Using linked government survey and administrative data, Stanford Health Policy’s Maria Polyakova documented new evidence of striking differences across demographic and socioeconomic groups in both economic and health impacts of the first year of the COVID-19 pandemic in the United States.
Overall, the pandemic was responsible for 22.3 additional all-cause deaths per 10,000 people ages 11-99 and for an average loss of jobs of 5.4 per 100 people ages 25-64 in its first year. As expected, extra deaths were concentrated among the older adults, while extra job losses were concentrated in the under-65 population.
In a new National Bureau of Economic Research working paper, Polyakova and colleagues found that within these age groups, the impacts of the pandemic on all-cause mortality and on employment varied considerably across different levels of society. But there was one notable regularity: The health and economic damages tended to be concentrated in the same demographic and socioeconomic groups, by industry and occupation, household income, race and ethnicity, and education.
Polyakova is an assistant professor of health policy and a faculty fellow at the Stanford Institute for Economic Policy Research.
“The United States has long exhibited striking variation in health and economic well-being across demographic groups, including geography, education, income, race, and ethnicity,” the authors wrote in the November 2022 paper. “The COVID-19 pandemic — which was both a health and an economic crisis — was no exception.”
Researchers found that non-white individuals and those with lower incomes or without a college degree experienced higher excess all-cause mortality as well as greater job loss during the first year of the pandemic. Perhaps surprisingly, geographically, mortality and economic damages by the end of the first year were not correlated. In fact, states that experienced high excess mortality from the pandemic did not necessarily experience higher job losses, and vice versa.
“Racial and ethnic minority groups, people without a bachelor’s degree, those with lower family incomes, and those working in service industries and jobs not suitable to working from home were impacted the most, both in terms of lower employment levels and higher excess all-cause mortality,” they wrote.
Polyakova and co-authors — Amy Finkelstein, Geoffrey Kocks & Victoria Udalova — went on to analyze what factors can account for different impacts by racial and ethnic groups as well as by educational groups. The researchers focused on the first 12 months of the pandemic in the United States — from March 2020 through February 2021 — which marked the period prior to widespread vaccination.
They considered two main classes of explanations for the differences in the pandemic’s impacts across demographic groups. One set focuses on differences in exposure to the pandemic; the other focuses on existing differences in health capital and human capital, which can affect outcomes, conditional on exposure. The researchers found that observable differences in living arrangements and the nature of work — which likely affected only exposure to the virus and to economic contractions — can explain 15% of the Hispanic-white difference in excess mortality, almost one-fourth of the non-Hispanic Black-white difference, and almost half of the difference between those with and without a college degree. These same factors can also explain 35% to 40% of the differences in economic damages between these groups.
In other words, measurable factors of exposure to the risk of the diseases or the associated economic ramifications were able to explain a substantial share of differences in who survived the pandemic and who was able to keep their job.
“These findings suggest that differences in exposure to the virus and to economic contractions played an important role in the disparate impacts of the pandemic,” the authors wrote.
At the same time, Polyakova said, “The nature of work and income were not able to explain all — or even most — of the observed differences in the pandemic’s effects, leaving an important role for other factors that we may not understand or be able to measure well.”
Such detailed analysis was possible due to the data linkage work of the U.S. Census Bureau. For the mortality analysis they used the U.S. Census Bureau’s version of the Social Security Administration’s Numerical Identification database, which provided individual-level data on the date of death. The researchers then linked these administrative records to a record of race and ethnicity from the Decennial Census and to the American Community Survey that provided information on education, occupation, industry, income and housing.
These findings underscore that public health can affect different parts of the society in dramatically different ways, and many of the reasons for such heterogeneous impacts are unrelated to medical factors.
“The findings shed light on which groups may be most vulnerable to both the health and economic shocks of a public health emergency, helping inform tradeoffs and focus of precautionary or protective policies,” Polyakova said.