April 21, 2020 12:56AM

No, Mr. President, Immigration Is Not Correlated with COVID-19 in the United States

Late on Monday night, President Trump tweeted that he will temporarily suspend all immigration to the United States to protect from the “attack from the Invisible Enemy,” COVID-19. Implicit in his tweet is that immigrants are spreading COVID-19 in the United States. The first part of this post shows that there is no relationship between COVID-19 cases or deaths at the county level and the share of the population that is foreign‐​born in each county. The second part explains our data and methods for the regression. The last part discusses the implications.

No Relationships Between COVID-19 and Immigrant Populations

The rate of COVID-19 cases and deaths by county in the United States is not correlated with the share of the population that is foreign‐​born. We ran two cross‐​sectional regressions that are graphically described in each figure below. Figure 1 shows that there is no relationship between the rate of COVID-19 cases by county and the immigrant share of the population in those counties. Figure 2 also shows no statistical relationship between the rate of deaths from COVID-19 by county and the immigrant share of the population. The numbers in Figures 1 and 2 are corrected for population density and state fixed effects (explained in the Data and Methods section below).

In both regressions, the coefficients for the foreign‐​born population share at the county level are both small in magnitude and statistically indistinguishable from zero (Table 1). In other words, the results are statistically insignificant. Rather, the most predictive variable in each model is a county’s population density, which is statistically significant at the 0.1 percent level in each regression. Since there are dense counties with disproportionately native‐​born populations and dense counties with disproportionately foreign‐​born populations, it was important in each regression to control for that variable to specifically see how immigrants were related to COVID-19 and because population density is a risk factor for the disease.

Data and Methods

The data on the county‐​level COVID-19 cases and deaths comes from The New York Times. These data list the cumulative number of COVID-19 cases and deaths by county by day and are current as of April 19, 2020. The share of the foreign‐​born population by county comes from the American Community Survey (ACS). We use the most recent 2014–2018 ACS 5‐​year pooled estimates, since the 5‐​year estimates provide estimates of the foreign‐​born population for all counties in the United States (single year estimates are only available for counties with populations of 65,000 or greater). The resulting dataset is a cross‐​section of all U.S. counties in the 50 states and the District of Columbia with a total of 3,142 counties and county equivalents.

We run simple cross‐​sectional regressions that plot the immigrant share of a county’s population on the X‐​axis and the COVID-19 death rate or case rate per 100,000 logged on the Y‐​axis. We control for county population density, as measured by the number of residents per square mile per county, because it is a key risk factor in the spread of COVID-19. Data on land area by county come from the Census Bureau 2018 Gazetteer files.

In each regression, the response variable is the natural log of the COVID-19 incidence or death rate per 100,000 population (with 1 added to avoid zeroes) and the explanatory variables are the foreign‐​born share and the log county population density. Since many states implemented mandatory stay in place orders to curtail the spread of COVID-19, we include state fixed effects to capture state‐​specific factors and compute heteroskedasticity robust standard errors that are clustered at the state level.

Discussion

The cross‐​sectional results above are correlations and not causal in nature. They are simple and should be taken with a grain of salt, especially so early in the pandemic. However, the results suggest that the share of a county’s local immigrant population does not accelerate the spread of COVID-19 across U.S. counties. Of course, we regressed a measure of the stock of immigrants in the United States and not the flow. The flow of immigrants is what President Trump’s executive order will stop (if it does anything), but those immigrants will flow to counties of the United States where immigrants already live. Since they’ve been flowing there since COVID-19 jumped to humans and there is no relationship between the immigrant share in those counties and COVID-19, the above regressions do help peel back the fog of ignorance.

Patient Zero, the first known person to have COVID-19 in the United States, returned from visiting family in Wuhan, China on January 15, 2020. He came down with a fever on January 19th and immediately went to a doctor’s office. Details about this person are not available, but since he “returned” to the United States it’s fair to say he probably didn’t get sick on his first trip here. According to the available evidence, a new immigrant didn’t bring COVID-19 to the United States.

The federal government has the legitimate power to restrict immigration or other travel across borders to limit the spread of serious contagious diseases and to protect public health. However, blanket bans like those imposed by the administration in recent months and those coming on April 21 are akin to closing the barn door after the horse has escaped. Most research on travel bans in response to pandemics finds that they don’t limit the spread of diseases, in part because they are always imposed after the disease has spread. At most, a reduction in travel from China might have delayed the onset of a pandemic for a few weeks after Trump imposed a virtual ban on travelers and immigrants from China on February 2, 2020. But further immigration bans now will have no effect.

The government should effectively use its power in cases where it can and should legitimately restrict immigration further. But Trump’s order is more for show as “the government has suspended nearly all legal immigration and travel to the United States” already, according to our colleague David J. Bier. Quarantines for new immigrants and travelers make sense. They could be carried out in either in their home countries or once they arrive here in many of the vacant hotels idling near airports. The immigrants or their sponsors could pay for the quarantines themselves and for a test when they want to enter the population. Additional health checks should be required for new immigrants in the short term.

Banning all immigration is not a serious response to the threat of COVID-19 at this stage of the pandemic. If Trump banned all immigration, travel with the rest of the world, and prevented U.S. citizens from returning except under strict quarantine in early January, he might have prevented or at least delayed the spread here, but that was long before the world knew of a serious pandemic. And remember that about 110,000 immigrants entered the United States during the more serious 1918 pandemic and the government didn’t cancel the temporary migrant worker program for Mexicans that year. During the greatest national emergency in the 20th century, the United States let in 170,952 immigrants on green cards and 227,668 bracero guest workers to work in the essential economic sector of agriculture. At a very minimum, we can allow some legal immigrants and temporary guest workers to enter and work in essential sectors of the economy if they pass rigorous health screenings. Let’s make sure immigration doesn’t threaten public health rather than banning it altogether.