What Is Affecting the CPS Data on Shifts in Immigrant and Native-Born Populations?

In September 2025, the Congressional Budget Office (CBO) released an updated projection for net immigration, forecasting that the immigrant population age 16 and older would grow by 408,000 in 2025. However, according to Current Population Survey data, the U.S. immigrant population age 16 and older fell by 1.86 million people between January and November 2025—dropping from 50.40 million to 48.54 million. This decline was entirely concentrated among non-naturalized immigrants.The relevant population considered in this blog post corresponds to the one used in the calculation of the official unemployment rate, namely the civilian noninstitutionalized population age 16 and older. Following the Bureau of Labor Statistics definition, the foreign born are those residing in the United States who were not U.S. citizens at birth. That is, they were born outside the United States or one of its outlying territories such as Puerto Rico or Guam, to parents who were not U.S. citizens. The native born are people who were born in the United States or one of its outlying territories, such as Puerto Rico or Guam, or who were born abroad of at least one parent who was a U.S. citizen. (See the table below).

At the same time, CPS data show the native-born population age 16 and older rose by 3.8 million, from 222.3 million to 226.1 million—this increase was far more than the CBO's projected growth of about 1 million for this group over this period.Our calculations based on CPS microdata yield slightly different estimates for the native- and foreign-born population than those reported in BLS employment situation reports. In particular, the increase in the native-born population is 3.83 million (instead of 3.80 million in the microdata) and the decrease in the foreign-born population is 1.88 million (instead of 1.86 million in the microdata). These minor discrepancies arise from the Census Bureau's age perturbation process—a disclosure limitation technique that slightly alters ages in public use files to protect respondent confidentiality. This process changes the composite weights between internal and public use files. While total population estimates remain consistent, subgroup breakdowns (such as by citizenship status) show small differences when weighted counts are calculated. Furthermore, these numbers suggest a staggering, almost threefold increase in native-born population growth compared with average growth (2014-19) before the COVID-19 pandemic.

These numbers have sparked considerable debate among economists and researchers. Some, like Jed Kolko at the Peterson Institute for International Economics, contend the numbers are implausibly large and likely reflect measurement issues rather than an actual mass exodus.

To gain a better understanding of these numbers, both for natives and immigrants, this blog post looks carefully at how the CPS works. We then compare the CPS estimates for net immigration with our own estimates based on a different source: U.S. payroll data. Our analysis suggests that the drop in the immigrant population is overstated largely because of a drop in participation of non-naturalized immigrants in the CPS who remain in the country but are wary of participating in government data collection. As a consequence, the increase in the native-born population is also overstated: The CPS weighting system must match predetermined population totals, so when immigrant numbers fall, native-born numbers must rise.

What the CPS Can and Cannot Measure

To begin with, it is important to note that the CPS records the number of foreign-born people in the U.S. in a given month. Changes over time therefore reflect net immigration—the sum of inflows (a positive number) and outflows (a negative number), but the CPS cannot distinguish between the two.For example, in December, the U.S. Department of Homeland Security reported 605,000 deportations (an outflow), while the U.S. State Department reported that between January and May 2025, about 50,000 immigrant visas were issued monthly (an inflow). Detailed data on all relevant in- and outflows separately are not readily available.

How Does the CPS Construct Population Aggregates?

The CPS surveys about 60,000 households each month and uses a sophisticated weighting system to scale these responses to match the entire U.S. population. Think of it this way: If one person in the survey represents 30 people in real life, that person’s “weight” is 30.

Here’s how the CPS produces its population estimates. The Census Bureau creates what it calls “population controls“—benchmarks that the CPS must match—for specific demographic groups defined by age, sex, race, and Hispanic origin. These controls are estimates that take the official population numbers from the latest decennial census and adjust them annually with data from official records, such as birth and death records, in following years. Each month, the CPS adjusts its survey weights so that the survey’s respondents align with the Census Bureau’s population projections for each of these groups, i.e., factoring in how much each of these demographic groups should grow over time.

Here’s the critical limitation: The Census Bureau does not produce population controls or growth projections for citizenship or nativity (whether someone is a U.S. citizen or whether someone was born in the U.S. or abroad). This means the CPS cannot directly adjust its weights to maintain the balance between native-born and foreign-born populations, which itself is not perfectly known to begin with. Instead, these groups are affected only indirectly through the adjustments made for age, sex, race and ethnicity.

What the Survey Responses Actually Show

Looking at the actual survey responses reveals something important. In January 2025, the CPS had 68,069 native-born respondents, 6,142 naturalized immigrant respondents, and 5,867 non-naturalized immigrant respondents. By November 2025, the CPS saw a drop in respondents across all groups relative to January: native born, down by 6.2%; naturalized immigrants, down by 5.6%; and non-naturalized immigrants, down by 16.6%. Put differently, non-naturalized immigrants’ participation dropped roughly 10 percentage points more than participation of the two other groups. Hence, there is indeed a real decline in this group, both in terms of survey participation and in the weighted population estimates derived from the survey.Interestingly, the larger decline in non-naturalized immigrant respondents relative to the two other groups is much more pronounced in the first interview cycle (interview months 1-4), when households are new to the survey. Respondents in the second interview cycle (interview months 5-8), who had previous experience with the CPS, show participation patterns more similar to those of native-born and naturalized respondents. This suggests that unfamiliarity with the survey may increase reluctance to participate among non-naturalized immigrants in the current enforcement environment.

For the non-naturalized immigrant population, the raw survey data and weighted data thus tell a consistent story. For the native-born population, however, the data are inconsistent. Native-born survey respondents saw a drop by November, yet the weighted native-born population increased by 3.8 million. This is a massive increase; it corresponds to an increase that is 2.6 times larger than the prepandemic average. (See the table above.)

This huge increase in the weighted native-born population, despite no corresponding increase in actual survey responses, is the statistical artifact we're identifying. However, this increase is a consequence of the weighting system having to balance the books: When the foreign-born population declines and the total population is predetermined by Census projections, the native-born population must rise to make the numbers add up.

Testing the Numbers: Employment Growth as a Reality Check

The larger decline in non-naturalized immigrant respondents raises an important question: Does this reflect a surge in the number of people who have actually left the country (in net terms), or does it reflect the number of people who are still here but have become increasingly reluctant to respond to government surveys? In an environment of heightened immigration enforcement, both scenarios are plausible and likely at play at the same time. Some immigrants may have left voluntarily, anticipating stricter enforcement or fearing deportation. Others may remain in the country but have grown wary of participating in any government data collection, even though the CPS is confidential and used only for statistical purposes.

If the decline is primarily driven by survey reluctance rather than actual departures, then the reported population decline would overstate the real exodus. To get a better sense of what is actually happening, we can look at employment data as a reality check.

Between January and November 2025, the U.S. economy added 499,000 jobs according to the Current Employment Statistics (CES), which is based on a monthly survey of about 121,000 businesses and government agencies across the country. However, the household survey (CPS) typically measures slightly higher employment levels than the establishment survey due to several well-understood methodological differences, including how they treat multiple jobholders and self-employment. In 2025, CPS employment has been running about 2.6% higher than CES employment. Adjusting for this difference, the 499,000 jobs in the establishment survey would correspond to about 512,000 jobs in household survey terms.

Now, using the CBO’s estimate of 1 million for native population growth (not the CPS’ 3.8 million), and applying the average seasonally adjusted native-born employment-to-population ratio of 59% in 2025, we’d expect native-born workers to have filled about 590,000 jobs.We applied an X13-ARIMA seasonal adjustment to the native-born and foreign-born employment-to-population ratios hosted on FRED.

Yet the 590,000 figure is more than the actual jobs created. This gap of 78,000 suggests that immigrants left the workforce and native-born workers filled those vacated positions. Using the foreign-born employment-to-population ratio of 63.5%, this implies net negative immigration of only about 123,000 people—far less than the CPS weighted numbers suggest.

Now, Federal Reserve Chair Jerome Powell reported in his December Federal Open Market Committee press conference that his staff estimates job growth in 2025 was actually lower and will be revised downwards as part of the annual benchmark revision (which will be released as part of the employment situation report for January 2026). Based on the Fed's estimates, between January and November 2025 only 187,000 jobs were created. Repeating the same calculations as above, this would be consistent with net immigration of about -627,000, which again is much smaller than the CPS numbers suggest.

What This Means for Understanding Immigration Data

These findings highlight an important feature of how the CPS produces population estimates. The survey does an excellent job tracking labor markets and overall demographics, but it faces inherent challenges when measuring subgroups by nativity or citizenship for which the Census Bureau does not produce population controls. This is not a flaw in the CPS methodology; it’s simply a limitation that arises from the survey’s design. When immigration patterns shift dramatically, whether through surges or declines, the weighting system must work harder to reconcile predetermined population totals with actual survey responses.For examples of the challenges in measuring the effects of shifting immigration levels, see Alexander Bick’s January 2025 blog post series: "The Recent Surge in Immigration and Its Impact on Unemployment" and "The Recent Surge in Immigration and Its Impact on Measured Productivity Growth."

We have two main takeaways. First, the CPS population estimate showing a 1.86-million decline in the number of immigrants mostly reflects a decrease in the participation of immigrants in the CPS. The question is whether the decline in surveyed immigrants indeed means an actual drop in net immigration or simply fewer immigrants choosing to participate in the survey.

Our analysis in the previous section suggests that consistency with U.S. job growth implies net immigration ranged between -123,000 and -627,000. These are still substantial numbers, but far below the CPS estimate of -1.86 million. This suggests that the major force for the large negative net immigration in the CPS is a drop in participation of non-naturalized immigrants who remain in the country but may be wary of participating in government data collection. Second, given this decline in participation, the methodology of the CPS requires an offsetting increase in the native population (and thus also natives’ job growth) that is implausibly large compared with the prepandemic period.

As policymakers, journalists and researchers rely on these data to understand immigration and employment trends, it’s essential to look beneath the surface at both survey participation patterns and corroborating evidence from other sources. Getting these numbers right—and understanding their methodological limitations—is more important than ever as immigration policy continues to be a major topic of national debate.

Notes

1. The relevant population considered in this blog post corresponds to the one used in the calculation of the official unemployment rate, namely the civilian noninstitutionalized population age 16 and older. Following the Bureau of Labor Statistics definition, the foreign born are those residing in the United States who were not U.S. citizens at birth. That is, they were born outside the United States or one of its outlying territories such as Puerto Rico or Guam, to parents who were not U.S. citizens. The native born are people who were born in the United States or one of its outlying territories, such as Puerto Rico or Guam, or who were born abroad of at least one parent who was a U.S. citizen.
2. Our calculations based on CPS microdata yield slightly different estimates for the native- and foreign-born population than those reported in BLS employment situation reports. In particular, the increase in the native-born population is 3.83 million (instead of 3.80 million in the microdata) and the decrease in the foreign-born population is 1.88 million (instead of 1.86 million in the microdata). These minor discrepancies arise from the Census Bureau’s age perturbation process—a disclosure limitation technique that slightly alters ages in public use files to protect respondent confidentiality. This process changes the composite weights between internal and public use files. While total population estimates remain consistent, subgroup breakdowns (such as by citizenship status) show small differences when weighted counts are calculated.
3. For example, in December, the U.S. Department of Homeland Security reported 605,000 deportations (an outflow), while the U.S. State Department reported that between January and May 2025, about 50,000 immigrant visas were issued monthly (an inflow). Detailed data on all relevant in- and outflows separately are not readily available.
4. Interestingly, the larger decline in non-naturalized immigrant respondents relative to the two other groups is much more pronounced in the first interview cycle (interview months 1-4), when households are new to the survey. Respondents in the second interview cycle (interview months 5-8), who had previous experience with the CPS, show participation patterns more similar to those of native-born and naturalized respondents. This suggests that unfamiliarity with the survey may increase reluctance to participate among non-naturalized immigrants in the current enforcement environment.
5. We applied an X13-ARIMA seasonal adjustment to the native-born and foreign-born employment-to-population ratios hosted on FRED.
6. For examples of the challenges in measuring the effects of shifting immigration levels, see Alexander Bick’s January 2025 blog post series: "The Recent Surge in Immigration and Its Impact on Unemployment" and "The Recent Surge in Immigration and Its Impact on Measured Productivity Growth."