A Look at the New Center for Migration Studies Illegal Population Estimates

By Steven A. Camarota on February 8, 2016

The Center for Migration Studies (CMS) recently released a report entitled, "US Undocumented Population Drops Below 11 Million in 2014, with Continued Declines in the Mexican Undocumented Population". The CMS write-up emphasizes a "continued decline" in the illegal population, but the change from 2012 to 2014 they report is less than 2 percent. More important, the data they use to estimate illegal immigration has a margin of error, so it is not really possible to draw the conclusion that there has been a significant change since 2012. Among the things to keep in mind when reading the CMS report are:

  • CMS estimates 10.9 million illegal immigrants in the country in 2014. The margin of error from the public-use file of the American Community Survey (ACS), the survey they use, is about ±106,000 for a population of this size. Therefore, statistically, it is not possible to know if the illegal population is actually below 11 million as their headline states.

  • Given the size of the margin of error, the 100,000 decline in illegal immigrants from 2013 to 2014 that CMS reports is almost certainly not statistically significant. Even the decline of 198,000 from 2012 to 2014 is probably not statistically significant. Thus, it is not clear if the illegal population "continued" to decline as the report argues.

  • It must be remembered that a margin of error of ±106,000 only deals with the statistical properties of the ACS, not the added error associated with trying to identify illegal immigrants in the survey, which is how CMS creates its estimates. The actual error must be larger.

  • A stable or even declining illegal population does not mean that new illegal immigrants are not coming. CMS, the Pew Research Center, and CIS have all estimated that 300,000 to 400,000 new illegal immigrants arrive each year, though the number coming is clearly lower than it was a decade ago. New arrivals are offset by deportations, return migration, legalizations, and deaths.

  • The decline in the illegal population reported by CMS, if correct, would tend to undermine an argument for legalizing illegal immigrants because it shows that the problem can be solved over time without a radical solution such as an amnesty for those in the country illegally.

  • Unlike CMS, the Pew Research Center does not report a decline in the illegal population in recent years. Pew's figures show that from 2009 to 2014 the illegal population "remained essentially stable". They also show an illegal population in 2014 of 11.3 million — 400,000 larger than CMS.

  • The CMS estimate only goes through the middle of 2014, so most of the 237,543 unaccompanied minors and family members from Central America who have crossed the southern border from 2012 to 2016 are not included in their estimates.

The new estimates of the illegal population from CMS show an illegal population of 10.9 million in 2014. The report was authored by Robert Warren, a well-respected analyst who formerly worked for the Immigration and Nationalization Service. While at INS he developed the government's estimates of the nation's illegal immigrant population.

Below, I discuss some issues with the new estimates. But it important to note at the outset that if the illegal immigrant population really has steadily declined in the way that CMS reports, then it would tend to undermine the argument for an illegal immigrant amnesty. One of the central arguments for legalizing illegal immigrants is that there is really "no choice": The illegal population is here to stay, so it is better to legalize it. But a slow decline may be the best possible solution because it represents a slow change, which is much easier for all parties involved to adjust to than something sudden. Moreover, allowing the illegal population to decline over time avoids radical action like an amnesty. A slow decline from 2010 to 2014 would also mean that even with the very weak enforcement efforts of the Obama administration the illegal population can be reduced. If there was a robust enforcement regime, the problem would almost certainly be solved more quickly.

One of the biggest problems with the CMS report is the way the findings are presented. The headline and the accompanying article emphasize a "continued" decline in the illegal population. But this conclusion is not supported by data they present. The illegal estimates from CMS are based on the public-use file of the American Community Survey, and like any survey it has a margin of error. Although CMS does not provide it, for a population of 10.9 million illegal immigrants drawn from the public-use file of the ACS, the margin of error must be a little over 100,000. We can estimate the margin of error for the illegal population by using the total foreign-born Mexican population in the 2014 ACS as a proxy population. In 2014 the ACS showed 11.7 million Mexican immigrants, with a margin of error of ±110,000. If we simply use the same procedure for calculating the margin of error for an illegal population of 10.9 million, the margin of error would be ±106,000 for 2014. This assumes a 90 percent confidence level. If we assume a 95 percent confidence level the margin of error is ±127,000. The illegal population is very similar in characteristics to the overall Mexican immigrant population so the confidence interval would have to be nearly identical.

Although they only report a single-number estimate of 10.9 million with no margin of error, the actual range for their 2014 estimate must be roughly 10.8 to 11 million. Thus, statistically it is not correct to write that the illegal population has dropped below 11 million as their headline states. This same problem exists when they discuss trends. The Center for Immigration Studies was one of the first to report a drop in the illegal population after 2007. But CMS argues this decline continued, showing a 813,000 decline from 2010 to 2014, with half occurring in just one year — 2010 to 2011. The decline over the entire period 2010 to 2014 is statistically significant. But given the margin of error in the survey, the 100,000 decline from 2013 to 2014 is not statistically significant. Even the 198,000 decline from 2012 to 2014 is almost certainly not significant. (Remember the margin of error has to be applied to the values for both years when determining if a change is statistically significant.) Thus, when CMS argues the decline has "continued" in recent years, it is going beyond what their data shows.

It must also be remembered that a margin of error of ±106,000 only deals with the statistical properties of the survey, not the added error associated with trying to distinguish illegal immigrants in the American Community Survey. In a 2013 International Migration Review article Robert Warren explains in detail the method he used to create his estimates of the illegal immigrant population. His estimates include calculation for in-migration, outmigration, and mortality. He must also incorporate administrative data from the Department of Homeland Security (DHS) on lawful permanent residents, temporary non-immigrants (e.g. guestworkers and students), and refugee admissions. All of this information introduces additional error and needs to be considered when evaluating small changes like those from 2013 to 2014 or from 2012 to 2014.

Having said this, it is certainly possible that the illegal population dropped over this time period in the way CMS describes. There is general agreement that the illegal immigrant population is no longer growing as it did 1990 to 2007. But there is certainly no consensus that it continued to drop in the last five years. The Pew Research Center reports an illegal population of 11.3 million for 2014, and referring to the time period 2009 to 2014 Pew states that the illegal population, "has remained essentially stable for five years." The Department of Homeland Security shows a much smaller decline in the illegal population from 2010 to 2012 than CMS. DHS estimates show a fall-off of 200,000 over this time period, in contrast to the 600,000 shown by CMS. (Unfortunately, 2012 is the last year for which DHS has published estimates.)

This is not to argue that Pew and DHS are correct and CMS is mistaken. But the changes CMS reports are relatively modest 2012 to 2014 — less than 2 percent. And the other research not only produces somewhat different point estimates, there is even disagreement about tends. This is to be expected given the uncertainty inherent in estimating illegal immigration. It may also be worth noting that the public-use ACS file shows that the total Mexican immigrant population (legal and illegal) grew by 154,000 between 2013 and 2014. This is not the trend one would expect if illegal immigration is declining from that country, as CMS argues; but this change, it must be pointed out, is not statistically significant.

Finally, CMS estimates only go through the middle of 2014, so they do not include most of the surge in unaccompanied minors and families from Central America coming across the southern border. A total of 237,543 have arrived as of January 2016. CMS does report that the number of illegal immigrants from Central America increased by 62,000 from 2013 to 2014. This is the only region of the world for which they show a large increase from 2013 to 2014. Nonetheless, most of the new illegal immigrants from Central America arrived after mid-2014 and so are not included in the new estimates.

The new CMS estimates are carefully done and may well capture a continued decline in the illegal immigrant population as they contend. But given the limitations of the data, the results should have been reported in a more circumspect manner. Because of these limitations, other research has come to somewhat different conclusions about the numbers and trends in the population of illegal immigrants in the United States.


I would like to thank Robert Warren for answering my questions about his methodology in a series of emails.