Household Overcrowding Facilitates the Spread of Covid-19

By Jason Richwine, Steven A. Camarota, and Karen Zeigler on October 8, 2020

A new report from the Center finds that immigrants comprise nearly half of all workers living in overcrowded housing. In many occupations deemed essential during the pandemic, immigrant workers account for an even larger share of those in overcrowded conditions. For example, although immigrants are 32 percent of butchers and meat processors, they comprise 64 percent of all butchers and meat processors who live in overcrowded conditions.

Overcrowding is a problem for many reasons, especially its role in spreading communicable diseases such as Covid-19. Indeed, major media outlets, such as the New York Times and the Los Angeles Times, have been raising the issue of overcrowding throughout the pandemic.

As the Center's new report discusses, a number of academic studies pre-dating the pandemic found that the share of a local population living in overcrowded conditions increases the spread of respiratory viruses. As for Covid-19 specifically, a study published in August by the Journal of the American Medical Association found that overcrowding was among the factors that increased a pregnant woman's chance of testing positive in New York City. An analysis published by the Donahue Institute at the University of Massachusetts (UMDI) found that overcrowded housing was "the most indicative measure of Covid-19 spread that UMDI observed".

The purpose of this post is to supplement the research by measuring the relationship between overcrowding and Covid-19 cases on a nationwide basis. Using housing data from the American Community Survey (ACS), Figure 1 plots county-by-county positive tests per capita (vertical axis) against each county's share of households that are overcrowded (horizontal axis). Each dot represents one of 3,134 counties. The correlation of 0.23 indicates a positive relationship.


Graph: Covid-19 Correlation with Overcrowding by County


There are two reasons that the relationship is probably even more positive than what Figure 1 implies. First, ACS sample sizes are small in less-populated counties, leading to considerable measurement error in the overcrowding variable. Second, the data contain strange outliers. Most of the dots toward the bottom right of the plot (high overcrowding, low viral spread) are majority-Native American areas of Alaska, some of which are census areas, not counties. Another is entirely in the Pine Ridge Reservation in South Dakota. Moving over to the top left of the plot (low overcrowding, high viral spread), we find counties with state prisons. Since overcrowding data excludes prisons, but the virus data includes them, these counties could be skewing the results.

Figure 2 repeats the analysis using only the 975 counties with at least 50,000 people, thus eliminating both the imprecisely measured counties and the strange outliers. This causes the correlation to rise to 0.44.


Graph: Covid-19 Correlation with Overcrowding in Large Counties


Though less than one-third of all counties in the United States, these 975 large counties account for nearly nine out of 10 U.S. residents. Therefore, in the counties where measurement of overcrowding is more statistically robust, and where the vast majority of Americans live, there is a relatively strong correlation between Covid-19 positivity rates and household overcrowding.

Many other factors are associated with the spread of disease, of course, but our simple analysis here is consistent with the finding that household overcrowding helps spread Covid-19, just as it helps spread other respiratory diseases. The connection to immigration is clear. As the Center's new report shows, immigration has added significantly to overcrowding in the United States, particularly among workers thought to be essential during the current pandemic. Efforts to reduce our nation's vulnerability to future pandemics will need to address the role that immigration plays in the overcrowding problem.

Methodological Notes

The source of our county-level data on housing and population is the most recent five-year file of the American Community Survey, accessed through It covers 2014 through 2018.

Following the Census Bureau's overcrowding definition, we designate as overcrowded any domicile with more than one person per room — excluding bathrooms, porches, balconies, foyers, halls, and unfinished basements.

Cumulative Covid-19 positive tests by county come from the New York Times online database, updated through September 1, 2020. We use positive tests as a proxy for the actual cases of Covid-19 in each county. However, please note that positive tests are affected not only by the prevalence of the virus in a particular county, but also by the county's volume of testing.

Due to the way the Times collects data in certain cities (see "Geographic Exceptions" at the link above), the five counties of New York City are treated as one super-county. In Missouri, the three counties that overlap Kansas City are combined, as are the two counties that overlap Joplin.