Stores) and one discretionary (Bars/Taverns/Night Clubs) and conducted a consumption analysis of key cities, including Leeds, Sheffield, and York.  

Grocery and Food Stores

During the early COVID-19 lockdowns, panic buying increased total transaction frequency by 10.69% and purchase quantity by 5.2%. By late 2020, however, per-transaction spending declined by 12%, reflecting reduced stockpiling and increased financial caution as COVID-19 cases surged. Spending was the highest in Leeds, Sheffield, and York, with stronger trends in areas with high part-time employment, suggesting work pattern shifts.  

From Q1 2021, increasing vaccination rates stabilised spending and regularised shopping behaviours returned. Interestingly, the subsequent Omicron variant did not cause significant disruptions in expenditures. Regions with higher full-time employment maintained consistent high (or low) spending patterns with equally influential part-time counts, indicating diverse consumer behaviour.  

Bars/Taverns/Nightclubs

Between Q2 2020 and Q1 2021, the industry experienced multiple declines in consumer spending due to lockdown restrictions, resulting in closures and reduced transaction frequency. Despite this, temporary re-openings in Q3 2020 saw ~13% higher spending per transaction, suggesting that limited capacities led to maximise socialisation opportunities and hence spending during fewer visits. Renewed restrictions and rising COVID-19 cases (258.8 per day) in late 2020 further delayed recovery.  

Spending gradually recovered from Q2 2021, reflecting growth in consumer adaptation to ongoing pandemic conditions. Although the Omicron wave briefly slowed spending during Q4 2021 and Q1 2022, the impact was less severe than earlier lockdowns. By late 2022, spending frequency and quantity nearly reached pre-pandemic levels, supported by high vaccination rates, relaxed restrictions, and improved public confidence.  

Leeds, Sheffield, and York led the recovery in nightlife spending, which extended beyond city centres into suburban areas. Postcodes with substantial full-time employment reported higher average spending per transaction, while younger demographics contributed through increased part-time employment. The MasterCard data highlights key spending patterns—such as the correlation between full-time jobs and consistent expenditure—demonstrating the dashboard's ability to provide nuanced insights. 

City-wise Consumption Analysis

Grocery and Food Stores

Leeds: Early lockdowns (Q2-Q4 2020) witnessed panic-buying with total per-transaction spending rising by 30%, which stabilized owing to vaccination rollouts. There was a concentrated mix of central and suburban high-spending areas, where part-time employment mostly exceeded full-time. 

 

Sheffield: Experienced pronounced sensitivity to COVID-19 waves, with spending fluctuations reflecting reactions to pandemic peaks, stabilising gradually post-pandemic. High-spending areas were concentrated in central areas towards Rotherham, with relatively low employment counts suggesting influence of factors like store density and commuting patterns. 

York: This city had a more balanced employment profile, with sharp initial increases in frequency and quantity but smaller average transaction sizes suggesting frequent but smaller purchases during early lockdown. Post-pandemic stabilization occurred steadily, supported by a diversified employment base, seasonal tourist-driven peaks, and sustained vaccination progress. 

 Bars/Taverns/Nightclubs

Leeds: Leeds quickly rebounded after initial severe declines in spending frequency and per transaction spend (57% decline), recovering strongly post-2021. The top spend areas (LS1 and LS2) were dominated by full-time employment, suggesting that nightlife spending was primarily co-located with working professionals.  

Sheffield: The city experienced delayed and fluctuating recovery due to prolonged business disruption. Omicron had a visible impact like York – but vaccinations minimised the effect. Spending dispersed more geographically than Leeds, with notable influence from part-time employment, suggesting a diverse workforce engagement.   

 

York: York saw the steepest decline in spend frequency (How Often Do Customers Buy?) with a gradual nightlife recovery. Spending was more balanced beyond the city centre (Where Are Customers Buying More?). A near equal (6 out of 10 postcodes) full-time/part-time employment split (Who Buys More?) indicates a steadier nightlife economy than Sheffield and Leeds, highlighting the dashboard’s ability to analyse regional economic resilience. 

Overall, the dashboard highlights the evolution of spending across essential and discretionary sectors and underscores the importance of dynamic, region-specific analyses in understanding economic resilience.  

Value of the research   

1. Supporting Policy and Decision-Making: The insights from the dashboard will assist policymakers identifying regions and industries most vulnerable to economic shocks, for example, area surrounding a university, allowing targeted economic recovery efforts.  
2. Business Applications: The industry-specific insights enable businesses to quickly identify changing post-pandemic consumer behaviours and adapt decisions around products and services.  
3. Socio-economic Resilience: The dashboard offers insights that help policymakers and local authorities devise strategies to enhance socio-economic equity and build resilience against future economic disruptions.  

Insights 

• Essential expenditures stabilised quickly after initial lockdown volatility, reflecting shifts in consumer confidence.  
• Discretionary sectors faced prolonged impact, but vaccination programmes and eased restrictions facilitated a steady recovery.  
• Future research can integrate additional demographic factors to refine understanding of behaviours and resilience post-pandemic.  
• Predictive modelling can be implemented to forecast spending behaviours that can have further applications.  

Research theme 

• Health  
• Societies 
• Food 

Programme theme 

• The Science of Data Science 
• Visualisation/ Extended Reality 
• Mathematical and Computational Foundations 

Team 

• Gauri Venkatachalapathi - Data Scientist, Leeds Institute for Data Analytics, University of Leeds. 
• Dr. Moises Sanchez - Lead Research Data Scientist, Consumer Data Research Centre, Leeds Institute for Data Analytics, University of Leeds 
• Dr. Stephen Clark – Research Fellow, Consumer Data Research Centre, Leeds Institute for Data Analytics, University of Leeds 

Partner(s) 

N/A 

Data Provider  

MasterCard (MasterCard GeoInsights Spending Data) 

Funder 

Funded by ESRC Consumer Data Research Centre 

This work has been facilitated by the Leeds Institute for Data Analytics (LIDA) Data Scientist Development Programme, which employs early-career data scientists to deliver real-world data-driven impact in the interests of the public good. 

References 

1. Aladangady, A., Aron-Dine, S., Dunn, W., Feiveson, L., Lengermann, P. and Sahm, C. (2019). From Transactions Data to Economic Statistics: Constructing Real-time, High-frequency, Geographic Measures of Consumer Spending. Finance and Economics Discussion Series, 2019.0(57). doi: https://doi.org/10.17016/feds.2019.057.  
2. Chetty, R., Friedman, J., Stepner, M., Libsch, A., Taska, B., Witten, J., Natesan, A., Palaniappan, R., Sandza, R., Vogeley, A., Foo, C., Swaminathan, K., Gilbertson, D., Nichols, M., Sifain, S., Doel, D., Thorpe, R., Mcrae, B., Sharma, S. and Autor, D. (2020). NBER WORKING PAPER SERIES THE ECONOMIC IMPACTS OF COVID-19: EVIDENCE FROM A NEW PUBLIC DATABASE BUILT USING PRIVATE SECTOR DATA. 
3. Chronopoulos, D.K., Lukas, M. and Wilson, J.O.S. (2020). Consumer Spending Responses to the COVID-19 Pandemic: An Assessment of Great Britain. SSRN Electronic Journal. doi: https://doi.org/10.2139/ssrn.3586723. 
4. Dunn, A., Hood, K., Driessen, A., Chen, J., Cornwall, G., De Francisco, E., Fernando, L., Fixler, D., Gholizadeh, M., Gottlieb, J., Knepper, M., Samuels, J. and Shapiro, A. (2020). Measuring the Effects of the COVID-19 Pandemic on Consumer Spending Using Card Transaction Data. 
5. Flynn, M., Ballantyne, P., Anderson, R. and Singleton, A. (2024). Eurovision’s economic impact in Liverpool Insights for future large-scale events Eurovision’s economic impact in Liverpool: insights for future large-scale events Key takeaways. [online] Available at: https://www.liverpool.ac.uk/media/livacuk/publicpolicyamppractice/pbseries3/PB315.pdf.