The COVID-19 pandemic significantly altered productivity and daily life patterns, leading to changes in energy demand. While the overall reduction in energy consumption across various sectors is well-documented (e.g., gasoline, jet fuel), the impact on electricity consumption (EC) requires further investigation, particularly considering its regional variations. Previous studies have shown a link between gross domestic product (GDP) and EC, but this relationship is also influenced by the economic structure of a region. Metropolitan Statistical Areas (MSAs), with their high population density and diverse industries, provide a suitable scale for analyzing the interplay between economic structure and EC changes during the pandemic. This study focuses on the initial two months of the pandemic (April-May 2020) in the continental U.S., a critical period representing an unprepared response to a large-scale socioeconomic disruption. By analyzing county-level data aggregated to the MSA level, this research aims to uncover the relationship between economic structure and EC variations during this period, providing valuable insights for future crisis management and power grid planning.

Existing literature establishes a strong correlation between electricity consumption and economic activity, often using GDP as a primary indicator. However, a more nuanced understanding acknowledges that economic structure significantly impacts this relationship. Studies have shown that the composition of industries within a region influences the sensitivity of EC to overall economic fluctuations. For instance, regions heavily reliant on specific industries may experience different responses to economic shocks compared to those with more diversified economies. The COVID-19 pandemic, with its unprecedented lockdowns and work-from-home transitions, provides a unique opportunity to investigate this interplay. Prior research has documented declines in overall energy demand and shifts in peak demand hours during the pandemic, but comprehensive nationwide data at the MSA level has been limited. This study aims to fill this gap by providing a detailed analysis of EC variations across different economic structures within MSAs.

This study employs a multi-step methodology to estimate MSA-level electricity consumption changes between April-May 2019 (pre-pandemic) and April-May 2020. First, state-level electricity consumption data (total and residential) from the U.S. Energy Information Administration (EIA) was disaggregated to the county level using a linear regression model. County-level GDP data and population estimates were used as predictors for total and residential EC, respectively. The model parameters were initially calibrated using county-level data from California, which possesses more readily available and detailed data. These calibrated parameters were subsequently applied for EC estimations at the county level in other states, taking into account variations in GDP structure via a penalty coefficient based on information technology intensity for each MSA and industry. Once county-level EC estimates were obtained, they were aggregated to the MSA level. Second, a COVID-19 incidence map was created for each MSA, using a seven-day moving average of new cases per 100,000 people, categorized into four levels (low, medium, high, critical). Third, K-means clustering was used to group MSAs into eight clusters based on their economic structures, employing 20 GDP-related variables from 2019 data. Missing values were handled by filling with averages or using a proportion based on other available data and a fifth-root transformation was applied to mitigate right skewness. Statistical tests, such as the Wilcoxon rank-sum test, were used to compare the EC changes between clusters and against the overall MSA average. The study also conducted sensitivity analysis using updated data sources to assess the robustness of the findings. The maps are plotted using data from the U.S. Census Bureau.

The study revealed a significant spatial heterogeneity in electricity consumption changes across U.S. MSAs during the initial phase of the COVID-19 pandemic. Total electricity consumption decreased overall, but the magnitude of the decrease varied significantly by region and economic structure. The most significant decreases were observed in the Midwest (-8.88%), followed by the Northeast (-7.45%) and West Coast (-6.9%). However, some regions, particularly in the South, experienced slight increases. Residential electricity consumption, in contrast, exhibited a nationwide increase, reaching its largest increases in Arizona and Nevada. Analysis of economic structure clusters showed statistically significant relationships between the composition of industries within an MSA and the change in total electricity consumption. MSAs with a high proportion of mining and real estate/leasing industries experienced smaller decreases in total EC, while those with high shares of manufacturing showed more substantial reductions. The change in residential EC, however, showed no significant relationship with the economic structure or COVID-19 incidence level. The findings are validated against data from regional grid operators (CEC, MISO, and PJM), demonstrating consistency in the overall trend of total EC decrease and residential EC increase. Sensitivity analysis indicated the robustness of the patterns despite data updates.

The results highlight the complex interplay between pandemic-related restrictions, economic structure, and electricity consumption. The observed decrease in total electricity consumption is primarily attributed to reduced commercial and industrial activity due to lockdowns. The increase in residential consumption is consistent with the widespread adoption of work-from-home arrangements and increased time spent at home. The findings concerning the relationship between economic structure and total EC change underscore the importance of considering sectoral composition when forecasting electricity demand. The lack of a clear relationship between residential EC change and economic structure or COVID-19 incidence level suggests that behavioral changes related to lockdowns were a more dominant factor influencing residential EC compared to economic factors during this time. These results have crucial implications for power grid operators, policymakers, and researchers in understanding and preparing for future disruptions. The significant spatial variations warrant region-specific approaches to energy planning and management during crises.

This study provides a comprehensive analysis of electricity consumption changes across U.S. MSAs during the early stages of the COVID-19 pandemic, revealing a complex interplay between economic structure and pandemic-induced behavioral shifts. The methodology presented offers an effective approach to estimating MSA-level EC variations under crisis conditions. The results emphasize the need for incorporating regional economic factors and behavioral changes into future energy demand models, enabling more accurate predictions and robust grid management strategies during both pandemics and other large-scale societal disruptions. Future research can explore the long-term implications of pandemic-induced behavioral changes on electricity consumption patterns and refine the EC estimation models with additional data sources and more complex modeling techniques.

The study's estimations rely on assumptions regarding the extrapolation of linear relationships between GDP/population and EC from California data to other states. While efforts were made to adjust for regional variations via a penalty coefficient based on information technology intensity, further validation with county-level data from other states would strengthen the results. The model could also benefit from incorporating additional variables, such as climate data and other economic indicators, for enhanced predictive accuracy. Furthermore, the focus on the initial two months of the pandemic limits the understanding of long-term impacts on consumption patterns.