Unequal household impacts of the COVID-19 pandemic

We are fascinated with human behavior here on Consumer Corner. And as terrible an upset  as the COVID-19 pandemic was – in the moment and ever since – there is a lot to be learned from careful study of how behaviors adapt in response to shocks.

We started gardening and worried about food supply chains:

• “Mary, Mary with COVID so Scary, How Does your Garden Grow” by Courtney Bir
• “Introducing the Purdue Food and Agriculture Vulnerability Index for COVID-19” by Jayson Lusk

Then we dug into behavioral adaptations and the sticky (but somewhat odd) shopping behaviors we were developing:

• “Straight Talk About Consumer Behavior Under Duress” by Nicole Olynk Widmar
• “Human Behavior, Beliefs and Practices in the COVID-19 Era” by Nicole Olynk Widmar and Dr. Courtney Bir
• “We’re Talking About Toilet Paper & Meat Again” by Nicole Olynk Widmar and Dr. Courtney Bir
• “Consumer Spending is Today’s Statistic; Consumer Behavior is Far Longer Lasting” by Nicole Olynk Widmar and Torrie Sheridan

We worried about the uptick in alcohol consumption:

• “Societal Values & Personal Risky Behaviors Unchanged in 6 Months of Pandemic Era, Except We’re Drinking More Alcohol” by Nicole Olynk Widmar and Dr. Courtney Bir

And tried to broach conversations about the impacts on kids:

• “Internet “Have Nots” (Including Many Children) are Suffering Digital Divide Consequences in the COVID-19 Era” by Nicole Olynk Widmar and Dr. John Lai

But our societal understanding of the impacts of the pandemic era on children is still evolving. We are seeing kids born in the acute phases of the pandemic now starting school – and those who navigated high school online at times moving through college. Different generations have always been shaped by the experiences they collectively experienced; this will be no different. The difference is that we don’t yet know what shape these behaviors will take.

Years ago, we shared insights from data collected during the early, acute phases of the pandemic. We have since returned to those datasets to dive deeper into questions about households with children and their experiences during COVID-19. Specifically, we seek to better understand COVID-19’s impacts on U.S. households, with particular interest in differences between households with children and those without. We are motivated by the possibility of disproportionate and persistent impacts of the pandemic and post-pandemic period on women, individuals working in certain industries, and caretakers.

Households with children faced unique challenges, magnified during the COVID-19 pandemic, as they balanced household economics alongside caregiving and child education. A June 2020 survey examined household composition, pandemic impacts (including mask wearing), and compared challenges faced by households with children to those without. The survey yielded 1,198 responses: 347 from households with children (under 18) and 851 from households without children. Across all household activities studied, mean self-reported pandemic impacts were higher for households with children than for those without. Having a child in the household or being female increased the reported impact on respondents’ ability to find meat, milk, and other perishable grocery items. If the respondent was female, the probability of taking on schooling activities for a child increased, supporting concerns about caregiver strain and potential employment and economic consequences during a pivotal time in the pandemic.

Among households with children, 20% reported being unable to access their usual childcare due to the pandemic. In addition, 53% took on schooling responsibilities and 28% had to cut or eliminate work hours. These findings underscore disparities in caregiving roles based on demographics and emphasize the need for informed decision-making in social, health, and support systems to address these inequities. 

As public institutions, including schools, closed in response to the rapidly spreading COVID-19 virus, many parents and caretakers grappled with balancing caregiving and educational activities alongside employment and household financial stability. These impacts of pandemic-related disruptions were not equally experienced across households. Likewise, they were not equally experienced within households. Food insecurity during the COVID-19 pandemic in 2020 was found to be worse in households with children than in households without children (Bir et al., 2024). Food-insecure families were challenged during the pandemic as schools, social networks, and charitable food sources (e.g., food pantries) were strained or inaccessible (Morales et al., 2021).

While these experiences are not universal and many household and individual nuances must be acknowledged, there are clear implications for economic and social household functioning when policies fail to acknowledge unequal or unintended consequences. Obeng et al. (2022) conclude that the COVID-19 pandemic led to gendered consequences for working women; without childcare or other support systems to alleviate conflict between workforce and household duties, this may lead to economic vulnerability for working mothers.

Data collection took place during the beginning of the relaxation of social distancing in the United States, from June 12 to June 20, 2020. A total of 1,198 completed responses were obtained from individuals over the age of 18. Of these, 347 (29%) reported having children in the household, while 851 (71%) reported no children under the age of 18 living in the household at the time of data collection. This survey was designed so the respondent sample matched U.S. Census proportions for sex, age, education, income, and region of residence based on the most recent data available at the time (U.S. Census, 2016). This analysis focuses on potential differences in the impact of COVID-19 on daily life and beliefs surrounding face masks between those who indicated children in the household and those who did not.

Because states did not experience the same level of COVID-19 cases or impacts, they were grouped using three criteria at the time of data collection:

1. States with more than 40,001cases
2. Top 10 states by cases per capita
3. Top 6 states with rapid post-Memorial Day increases in cases.

As of June 17, 2020, 17 states had over 40,001 cases of COVID-19: California, Connecticut, Florida, Georgia, Illinois, Indiana, Louisiana, Maryland, Massachusetts, Michigan, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Texas, and Virginia (CDC, 2020a). To facilitate comparisons across states with varying population sizes, case counts as of June 17, 2020, were divided by estimated 2019 population to generate per-capita adjusted case levels (U.S. Census Bureau, 2016). The top 10 states by cases per-capita were Connecticut, Delaware, the District of Columbia, Illinois, Louisiana, Maryland, Massachusetts, New Jersey, New York, and Rhode Island. Following reopening plans and the post-Memorial Day weekend, six states recorded numbers of new cases: Arizona, Florida, Nevada, Oklahoma, Oregon, and Texas (CBS News, 2020).

In addition to demographics, respondents indicated agreement (1 = strongly disagree to 5 = strongly agree) with:

• Someone in my household, or that I frequently spend time with, is at higher risk of complications of COVID-19.
• I am in the higher-risk group for complications of COVID-19.

Mean responses were calculated for households with children and those without, and t-tests were used to compare groups.                    

To gauge beliefs surrounding mask wearing, respondents indicated beliefs for multiple statements, including: masks have a potential role in U.S. society related to the spread of viral disease; wearing a mask helps prevent the spread of COVID-19; wearing a mask helps prevent me from getting COVID-19; wearing a mask helps prevent me from spreading COVID-19; wearing a mask will help prevent future lockdowns in my community; there is social pressure in my community to wear a mask; wearing a mask does not prevent the spread of COVID-19; and wearing a mask has negative health consequences for the mask wearer. Tests of proportions were used to statistically compare the percentage of respondents with children in their household and those without who believed in the statements.

Respondents also indicated COVID-19’s impact on five areas of daily life using a scale from 1 (not impacted) to 5 (impacted), with a “does not apply to me” option. Activities included:

• daily activities outside of work/school,
• ability to buy paper products (e.g., toilet paper, paper towels),
• ability to find meat, milk, and perishable grocery items,
• ability to execute travel plans, and
• activities related to respondent’s work/school.

Although not explicitly stated, it was assumed impacts were not positive. There may have been instances where people experienced positive impacts from COVID related shutdowns and changes; however, those people may have selected “does not apply to me.” The percent of respondents who selected each option was determined for those who indicated having children in the household and those that did not. After removing “does not apply” responses for a given activity, mean impact levels were calculated and compared using t-tests.

A series of ordinary least squares (OLS) regressions were employed to evaluate the relationship between demographics, mask-related beliefs, and impact levels. Given impact level L activity i and respondent n the equation can be given as:

L_in = b1*Kids_in + b2*Female_in + b3*Age_in + b4*Income_in + b5*HighCase_in + b6*HighCap_in + b7*HighIncrease_in + b8*YesMasks_in + e_in

where Kids_in is a dummy variable indicating whether the respondent reported children in the household; Female_in is a dummy variable indicating whether the respondent selected female or male; Age_in is a continuous variable ranging from 1 (18-24) to 6 (65+); Income_in is a continuous variable ranging from 1 (income of $0-$24,999) to 5 (income of $100,000+); HighCase_in is a dummy variable indicating residence in a state with more than 401,000 cases of COVID-19; HighCap_in indicates residence in a state with high per-capita COVID-19 cases; HighIncresase_in indicates residence in a state that experienced a spike in COVID-19 cases post-Memorial Day 2020 and YesMasks_in indicates the respondent believed masks had a role in U.S. society related to the spread of COVID-19. The error term is represented by ε_in.

Respondents reporting a child in the household were also asked whether four additional statements applied due to COVID-19:

• My child was no longer able to attend daycare or stay with a family member for childcare.
• I had to take on schooling activities for my child.
• I was not able to continue working, or had to cut back on work hours, due to childcare responsibilities.
• My childcare and educational routine did not change due to COVID-19.

A multivariate probit model (Cappellari and Jenkins, 2003) was used to estimate relationships between these outcomes and demographics.

Demographics of respondents who indicated there were children in the household differed statistically from those who did not for several categories (Table 1). Fifty-eight percent of respondents from households with children were female. Respondents reporting children in the household were between 25 and 54 years of age. For the statement I am in the higher risk group for complications of COVID-19, respondents with children reported lower perceived risk (2.571) than those without children (2.976).

^ΨIndicates the percentage or mean of respondents who had children in their household and those that did not is statistically different at the <0.05 level

¹Indicated on a scale from 1 (strongly disagree) to 5 (strongly agree).

For the statement masks have some potential role in U.S. society related to the spread of viral disease including COVID-19, there was no statistical difference between respondents with children (82%) and those that without (84%) (Table 2). A lower percentage of respondents with children agreed that wearing a mask helps prevent the spread of COVID-19 and helps prevent them from spreading COVID-19. Additionally, a lower percentage of respondents with children agreed that masks would help prevent future community lockdowns related to COVID-19 (41% versus 50%). A higher percentage of respondents with children (17%) agreed wearing a mask has negative health consequences for the wearer when compared to those without children (11%).

Table 3.2. Comparison between the percentage of respondents with and without children in the household who believe the following statements regarding masks (percentage of respondents)

ΨIndicates the percentage of respondents with children in their household that selected yes to that statement is statistically different from the percentage of people who did not indicate there were children in the household and said yes to that statement. <0.05 level

For all household activities studied, the mean score on a scale from 1 (not impacted) to 5 (impacted) was statistically higher for those with children in the household when compared to those without (Table 3). For both groups of respondents, with and without children, the ability to find meat, milk and perishable grocery items had the lowest mean impact score.

Table 3.3. Mean impact level of COVID-19 on daily life (where 1 was not impacted and 5 was impacted) for all those who did not select “does not apply to me” in response to impact; reported for those who reported children in the household (Kids N=347) and those who did not report children in the household.

^ΨIndicates the mean is statistically different between those who have children and do not have children for that activity at the <0.05 level.

Considering the OLS model for impacts on daily activities outside of work or school, having a child increased the impact score by 0.375 (Table 4). Being female and believing masks have a role in society to prevent the spread of COVID-19 increased scores by 0.350 and 0.532, respectively. As age increased, the impact score decreased. As income increased, the score increased for daily activities outside of work or school.
For the model of ability to buy paper products, having a child increased the impact score by 0.272; being female increased the score by 0.284; and believing masks have a role increased the score by 0.455. For the model of ability to find meat, milk, and perishable grocery items, having a child increased the impact score by 0.360; being female increased the score by 0.273; and believing masks have a role increased the score by 0.335. Again, the score decreased as age increased. For ability to execute travel plans, the impact score increased with income and for those who believed masks have a role in society. For activities related to work or school, having children (0.474), being female (0.235), living in a high-case state (0.308), and believing masks had a role (0.484) all increased the impact score. The impact score also increased with income and decreased with age.

Table 3.4. Ordinary Least Squares Model of the Impact of COVID-19 on Respondents Activities on a Scale of 1 (not impacted) 5 (impacted). Respondents Who Indicated the Activity Applied to Them, N Given in Table.

Twenty percent of respondents with children (n = 347) indicated their child was no longer able to access childcare. Additionally, 53% reported taking on schooling activities for their child, and 28% were not able to continue working or had to cut back on hours due to childcare responsibilities. Finally, 21% indicated their childcare and educational routine did not change due to COVID-19. More women than men reported having taken on schooling activities for children, although both sexes reported impacts from loss of childcare and schooling.

Respondents in households with children reported lower agreement with statements that they were at high risk for COVID-19 complications. This may be partially explained by the younger ages of household members.

At the time of data collection (June 2020), there was less agreement in households with children that wearing a mask helps prevent the spread of COVID-19, helps prevent the wearer from spreading the disease, and helps prevent future community lockdowns. Lower agreement that masks could help prevent future lockdowns is puzzling. Possible explanations include pessimism associated with school opening plans or prolonged periods without childcare options. It is also possible that parents’ lived consequences during the ongoing pandemic shaped beliefs about whether individual actions could yield broader societal outcomes.  

The proportion of respondents with and without kids in the household felt there was social pressure to wear masks was statistically equivalent (33% and 30%), as was agreement that wearing a mask does not prevent the spread of COVID-19 (16% and 13%). Very few negative health consequences for mask wearers have been reported by medical professionals (Marfin, 2020). The higher proportion (17% versus 11%) of respondents with children reporting negative health outcomes may reflect concerns about masks on very young children (under 2 years of age) or on children unable to remove a mask without assistance, for whom masks were not recommended by the CDC at the time (CDC, 2020b).

Higher self-reported impacts for all activities among households with children suggests heightened stress in households with kids during the pandemic. Activities outside of work or school and travel are highly correlated with household income and other demographics. Older respondents may have different shopping behaviors, for reasons ranging from intentional reductions in trips into public spaces (Miller, 2020) to shopping behaviors formed prior to 2020. Higher-income households may have greater ability to fund large bulk purchases, which can be out of reach for households unable to buy ahead. Constrained cash flow necessitates smaller scale purchases and thus a higher probability of difficulty finding items during peak demand periods. Regardless, the presence of children was statistically significant in explaining higher reported impacts across the diverse set of outcomes investigated.

Over half of respondents with children took on schooling activities, and 20% reported loss of daycare or family childcare. According to the Office of Administration for Children and Families, in spring of 2020, 63% of childcare centers and 27% of family childcare homes were closed (Lin and McDoniel, 2023). Twenty-eight percent of respondents reduced working hours or stopped working entirely due to childcare responsibilities. The probability that a child could no longer attend daycare or stay with a family member increased with household income, potentially reflecting greater reliance on paid childcare pre-pandemic among higher-income households. Childcare can cost $10,000 or more per year when it can even be found at all (Rexrode and Weber, 2020). Many childcare businesses did not survive the early pandemic closings, and those that did faced additional cleaning and protective equipment expenses (Rexrode and Weber, 2020).

Although child age was not collected, the probability that childcare was impacted decreased with respondent age – perhaps reflecting older respondents with older children and fewer childcare needs. Supporting this hypothesis, the probability of taking on schooling activities increased as respondent age increased, potentially reflecting time allocation toward schooling for older children. Women reported taking on schooling responsibilities more often than men after shutdowns induced by COVID-19. Not all cases involved reducing or terminating employment in order to take on educational activities; many respondents reported juggling remote work alongside childcare and shifting or reducing hours.

Households with children reported higher impacts across household activities and procurement of essential goods. Fundamentally, two distinct conversations remain important for public policy. First, documented differences exist between households with and without children. Second, the impacts of the pandemic differed across respondent demographics; in particular, being female increased the probability that a respondent took on schooling responsibilities for children.

Taken together, these findings highlight differences in impacts for households with children relative to those without, while also underscoring that impacts within any given household are not uniform across household members. This analysis provides empirical support for widely discussed concerns about the disproportionate impacts of COVID-19 adjustments on women, including reported impacts on everyday activities and the taking on of caregiving activities. Loss of childcare and educational opportunities alongside lost professional opportunities for caretakers, may generate effects that take years to materialize in measurable ways. These impacts cannot be ignored.