Did the COVID-19 Pandemic Cause an Urban Exodus?
- Date Posted:
- Is Database:
- Database
The COVID-19 pandemic doubled monthly net out-migration from large cities to 56k/month in 2020, compared to 28k in 2017-2018, led by fewer new arrivals rather than more departures. Higher-income renters saw a 210% jump in net outflow, while major metros with over 5m residents were most affected.
Interesting analysis from Cleveland Fed on the pandemic's impact on urban migration. Net there has been outmigration from cities, largely by younger and higher income renters. However this effect has been driven primarily because of a decline in people moving to the cities (which may mean revert once pandemic is over.)
Core finding, “…As measured by net migration to and from urban neighborhoods, 2020 does appear to have seen an urban exodus relative to the migration flows during the preceding 10 years (Figure 1). The net flow of people out of US urban neighborhoods averaged nearly 28,000 people per month in March through September of the recent years 2017 to 2019. That number about doubled—to 56,000 people per month—in 2020 after the pandemic’s onset in March. Figure 2 shows that the decline of in-migration is actually the larger driver. The average of monthly out-migration was 276,000 in March through September 2020. This is 10,000 more than its average of 266,000 for the same months in 2017 through 2019. The larger change was for in-migration, which fell 18,000—from an average of 238,000 in 2017 to 2019—to 220,000 in 2020. Figure 2 shows the four-quarter moving average of the estimated flows to smooth out seasonality. Doing so also balances the dip and spike that occurred because the pandemic delayed many moves from the spring into the summer….”
For tracts that have a majority pre-WWII housing, I also require that they have a population density of at least 2,000 people per square mile. This avoids including exurban edges where old rural homes have not yet been outnumbered by homes in new subdivisions. I selected the 7,000 people per square mile threshold because it was the average density of major cities in the pre-WWII period.
Brevoort, Kenneth P., Philipp Grimm, and Michelle Kambara. 2016. “Credit Invisibles and the Unscored.” Cityscape: A Journal of Policy Development and Research. 18 (2).https://www.huduser.gov/portal/periodicals/cityscpe/vol18num2/ch1.pdf.
Cullen, Julie Berry, and Steven D. Levitt. 1999. “Crime, Urban Flight, and the Consequences for Cities.” The Review of Economics and Statistics. 81(2): 159-169.https://doi.org/10.1162/003465399558030.
Did the COVID-19 Pandemic Cause an Urban Exodus?
One constant through the upheavals of 2020 was the steady stream of media reports about residents’ fleeing dense urban areas.1 In this data brief, I use the Federal Reserve Bank of New York/Equifax Consumer Credit Panel (CCP) and find that migration flows were in fact very unfavorable for urban neighborhoods in 2020. However, people’s taking flight from urban areas is only part of the story.
Initially, the urban exodus stories reported that people were afraid of contracting the novel coronavirus in elevators and subways. Then the narratives suggested remote work had freed office workers from long commutes, allowing them to relocate. With both remote workers and students at home full time, a desire for home offices purportedly rose, and low interest rates made buying a larger suburban home attractive. Urban amenities such as restaurants and theaters were shuttered. Later, the protests sparked by the killing of George Floyd and others were cited for motivating some urban residents to leave. Most major cities experienced increases in violent crime during 2020, and crime rates have historically predicted migration changes.2 The proposed and enacted cuts to police funding were also cited as a reason to leave by some people who feared that crime would increase further.
The media reports have used a variety of data sources to measure the urban exodus, ranging from real estate listings to the hiring of movers and renting of moving vans. For this District Data Brief, I use the CCP, which is arguably a more accurate source for measuring migration. The CCP is a nationally representative anonymous random sample of 5 percent of US consumers with a credit file, resulting in a sample of more than 10 million adults. Approximately 9 of 10 adults have credit records, so CCP-based estimates are more representative than measures that reflect only homebuyers or only clients of a specific company. The populations the CCP underrepresents include low-income people who do not use traditional lenders (for example, banks and credit cards) and college-aged adults whose parents are still handling their finances.3 The techniques and definitions I use in this brief are the same as those I detailed in two 2019 Economic Commentaries (Whitaker, 2019a and 2019b).
Lenders report their borrowers’ current addresses to Equifax each month, and the CCP reports each borrower’s census tract. Using census tracts allows us to observe moves into and out of urban neighborhoods with much more precision than the many data analyses that rely on county-level migration aggregates. Almost all counties include both high-density and low-density neighborhoods. In the CCP, we can observe urban-suburban transitions that do not cross a county line, and we can distinguish those transitions from suburb to suburb moves that do cross a county line. I define urban neighborhoods as census tracts in metro areas with populations of more than 500,000 that have either a population density of more than 7,000 people per square mile or the majority of their housing stock built before World War II (WWII).4 Having either of these characteristics makes it more likely the neighborhood will be recognizable as urban. Neighborhoods built before WWII were designed for pedestrians, so residential, commercial, and retail buildings are within walking distance. Very-high-density neighborhoods built after WWII can also offer an urban lifestyle if they can support walkable retail and high-frequency public transit. In nonurban neighborhoods, zoning usually separates residences from businesses, so almost all trips require a car.
As measured by net migration to and from urban neighborhoods, 2020 does appear to have seen an urban exodus relative to the migration flows during the preceding 10 years (Figure 1). The net flow of people out of US urban neighborhoods averaged nearly 28,000 people per month in March through September of the recent years 2017 to 2019. That number about doubled—to 56,000 people per month—in 2020 after the pandemic’s onset in March.

An increase in the net out-migration from urban neighborhoods can be driven either by an increase in people moving out or by a decrease in people moving in. Figure 2 shows that the decline of in-migration is actually the larger driver. The average of monthly out-migration was 276,000 in March through September 2020. This is 10,000 more than its average of 266,000 for the same months in 2017 through 2019. The larger change was for in-migration, which fell 18,000—from an average of 238,000 in 2017 to 2019—to 220,000 in 2020. Figure 2 shows the four-quarter moving average of the estimated flows to smooth out seasonality. Doing so also balances the dip and spike that occurred because the pandemic delayed many moves from the spring into the summer.

We can examine the net out-migration from urban neighborhoods several ways. Figure 3 shows the net out-migration by neighborhood income, renter or homebuyer status, resident age, and population. For neighborhood income, I divide the flows by whether the urban neighborhood had above- or below-median household income. In most years, the net flows out of below-median-income neighborhoods are larger than the net flows out of above-median-income urban neighborhoods, but this pattern reversed in 2020. The net outflow of people who are purchasing a home at their destination neighborhood increased 62 percent, but the net outflow of those renting at their destination increased by 210 percent. By age category, young adults exhibited the largest increase in net outflows. The most populous metro areas, those with more than 5 million residents, saw larger increases in net outflows than did smaller metro areas.

Figure 4 shows the difference in the urban outflow from 2020 over the average outflow from 2017, 2018, and 2019 for the months of April through September. The differences are plotted over four measures suggested to influence out-migration: deaths from COVID-19 in the metro area, the share of the labor force that can work remotely, small-business closures, and an increase in homicides in the central city. The details of how each of these is measured can be found in the appendix.
The correlations run in the direction we would expect for three of the relationships. Net out-migration was greater for metro areas that had more deaths from COVID-19 and those with more telework-capable occupations. Metro areas that had more small businesses remain open experienced less net out-migration. The increases in homicides in the metro area’s central city was not associated with increased net out-migration.

The final proposed cause of the urban exodus, the protests and civil unrest, is more difficult to measure. While protests occurred in every major city in the country, some were considerably more destructive or long lasting. Figure 5 shows the net out-migration series for six metro areas that had very high-profile civil unrest, and the figures for other metros with populations of more than 1 million are included in the appendix. The red vertical lines in the graphs mark the beginning of the protests in May. In Chicago, New York, Philadelphia, and Seattle, the most recent net out-migration estimates are more than twice as high as the average during the preceding three years.

Among major cities in the Cleveland Fed’s District, Pittsburgh and Columbus saw increases in net out-migration in 2020 (Figure 6). For Columbus, the decline in inflows accounted for about two-thirds of the net change. Pittsburgh’s gross outflows were almost unchanged, while its 2020 inflows fell by 27 people per 100,000. In the other four of the six largest metro areas in the Fourth District, 2020 was not remarkably different from the recent past.

Conclusion
Did the COVID-19 pandemic start an urban exodus? The estimates presented here strongly suggest that migration flows were unfavorable for urban neighborhoods during 2020. However, the word “exodus” emphasizes that migrants are leaving a particular area in large numbers. If we look at the gross flows underlying the trends in Figures 3 through 6 (available in the appendix), we can see that the declines of in-migration are almost always greater than the increases in out-migration. Out-migration did increase in many urban neighborhoods, but the magnitudes probably would not fit most definitions of an exodus. What is certain is that hundreds of thousands of people who would have moved into an urban neighborhood in a typical year were unwilling or unable to do so in 2020. These people may be harder to identify, label, and interview, but they may be best positioned to tell the real story.
Footnotes
In the appendix, I provide 16 examples of these articles.
Stephan Whitaker, "Did the COVID-19 Pandemic Cause an Urban Exodus?," Federal Reserve Bank Of Cleveland, February 5, 2021, https://www.clevelandfed.org/newsroom-and-events/publications/cfed-district-data-briefs/cfddb-20210205-did-the-covid-19-pandemic-cause-an-urban-exodus.aspx
Takeaway, “….Did the COVID-19 pandemic start an urban exodus?The estimates presented here strongly suggest that migration flows were unfavorable for urban neighborhoods during 2020.However, the word “exodus” emphasizes that migrants are leaving a particular area in large numbers. If we look at the gross flows underlying the trends in Figures 3 through 6 (available in the appendix),we can see that the declines of in-migration are almost always greater than the increases in out-migration. Out-migration did increase in many urban neighborhoods, but the magnitudes probably would not fit most definitions of an exodus….”
The out migration was driven by younger, richer, renters, “…Figure 3 shows the net out-migration by neighborhood income, renter or homebuyer status, resident age, and population. For neighborhood income, I divide the flows by whether the urban neighborhood had above- or below-median household income. In most years, the net flows out of below-median-income neighborhoods are larger than the net flows out of above-median-income urban neighborhoods, but this pattern reversed in 2020. The net outflow of people who are purchasing a home at their destination neighborhood increased 62 percent, but the net outflow of those renting at their destination increased by 210 percent. By age category, young adults exhibited the largest increase in net outflows. The most populous metro areas, those with more than 5 million residents, saw larger increases in net outflows than did smaller metro areas…”


















Ed Comment:Add how the pandemic broke the valley’s hold on tech jobs. I think this will accelerate something I fear—the offshoring of tech jobs abroad, especially the most technical jobs. That’s going to lead to the more entrepreneurialism abroad or the most valuable kind. The more our tech giants can zoom in the talent and breakthrough, the less they will care about US working. There will be tech hubs in every country attracting their best talent.