‘Snapshots’ of Migrants in Mexico Suggest U.S. Undocumented Population Is Much Larger than Previous Estimates
The population of undocumented immigrants in the United States is generally estimated to be around 11 million. But a new study from Yale SOM’s Edward Kaplan and Scott Rodilitz, making use of data on migrants who have returned to Mexico, suggests that the real number is an estimated 19.6 million, consistent with previous research by Kaplan using different methodology.
The U.S.-Mexico border in Southern California in April 2019. Photo: Mario Tama/Getty Images.
A new study by Yale SOM’s Edward Kaplan and Scott Rodilitz, based on data from a long-running survey of migrants who have returned from the United States to Mexico, estimates that the number of undocumented immigrants in the United States is 19.6 million, far exceeding widely accepted estimates.
Kaplan is the William N. and Marie A. Beach Professor of Operations Research, professor of public health, and professor of engineering; Rodilitz is a doctoral student in operations at Yale SOM. Their study appears in the journal Risk Analysis, the flagship publication of the Society for Risk Analysis.
The study draws on the Mexican Migration Project (MMP), which has surveyed communities in Mexico since 1982 (and annually since 1987), asking about the history of household members’ migration to and from the United States. MMP data alone cannot be used to directly estimate the number of immigrants in the United States, the authors say, because it is based on a “snapshot” of the population living in Mexico at one point in time.
“The MMP sampling creates a large bias in the data,” Kaplan says. “The observed durations of time undocumented migrants in the samples spent in the United States will greatly underestimate the true lengths of stay taken over all undocumented border crossers, while the trip departure dates observed among those sampled in Mexico will be biased backwards in time relative to the true trip departure dates among all of the migrants.”
Instead, Kaplan and Rodilitz use the data to create a probabilistic model of undocumented migration between Mexico and the United States, calculating the trip timing and duration most likely to lead to the results seen in the MMP survey. “With a probability model of true migrant trip departure and sojourn time distributions, we can determine what the data in the MMP surveys would look like solely due to the physical constraints that must be satisfied for a migrant to be included in the survey,” Rodilitz says. “This enables us to work backwards from the data and infer the true population flows.”
They conclude that from 1980 until 2005, when border security was tightened, most Mexican migrants traveled to the United States for shorter periods; since 2005, fewer have crossed the border but those who do stay longer. According to the model, the total number of immigrants currently in the United States who entered over the border is likely between 9.4 million and 19.8 million, with a “point estimate”—the most likely scenario—of 14.6 million. An additional 5 million immigrants are estimated to have overstayed a visa, leading to a total estimate of 19.6 million undocumented immigrants.
The estimate is nearly double the frequently cited 2018 estimate of 10.7 million total undocumented immigrants by the Pew Research Center. On the other hand, it falls within the range of estimates in a 2018 study by Kaplan, Yale SOM’s Jonathan S. Feinstein, and MIT’s Mohammad Fazel Zarandi. That study, which used mathematical modeling on a different set of demographic and immigration operations data, estimated a total of 16 million to 29 million undocumented immigrants, with 22.1 million as the mean.
“Our work thus adds additional evidence to the contention that there are many more undocumented immigrants in the United States than have been appreciated to date,” Kaplan and Rodilitz write.