By Paleo Pundit
Last year, I questioned Homeland Security Secretary Kristi Noem’s claim that 1.6 million illegal immigrants had “self-deported” since Donald Trump returned to office. The evidence offered for that assertion relied heavily on changes in the Census Bureau’s Current Population Survey (CPS), and I argued that the survey was being asked to bear more weight than it could reasonably support.
Since then, the evidence has changed, and any fair-minded critic such as Yours Truly should acknowledge it. There is now little question that immigration into the United States has declined substantially. Border encounters have collapsed, admissions of various kinds have fallen, and the Census Bureau itself has substantially reduced its estimates of net international migration. There is also evidence that emigration has increased. It is therefore increasingly plausible that the foreign-born population has stopped growing and may actually be declining.
But that is not quite the same thing as the much more dramatic conclusion now being offered by the Center for Immigration Studies (CIS).
Using the CPS, CIS reports that the foreign-born population declined from about 53.3 million in January 2025 to 50.4 million in July 2026—a drop of 2.9 million. CIS further estimates that roughly 2.3 million of this decline represents illegal immigrants. If correct, this would constitute a demographic reversal and imply that net international migration has been negative for an extended period.
There are good reasons, however, to treat those numbers cautiously.
The Census Bureau itself has warned against interpreting movements in the CPS foreign-born estimate as a simple count of immigrants entering or leaving the country. Discussing the decline that had already appeared in the survey during 2025, Census noted that changes could reflect actual emigration but could also reflect “survey non-response, coverage error or the impact of population controls.” That qualification is important because the CPS is a sample survey, not a population register or an administrative count of everyone residing in the country.
Population controls present a particular complication. In January 2025, the CPS incorporated unusually large revisions to the population estimates used to weight the survey. Those revisions added roughly 2.9 million people aged 16 and older to the estimated population and were driven substantially by revised estimates of the enormous immigration surge during the preceding years. The Bureau of Labor Statistics (BLS) explicitly cautioned that January 2025 estimates were therefore not directly comparable with December 2024 estimates.
This matters because January 2025 happens to be CIS’s starting point. Foreign-born status is not itself one of the characteristics to which CPS population weights are directly controlled. Nevertheless, changing the weights for age, sex, race and Hispanic origin can affect the resulting estimate of the foreign-born population. Another population-control adjustment occurred in January 2026, incorporating newer information about declining international migration.
This makes the CIS comparison more complicated than it first appears. During the period it examines, the Census Bureau changed the population estimates used to weight the CPS. Those changes were themselves influenced in part by new estimates showing that immigration had fallen. The revised population numbers then helped produce lower CPS estimates of the foreign-born population. CIS is now using that decline as evidence that millions of immigrants have left the country. In other words, some of the evidence used to estimate lower immigration may already be built into the CPS numbers that CIS is citing as proof of it.
Survey nonresponse creates another unresolved problem. CIS acknowledges the possibility that immigrants have become less willing to participate in a federal survey but argues that overall CPS response rates do not appear closely related to the decline and that immigrants who participate generally continue answering immigration-related questions.
That does not entirely answer the objection. The concern is not primarily that an immigrant agrees to participate in the CPS and then refuses to answer the birthplace question. It is that some immigrant households may be disproportionately unwilling to participate at all—particularly in an environment of heightened immigration enforcement.
CPS weighting can compensate for known demographic characteristics of nonrespondents, but it cannot easily correct for a characteristic it does not independently control. If foreign-born Hispanics became less likely to respond, for example, weighting could restore the appropriate Hispanic population without necessarily restoring the appropriate number of foreign-born Hispanics.
There is another useful check on the CIS interpretation. The Census Bureau’s Population Estimates Program independently estimates net international migration. Those estimates also show an enormous decline—from approximately 2.7 million in 2024 to 1.3 million in 2025 and a projected roughly 321,000 in 2026. That is evidence that something dramatic has changed, but these figures still imply positive net international migration, rather than the sustained negative migration seemingly required by the CIS interpretation of the CPS.
None of this means that CIS has discovered nothing. On the contrary, the evidence that immigration has fallen dramatically is now overwhelming, and the evidence that the foreign-born population itself has begun declining is stronger than it was when I wrote about this subject last year. I would therefore modify my earlier skepticism accordingly.
But direction and the magnitude of the change are two entirely different questions.
The CPS provides increasingly persuasive evidence of an immigration reversal. It does not provide anything approaching a precise head count showing that 2.9 million foreign-born residents have physically departed the United States. And the further conclusion that 2.3 million illegal immigrants have left is more uncertain still, since legal status is not directly measured by the CPS and must instead be estimated.
The fairest conclusion lies between the two extremes. The decline should not simply be dismissed as a statistical phantom. But neither should a volatile survey estimate—affected by sampling, nonresponse, changing population controls and assumptions about migration—be transformed into a confident declaration that 2.9 million immigrants have disappeared from the country.
The Trump administration can reasonably claim a reduction in immigration. Claims about millions of “self-deportations,” however, still require considerably more evidence.
Please consider supporting American Remnant: A green “Donate Today” button has been added at the end of each article (see below) appearing on the website. If you value what AR is doing, please consider supporting the website financially. $5, $10, or any amount that you can afford. Regular donations would especially be appreciated. Thank you!
