Anthony Behar/Sipa USA via AP Images
A medical worker administers a swab to a woman at a testing site in the Queens borough of New York City, which saw an early surge of COVID-19 cases.
The pandemic is exposing the racial fault lines that divide our society. More African American and Hispanic people are hospitalized than Anglos, and more people of color die from COVID-19 in every age group as a percentage of their respective populations. The data supporting this reporting is solid, but there may be more to the story.
By working with demographic and COVID-19 statistics in New York City, a small group of researchers and I have found that income, even more than age, race, or ethnicity, might be the most significant driver of COVID-19 deaths. We make this claim cautiously, because in the United States, public-health data rarely includes income or occupation, the information required for definitive proof. To determine the impact of class on COVID-19 death rates, we have had to use indirect methods of statistical evaluation. But it still appears to show a strong correlation.
Here’s how we analyzed the 22,000 COVID-19 deaths in New York City recorded through May 15, 2020.
The New York City Department of Health tracks the number of deaths (and cases) by ZIP codes (more than 170 of them). ZIP codes in New York can be fairly broad groupings of neighborhoods that may vary widely demographically. So we looked at the 2,047 census tracts in New York City, which more closely correspond to actual neighborhoods. These tracts are surveyed by the U.S. census each year for their defining social characteristics, which number in the hundreds of categories, including income, ethnic identity, age, and race.
New York City does not provide COVID-19 death rate data for these smaller tracts, but we applied the death rate for each ZIP code to each of the census tracts within that ZIP code. This is not perfect, but in most ZIP codes there is a preponderance of one kind of neighborhood, be it richer or poorer, lighter or darker. In this way, we gain the benefit of analyzing the detailed demographic data of each neighborhood provided by the U.S. census.
We first looked at the city’s richest and poorest neighborhoods. We compared basic demographic data for the approximately 230,000 people who live in the city’s wealthiest neighborhoods (U.S. census tracts with a median income above $240,000) with the similar number of people who live in the city’s lowest-income neighborhoods (median income below $25,000). As the table below shows, the results are stark. The COVID-19 death rate in the poor neighborhoods is more than two and a half times higher than the rate in the wealthy neighborhoods.
It is very tempting to conclude from this data that the real driver of the death rate is whether or not you are Anglo, because richer neighborhoods are much whiter and have fewer deaths. With differences this dramatic—and without more income and occupational detail—one can understand why the main reported story line has been that people of color are the most at risk. But our goal was to dig deeper, to find out if it really was race and ethnicity that drove the difference, or if income played a part, too.
To find an answer, we relied on a basic process taught in introductory statistics classes—multiple regression. This tool allows us to ask of the data, “What is the relative contribution of each causal factor, if all the other significant ones don’t change?” Our goal was not only to figure out which demographic variables best accounted for the death rate, but also to see how much each variable contributed to it.
We tested many different variables from among the hundreds that the Census Bureau collects, ranging from “not a citizen” to “average household size” to “speaks English less well” to “income below the poverty line” to “households with a broadband internet subscription.” In the end, we found that the death rate in New York City by neighborhood was best explained by: median income, the percentage of residents 65 and older, the percentage of African Americans, the percentage of those born in Latin America, and the percentage of those living in apartments with more than 1.5 people per room.
Each of these five factors has had a powerful and significant impact on the COVID-19 death rate (measured per 100,000 residents). But some have more influence than others.
Neighborhoods with approximately one-third more African Americans than the average NYC neighborhood have nine more deaths per 100,000, making the average death rate jump from 201 per 100,000 to 210. If the percent of crowded housing increased by about a third, the death rate also increased by about nine per 100,000. Being born in Latin America, a category that includes many undocumented workers, was associated with twice the risk of dying from COVID-19 than that faced by African Americans and those who lived in crowded housing. This is likely because it is far more difficult for undocumented workers, even essential ones, to gain access to medical and financial assistance.
Being old, of course, is a major risk factor no matter your ethnicity, place of origin, or income.
But income alone, a key indicator of class, was the most influential characteristic. Lower-income neighborhoods saw an addition of nearly 28 deaths per 100,000, increasing the average death rate by more than 10 percent, from 201 deaths per 100,000 to 229. Again, these factors are independent of each other. A low-income elderly Black person who was born in Latin America and lives in a crowded New York apartment would be in great danger of dying from COVID-19.
The data suggest that as the virus spreads, those who because of the nature of their jobs and their financial circumstances have to work outside of the home, and their families, are more likely to become infected and die. A disproportionate number of these low-wage workers are people of color, but by no means exclusively so. The common denominator isn’t only skin color. It is also income and class.
While it’s tempting to blame the current outbreaks across the U.S. on kids partying with reckless abandon, it should be noted that the disease is spreading as states open up their economies. Low-income workers who can’t earn a living while isolated at home are spending their workdays around each other and the public. Without access to income and occupational case data, we can’t know for sure how much of the spread of COVID-19 is due to work, or socializing, or both. But the NYC data strongly suggests that people who must leave the home to work face extra risk of infection and death.
These data also confirm what we already knew was true. We are NOT all in this together. The richer you are, the easier it is to protect yourself. If you have enough money, you don’t have to work in a meatpacking factory or a warehouse or an assembly line or a nursing home. If you are rich enough, you can buy isolation, instead of being forced into the fray.
Clearly our study is limited. (You can read a technical note and see all of our data here.) We did our best to go where the data led us. But for more definitive studies, it is necessary for public agencies to collect income, occupation, and other specific class-oriented statistics the moment a person has tested positive for coronavirus.
Public-health officials, journalists, and independent researchers need precise information on income and occupation to better understand the spread of COVID-19 and how to combat its social effects. Lives depend on it.
This report is produced by RunawayInequality.org. Special thanks to Professor Michael Parkin, Oberlin College, for reviewing the statistical analysis; Peter Kreutzer for extensive statistical support and editing; and Kris Raab, Elizabeth Royte, and Sharon Szymanski for additional editorial support.
