The actual size and the origin of the illegal alien population in the United States is uncertain and is difficult to ascertain because of difficulty in accurately counting individuals in this population. Figures from national surveys, administrative data and other sources of information vary widely. By all measures, the population of illegal aliens in the USA has declined substantially since 2007. The number of border apprehensions has substantially declined since 2000, the peak year, but more than doubled in the most recent fiscal year to approach 2007 levels.
Size
According to the nonpartisan Pew Research Center, the number rose rapidly in the 1990s, "from an estimated 3.5 million in 1990 to a peak of 12.2 million in 2007," then dropped sharply during the Great Recession before stabilizing in 2009. Pew estimated the total population to be 11.1 million in 2014, or approximately 3 percent of the U.S. population. This "is in the same ballpark" as figures from the United States Department of Homeland Security (DHS), which estimated that 11.4 million undocumented immigrants lived in the United States in January 2012. The estimate and trends are also consistent with figures reported by the Center for Migration Studies, which reported that the U.S. illegal immigrant population fell to 10.9 million by January 2016, the lowest number since 2003. A 2018 paper by three Yale School of Management professors estimated that the undocumented immigrant population was in the range of 16 million to 29 million, however the methodology presented in this study has been criticized as leading to vastly overstated results.
Estimation method
The "residual method" is widely used to estimate the undocumented immigrant population of the US. With this method, the known number of legally documented immigrants to the United States is subtracted from the reported U.S. Census number of self-proclaimed foreign-born people (based on immigration records and adjusted by projections of deaths and out-migration) to obtain the total, undocumented immigrant (residual) population. This methodology is used by the US Department of Homeland Security, the Pew Hispanic Center, the Center for Immigration Studies, and the U.S. Census Bureau. Since undocumented immigrants have many reasons for not answering the U.S. Census correctly and since penalties for answering the U.S. Census incorrectly are rarely enforced, it is accepted that it under-counts the number of undocumented immigrants. The users of this methodology assume that 10% of undocumented immigrants are not counted by census takers. The 10% assumption is based on a 2001 University of California survey asked of 829 people born in Mexico and living in Los Angeles whether they responded to census interviewers in the 2000 census with 40% of queried households refusing to answer the survey. Critics claim that the estimate is unreliable for a number of reasons: figures for outmigration are not tracked by the federal government; the proportion of undocumented immigrants who respond to the Census is unknown; the estimate that 10% of undocumented immigrants do not respond to the census is arbitrary and unsupported by a sufficient sample size and geographic spread; and that the self reporting of where one was born relies on the honesty of the person being questioned.
In 2016, the Pew Research Center estimated that, in 2014, there were 11.1 million undocumented immigrants in the US. These estimates are based on modeling using data from the American Community Survey (ACS) or the Current Population Survey conducted by the United States Census Bureau each year.
A 2018 paper by three Yale School of Management professors yielded similar trajectories of the undocumented immigrant population, with peak growth in the 1990s and early 2000s followed by a plateau from approximately 2008 onward. However, their model yielded an estimate of the numbers of undocumented immigrants of 22.1 million undocumented immigrants as the mean — roughly twice as great as the estimates based on the ACS. Moreover, according to the three authors, their estimate has a 95 percent probability range of 16 million to 29 million. That result, however, was criticized for vastly overstating the true number and for failing to account for the circular flow rate.