COVID-19 has been difficult for everyone, but much harder for some. A new survey analyzed by the Federal Reserve Bank of Minneapolis highlights the pandemic’s health and economic toll, and reveals stark disparities in how Americans have been affected by the pandemic.
There have been “dramatic declines in well-being,” and some areas and groups have suffered far more than others. “Hispanic, younger, and lower-earning individuals all faced disproportionately worsening economic conditions, as did those with school-aged children.” Equally troubling, some respondents exhibiting symptoms or exposed to infected family members appear not to be taking precautions to protect themselves or others.
These conclusions are found in a May 2020 working paper authored by Abigail Wozniak, director of the Opportunity & Inclusive Growth Institute at the Minneapolis Fed. Wozniak notes a number of unexpected demographic disparities and points out geographic variations that are not well explained by location characteristics either before or during the pandemic.
“Disparities and Mitigation Behavior during COVID-19,” published by the Institute as part of its working paper series, analyzes data from the COVID Impact Survey (CIS), an ongoing survey first conducted in April 2020. The CIS was administered to a large nationwide sample, allowing estimation at the national level, but also for specific geographic locations including 10 states and eight metropolitan areas. It assesses a broad spectrum of well-being measures, asking respondents about their employment status, hours of work, social connection, food insecurity, and mental health.
The CIS is unique in that it gauges these broad well-being outcomes at local geographic levels and relates outcomes to COVID-19 exposure and behavioral changes. The working paper therefore contributes “new knowledge about how COVID exposure, personal risk, and risk to others relate to the massive changes in employment and broader well-being that followed the onset of the COVID-19 pandemic,” writes Wozniak.
To gauge changes over time as well as variation among locations and demographic groups, Wozniak compares answers to the April CIS questions against responses to earlier national surveys on which CIS questions were modeled. She finds that “nationally and for each of the 18 subnational sampled places in the CIS, nearly every measure of well-being that I study has deteriorated markedly.”
Nonetheless, Wozniak finds, location outcomes vary significantly. Some places have suffered far more economic fallout from the pandemic than others. This variation is not easily explained by location characteristics prior to COVID-19, nor by policy responses or behavioral change after the start of the pandemic.
The results lead Wozniak to conclude that place-based relief policies make less sense than targeting relief to individuals based on characteristics like family structure, earnings, employment, race/ethnicity, and age, as well as to individuals directly exposed to COVID-19. She recommends financial assistance to aid those recovering from the disease and to encourage infcted individuals to self-quarantine.
Geographic differences
To capture well-being on multiple dimensions, the working paper analyzes five outcomes—employment rate, hours worked, social interaction, food insecurity, and mental health. The initial analysis examines changes in well-being since prepandemic surveys and measures variation among areas. All locations—nation, state, and metro—experienced significant drops in employment levels since 2019 and marked declines in hours worked for those still employed. Food insecurity rose everywhere (see Figure 1), and respondents in all states reported decreased ability to cover emergency $400 expenditures.
Measures of mental health declined in all areas, although regular communication with friends, family, and neighbors rose substantially, a large improvement in social connectedness.
Locations differed significantly in measures of economic security. Texas and Birmingham saw large declines in hours worked, while Minnesota and Phoenix experienced no such change. Pittsburgh had a 10 percent increase in food insecurity, while Phoenix food insecurity rose by 29 percentage points. States varied considerably in respondent ability to cover emergency expenses.
What explains such wide spatial variation? Statistical analysis reveals that behavioral changes such as wearing masks and mandated restrictions had little relationship to differences among places in well-being. Nor is variation explained by characteristics of these places prior to the pandemic—such as median household income or income inequality.
Wozniak suggests that economic impacts of the pandemic are perhaps better predicted by variation in sectoral linkages. Blocked supply chains or demand links are likely to affect locations differently, resulting in substantial economic damage in some areas and minor impact elsewhere.
Demographic differences
Effective targeting of government aid calls for identification of those most in need. Demographic analysis of CIS data can help identify which segments of each location’s population will benefit most from assistance. Local and state policymakers can use location-specific CIS data to gauge how their populations are faring and distribute aid accordingly. The sharp rise in Phoenix food insecurity, for example, highlights a growing emergency for Arizona officials to address.
While all demographic groups experienced employment declines and fewer working hours, employment drops were largest for younger workers, nonwhites, and individuals with children. (See Figure 2.) Controlling for other characteristics, older and higher-earning respondents were less likely to be furloughed or unemployed, more likely to be working longer hours, and less likely to be seeking food aid.
Hispanics and individuals with school-aged children were more likely to experience job separation and to be working fewer hours. The data are less clear on Asian respondents, but suggest that they too may have experienced a larger drop in hours worked. Contrary to expectations, the data suggest that declines in working hours have been similar for black and white respondents.
Personal exposure and behavior
The working paper also analyzes the relationship between exposure to COVID-19 and behavioral changes. In short, have people with COVID-19 symptoms or with exposure to infected individuals taken steps to safeguard their health or that of others? The answer appears to be no.
Rather, the data suggest that those in households with COVID-19 infection have been less likely than others to leave work or reduce their working hours. Moreover, high-risk individuals have reportedly been no more likely to leave work or reduce hours than less-vulnerable population groups, nor have they been more likely than others to take additional protective measures. Similarly, respondents reporting recent fevers have been no more likely than others to work less or take avoidance measures. The data indicate higher levels of mental distress among such individuals, suggesting that “respondents are not unaware of the risks they face (or pose),” observes Wozniak.
Conclusion
While future results may reveal a changing story, analysis of the first wave of data from the CIS survey indicates numerous areas of concern and several courses of action.
All geographic areas and demographic groups have suffered significant declines in economic, physical, and mental well-being, but variation is high among both locations and groups.
Geographic variation appears to be driven not by local policies, behaviors, or characteristics, but by sectoral linkages that affect some locations more than others.
Loss of jobs and working hours in the wake of the pandemic is experienced disproportionately by Hispanics, possibly by Asians, and by young workers and those in households with school-age children. Older, higher-earning workers have suffered less. Blacks and whites have experienced similar declines in employment and hours worked.
Cautioning against place-based policies, Wozniak instead recommends relief policies targeted to vulnerable individuals identified by family structure, income prior to the pandemic, and other characteristics such as race/ethnicity, age, and employment sector. She further suggests incentive payments or similar policies to motivate those directly affected by COVID-19 to engage in self-quarantine and other mitigating behaviors.