by Mark Gross, Chris Dick, and Beth Jarosz
Demography has historically been a quiet field with little time in the national spotlight. Between the three of us, we have spent nearly 5 decades working at the nexus of applied demographic modeling and public policy. We have built population estimates for the United States Census Bureau (Chris and Mark), built population estimates and forecasts outside of government (Beth and Chris), and have advocated for improvements in population estimation at both the federal and state levels (Beth and Chris). However, even with all of our combined experience, we were surprised by the news this week.
According to reporting from CNN, and confirmed through conversations we’ve had with state demographers, in order to receive FEMA Emergency Management Performance Grants, states “must now submit a population certification as of September 30 detailing their methodology and confirming that individuals removed under US immigration laws are not included in the tally.”
Regardless of how one views this new requirement philosophically, strictly on technical grounds it raises several concerns for us. First of all, the above statement from FEMA lacks clarity on who should be removed from the population estimate. Does it include only those who have already been removed? Those who have received removal orders? What about self-deportations? Even if we assume we have a clear answer to that question, multiple questions and concerns remain.
First: States don't have deportation data, and often don’t estimate immigration and emigration.
International migration is typically the most challenging of the components of population estimates to measure, and within international migration, emigration (people moving out of the U.S., including deportations) is the most difficult to measure in the best circumstances. Emigration typically cannot be measured directly because emigrants are no longer living in the United States, and therefore are no longer covered by any U.S. census or survey. Instead, demographers use mathematical models to estimate the number of people who leave the U.S. in a given year. And states typically rely on net migration estimates from the U.S. Census Bureau for the latest information, because they lack any better data for estimation purposes. Indeed, many states use methodologies that do not require international migration estimates because of these issues.
Even if states were able to track emigration more directly, the federal government data sources on immigration/emigration are not centralized or easily obtained (especially on short timelines). More immediately, the federal government has not updated its reporting on deportations. A common resource for data on ICE Enforcement and Removal Operation Statistics is only current through December 2024 (first quarter of fiscal year 2025). Another commonly-used source, Immigration Enforcement and Legal Processes Monthly Tables, also only reflects data through December 2024. Moreover, even if the data were current, they’re published by “area of responsibility,” not by state.
Second: The timeline is unreasonable.
The U.S. Census Bureau, which produces the official population estimates for the nation and states, typically engages in months of research, testing, and validation before making any methodological changes or incorporating new data into their estimates. To require states to craft new estimates on their own, and on such a short timeline, is likely to yield highly inaccurate estimates.
Perhaps more importantly, the new requirement is for estimates as of September 30, 2025, but given the time it takes to process the other elements of population estimates (birth certificates, death records, and moves), the data simply do not yet exist. In fact, the United States Census Bureau generally publishes their first estimates for a given year (in this case July 1, 2025) right at the end of the year. These state total estimates are generally released before December 31 of the year. This timeline is as fast as these data can be created given input data timelines and the need for quality assurance procedures.
Third: Estimating new population counts shouldn’t be a state activity.
As noted above, the U.S. Census Bureau provides official population estimates for the nation and states, ensuring that consistent methodology and data are used for all states. While many states produce their own estimates for internal purposes, states vary widely in their capabilities–some have state demographers and robust data and research teams while others subcontract to universities for their demographic analysis. But even if all states had teams of demographers at the ready to deploy a new estimation methodology, and consistent access to the required input data, the task simply does not belong at the state level. By asking states (and state equivalents like Washington, D.C. and Puerto Rico) to do the work, the Administration will get 52 different methodologies for the results, none of which will be comparable.
Further, it is not clear who at FEMA would be qualified to review and approve the complex methodologies used to produce population estimates. While the estimates produced by the U.S. Census Bureau certainly have limitations (usually due to the limited availability of appropriate data), their methodology has been continually reviewed and vetted by demographers, government officials, academics, and public data users over the course of decades. It is unclear why FEMA wouldn’t go to the United States Census Bureau to ask for this expertise, and also to update the estimates.
Fourth: Updated population counts don’t appear to serve a practical purpose.
According to the Trump administration, “Two million illegal aliens have left the United States in less than 250 days, including an estimated 1.6 million who have voluntarily self-deported and more than 400,000 deportations.” All of these data points, if verified, should be captured by the U.S. Census Bureau’s current Population Estimates methodology. Asking states to calculate independent estimates is duplicative and wasteful. Further, it is unclear from FEMA’s ask which populations should be subtracted from the population estimate.
Moreover, despite ramped-up enforcement, the amount of difference deportations would make in state population estimates is negligible. Even assuming a sizeable deportation level, such as the 400,000 cited above, deportations reflect just 0.12% of the total U.S. population.
While population estimates methodologies should and do incorporate small components of change, such as deportations in this case, overhauling the entire approach to population estimates to incorporate such a negligible component of population change does not make sense. This is especially true since deportations will be accounted for in the Census Bureau’s current methodology.
Since the new estimates produced by states are likely to be less accurate, incomparable, and incorporating the negligible impact deportations are having on population change, it begs the question of what the point of this demand is.