How does COVID-19 affect the life cycle environmental impacts of U.S. household energy and food consumption?

During the pandemic, 255 Mton CO₂ eq of life cycle GHG emissions were reduced by U.S. households (figure 1(a)), accounting for a 7.4% decrease, as compared to the GHG emissions before the pandemic. This reduction was due to decreased transportation (8.7% reduction) and food consumption (1% decrease), and increased direct energy consumption (2.3% growth) (figure 2). This decrease in GHG emissions because of reduced transportation was driven by the reduced use of gasoline and fuel products for vehicles (figure 3(a)), which accounted for 82%–83% of the total life cycle GHG emissions associated with transportation before and during the pandemic (tables S10 and S11). The remaining GHG emissions were associated with transportation services that were also halted during the pandemic. An increase in life cycle energy-related GHG emissions was mainly attributed to the growth of household electricity consumption (figure 3(b)). Electricity accounted for 62.8% of the total life cycle GHG emissions of household direct energy consumption before the pandemic (table S13), which increased to 68.3% during the pandemic (table S14).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Similar to the GHG emissions results shown in figure 1(a), the energy results in the same figure show a reduction of life cycle energy use (4.46 EJ, 8.7% reduction) as a result of 5.34 EJ reduction of non-renewable energy (driven by the declined gasoline usage, figure 3(a) and 0.87 EJ growth of renewable energy (driven by increased electricity consumption, figure 3(b). The contribution of household food consumption to reductions in energy use and GHG emissions was minimal (∼1%) (figure 2). The increase in the use of renewable energy should not be interpreted as an increase in renewable energy use in households but as an increase in renewable energy across the life cycle of electricity used by households. This interpretation is more consistent with the data collected by the U.S. EIA, which showed an 8.6% increase in renewable energy consumed for electricity generation [46]. On the other hand, the renewable energy directly consumed by the U.S. residential sector decreased by 5.7% (table S22) [46]. Although renewable energy consumption reached a record high of 12% of total U.S. energy consumption in 2020 [57], previous studies have pointed out the potential adverse long-term effects of COVID-19 on renewable energy transition by decreasing economy-wide demand [58] and decelerating clean energy innovation [4].

The net differences in life cycle GHG emissions and energy use between decreased transportation and increased direct energy consumption of U.S. households were quantified in this study. Such differences are likely to decrease during the post-pandemic period with resumed travel. Previous studies [4, 6] pointed out that workers and employers may be comfortable with remote working, and this may reduce travel but increase building energy use; however, they have not explored the net differences. The results of this study can be used as a foundation for future analysis of trade-offs regarding remote working and lifestyle changes in the post-pandemic period.

Four human health impact categories were investigated in this study: cancer and non-cancer-related human health impacts, respiratory effects, and smog formation. All four human health-related impacts were reduced during the pandemic due to reduced travel. The health impacts are in the unit of Comparative Toxic Unit for Humans (CTUh), which estimated an increase in morbidity, in other words, disease cases [59, 60]. Cancer-related human health impact decreased 99 CTUh and non-cancer-related impact showed a decrease of 790 CTUh (disease cases), which were far less than the deaths caused by COVID-19.

The respiratory effects were estimated in terms of particulate matter equivalent (PM_2.5 eq). This study showed a net decrease of 35 kton PM_2.5 eq (figure 1(a)) across the life cycle of food and energy consumed by U.S. households. According to the U.S Environmental Protection Agency (EPA), national PM_2.5 primary emissions excluding miscellaneous emissions (e.g. emissions from forest fires) were 1.33 Mton in 2019 and 1.32 Mton in 2020 [61]. These values were similar to the estimates of this study (1.1 Mton for 2019 and 1.09 Mton for 2020). The EPA's estimations were larger as they covered all emissions sources in the U.S. that had a broad scope. On the other hand, this study had a narrower scope as it specifically quantified the changes in life cycle respiratory effects associated with U.S. household consumption of food, energy, and transportation. The life cycle respiratory effects before and during the pandemic were dominated by food (figures 1(b) and (c)). These results are consistent with those in the reported literature that showed significant PM_2.5-related emissions contributed by the agriculture sector in the U.S [62]. A large portion of PM emissions were attributed to the production of FAH, such as meat and miscellaneous foods (figure 4(b)). Meat consumed at home, including beef, pork, poultry, and other meats, accounted for 36% of the life cycle respiratory effects of household food consumption during the pandemic (figure 4(b)). This is consistent with the literature demonstrating that 80% of air-quality-related health damages from food production were contributed by activities involved in the life cycle of animal-based foods, including animal production and growing biomass for animal feed [63]. Miscellaneous foods, including crackers, seasoning and dressing, cookies, snacks, and frozen food contributed 13%. This significant contribution of FAH to the life cycle respiratory effect is unlikely to be reduced even when the pandemic is over, given the similar patterns shown in figure S1 for 2019, unless U.S. households purchase less meat and miscellaneous food. Consumption-based strategies (e.g. reducing food purchases by minimizing food waste and stockpiling) and production-based efforts to reduce PM emissions and corresponding respiratory effects in agricultural practices are needed for effective air quality management in the post-pandemic period.

Zoom In Zoom Out Reset image size

Another human-health-related environmental impact is the smog formation potential related to ground-level ozone caused by air pollutants NO_x and volatile organic compounds (VOCs) [64]. Previous studies reported a reduction in free tropospheric ozone by up to 5 parts per billion globally during the pandemic due to reduced anthropogenic NO_x emissions, with substantial emission reductions in Asia and the Americas [65]. This observation is consistent with the results of this study. Figure 1(a) shows a net life cycle decrease of 9.17 Mton O₃ eq due to reduced travel that offsets increased household energy and food consumption. This is not surprising because the tailpipe emissions of on-road transportation contributed 40%–60% of the precursors of ground-level ozone in the U.S [66]. Globally, ∼8.8 million excess deaths were estimated to be caused by long-term exposure to PM_2.5 and ozone pollution [67]. Previous literature has also revealed a significant relationship between ground-level ozone and positive COVID cases [68]. Therefore, mitigating ground-level ozone is particularly meaningful during the pandemic and remains essential for post-pandemic recovery.

Freshwater, land use, and mineral and metal use showed different trends during the pandemic. The net increase of life cycle freshwater withdrawals is 8.6 Gton, accounting for 2% of national water use (based on the USGS estimation of national water use as 322 billion gallons per day in the U.S., equivalent to 445 Gton yr⁻¹ [69]). Energy consumption (particularly electricity, figure 3(b)) contributed to the greatest increase in life cycle freshwater withdrawals. This conclusion should not be interpreted as an increase in water consumption of the electricity sector driven by an increase in household energy consumption, as EIO-LCA only quantifies the life cycle environmental impacts associated with a specific product or consumption [70]; this study did not intend to use EIO-LCA to investigate the drivers of water consumption in the U.S. electricity sector.

The life cycle use of land, and minerals and metals showed a net decrease of 41 billion m²*year and 16.1 Mton, respectively, although the reduction of land use appeared to be minor (−1.3%; figure 2). Similar to other impact results, this decrease was mainly attributed to the lower usage of gasoline and other fuel products for vehicles (figure 3(a)). Land, minerals, and metals are mainly used in the production of fuels and relevant infrastructure; therefore, this reduction is temporary. Improving land management and material efficiency is essential to curb the increase in resource use when the pandemic ends.

Household food consumption dominated all three resource uses before and during the pandemic (figures 1(b) and (c)); however, it had minor contributions to the changes in resource use (figure 1(a)) and showed varied trends for different resources. Specifically, changes in U.S. household food consumption led to increased life cycle land use but decreased freshwater withdrawals and mineral and metal use (orange bars in figure 1(a)). Meat consumption (i.e. beef, pork, poultry, and other meats) drove the results of life cycle land use associated with food (∼61%; figure 4(b)). Most meats such as beef are consumed at home [71], which may explain why the shutdown did not bring considerable reductions in life cycle land use.

The trend was opposite for mineral and metal use, where the reduced FAFH led to more reductions than increased FAH. Before and during the pandemic, FAFH (including full- and limited-service restaurants, other drinking and eating places, and alcoholic beverages consumed outside homes) was the most significant contributor to the life cycle mineral and metal use. FAFH contributed 28.7% of the total mineral and metal use (figure S1), and this contribution reduced to 23.4% during the pandemic. However, this was still higher than that of the other individual food items consumed at home (figure 4(b)).

Regarding freshwater withdrawals, reduced FAFH due to lockdown compensated for the effects of increased household food consumption, as shown in figure 4(a). About 30% of the life cycle freshwater withdrawal associated with food was attributed to fresh fruits and vegetables consumed at home before and during the pandemic (figure 4(b)). Interestingly, the contribution of fresh fruits to the life cycle freshwater withdrawal decreased slightly from 16.2% in 2019 to 15.8% in 2020, while the contribution of fresh vegetables increased from 13.7% to 14.7% during the pandemic (figures 4(b) and S1). This can be explained by the different patterns of fresh vegetables and fruit consumption. According to the USDA, U.S. households consumed fewer fresh fruits but more fresh vegetables during the pandemic [45, 72].

These results indicate the trade-offs between FAH and FAFH, as well as the complications in life cycle resource use associated with various food items. Regardless of where the food is consumed, increasing resource use efficiency and directing consumers to less resource-intensive food choices is essential to avoid a dramatic increase in resource consumption when the economy returns to normal. Several studies have shown changes in customers' perspectives on energy management technologies [73, 74] and improved awareness of food waste during the pandemic [30, 31]. These changes may offer opportunities to limit the system-wide increase in resource consumption caused by higher occupancy during the pandemic and support better household resource management in the future.

Three types of commercial wastes were assessed in this study: commercial municipal solid waste (MSW), RCRA hazardous waste (Resource Conservation and Recovery Act), and construction and demolition debris (CDD). RCRA is a public law regulating solid wastes in the U.S. [75], and USEEIO provides RCRA hazardous waste generation factors for all industrial sectors across the U.S. economy [39]. Overall, 8.31 Mton of commercial wastes were reduced across the life cycle of U.S. household activities during the pandemic, including reductions in commercial MSW (5.36 Mton), RCRA hazardous waste (1.81 Mton), and commercial CDD (1.14 Mton). This decrease in commercial MSW was mostly attributed to food consumption, specifically, the reduced consumption of FAFH (figure 4(a)). Closure of full-service restaurants, which were the major contributors of commercial MSW both before and during the pandemic, helped the most in commercial MSW reduction (figures 4(b) and S1). For commercial RCRA hazardous waste and CDD, reduced travel dominated the decrease (figure 1(a)), although food consumption had a larger or similar contribution to the two waste categories across the life cycle compared to transportation (figures 1(b) and (c)). Some news articles attributed the reduction in commercial waste to the closure of businesses and isolation measures [76, 77], which was consistent with the findings of this study. However, comprehensive and quantitative real-world data on waste generation in cities are lacking. The latest waste generation data by the U.S. EPA are for the year 2018 [78]. Most landfills, institutions, and recycling facilities do not report recent data on their websites [79]. Collecting location-specific waste generation data will help tailor effective waste management strategies after restaurants re-open.

All wastes assessed in this study are commercial wastes across the life cycle of household activities. Residential wastes, such as food waste, were not included due to the lack of data [80]. The U.S. EPA estimated food waste generation to be as high as 63.1 million short tons in 2018 (equivalent to 57.2 Mton), 39% of which were generated by the residential sector [81]. Assuming a linear relationship between the food waste increase and the expenditure growth of food consumption at home (11.7% increase in 2020 compared to 2018 according to the data from the USDA [43]), the increase in food wastes during the pandemic was estimated to be 2.6 Mton; that is smaller than the reduction in commercial MSW estimated in this study (8.31 Mton reduction of commercial wastes as discussed previously). When restaurants and food services re-open, temporary MSW reduction is likely to vanish unless more effective waste management is implemented for post-pandemic recovery. More data collection efforts are needed to investigate consumer behavior and food waste generation at homes and restaurants.

Other environmental impacts evaluated in this study include acidification, eutrophication, ozone depletion, and freshwater ecotoxicity. Household food consumption affected all four environmental impacts before and during the pandemic (figures 1(b) and (c)). Furthermore, changes in food consumption dominated changes in ozone depletion and freshwater ecotoxicity (figure 1(a)). Acidification potential quantifies the mid-point impacts of precursors to acid rain, such as sulfur dioxide (SO₂) and nitrogen oxides (NO_x ) [82]. During the pandemic, there was a net decrease of 226 kton of SO₂ eq across the life cycle of household activities. Reduced travel significantly decreased these precursors, which compensated for the effects of increased consumption of energy and food, particularly electricity (accounting for over 90% of life cycle acidification potential related to household direct energy consumption in tables S13 and S14) as well as meat and dairy (∼60% of life cycle acidification potential related to household food consumption in figure 4(b)).

The remaining three environmental impacts showed net increases during the pandemic, as shown in figure 1(a). Eutrophication potential quantifies the impacts of excessive nutrients in water and soil [83]. As compared to 2019, net life cycle increase was estimated to be 6 kton N eq during the pandemic. As fertilizers are primarily used for agriculture, food production plays a dominant role in the life cycle eutrophication potential for household activities (figures 1(b) and (c)). Illinova et al showed that the fertilizer industry remained strong during the pandemic; moreover, the global demand for fertilizers has increased due to the growing food consumption [84]. Higher domestic energy consumption, particularly electricity, also contributed to an increase in life cycle eutrophication potential (figure 3(b)). The life cycle eutrophication potential of different power generation technologies was assessed in previous LCA studies, and it highlighted the need to improve the current electricity generation systems [85]. For ozone depletion, the net life cycle increase was estimated to be 3.4 ton CFC-11 eq. Methyl bromide, a primary pesticide in the agriculture sector [86], was the primary contributor to life cycle ozone depletion (table S16). Across different food types, fresh vegetables accounted for approximately 70% of the total life cycle ozone depletion before and during the pandemic (figures 4(b) and S1). Freshwater ecotoxicity potential (in CTUe, comparative toxic units for ecotoxicity) estimates the potentially affected fraction of species due to pollutants emitted to the aquatic environment [83]. This study showed a net increase in freshwater ecotoxicity impact during the pandemic, largely due to an increase in food consumption. The major emissions contributing to this impact include lambda-cyhalothrin, cyfluthrin, fenpropathrin, and many others used as pesticides (table S17).

Several impacts of COVID-19 that were not addressed in this study included: (a) supply chain disruption caused by increased household consumption and activities. For example, lumber prices in the U.S. and Canada reached a record high during the pandemic due to the high demand for home repair and renovation activities, closure of non-essential industries, and shortage of workers [87, 88]; (b) rapid growth of food delivery in the pandemic [89, 90], which could offset the benefits of reduced household transportation. Future research could focus on the environmental impacts of increased construction demand and food delivery when expenditure or consumption data are available. This study investigated the life cycle impacts of U.S. household energy use in 2019 and 2020, but it did not explore the energy efficiency differences between household cooking and restaurant cooking due to the lack of disaggregated data for cooking activities. Data from the Energy Star program showed large variations in the energy efficiency of conventional oven (72%–86%) [91] and residential cooking (85% for induction cooking tops, and 32% and 75%–80% for conventional cooking tops using gas and resistance heating elements, respectively) [92]. However, there were no data on the number and sizes of different cooking appliances used in the United States, which will need future data collection efforts to inform cooking and food choices in the pandemic recovery. In addition, some exogenous changes between 2019 and 2020 (e.g. change in heating degree days) could have affected the results, which can be removed when higher-resolution data are available in the future. Additionally, the impacts of household variability, such as households in different regions or income groups can be explored when region-specific household expenditure data by demographic groups are made available.

In addition to the hybrid LCA, other modeling options include global environmental multiscale models (GEMMs) and general equilibrium models (GEMs). The GEMM focuses on environmental predictions at regional or global scales. It has been used for weather and air quality forecasting; however, it may not be a suitable tool for tracking the impact of one specific activity in one region or supply chain activities across different regions [93–95]. The GEM can cover economic activities in the whole economy. However, previous studies have been limited to carbon emissions and have not considered the full life cycle perspective [96–98]. Several studies have linked GEM with LCA. However, they have mainly focused on specific industrial sectors because of the intensive LCI data requirement [97, 99]. Overall, these models can be coupled with hybrid LCA to address the time lag issue of EEIO in future research.