How do DHS residence-state estimates of unauthorized immigrants compare to ACS-based residual estimates at the county level?
Executive summary
state-estimates">DHS residence-state estimates are produced by subtracting administrative counts of lawful immigrants from Census Bureau measures of the immigration-statistics">foreign‑born—an approach shared with ACS‑based residual methods used by Pew, MPI and academic groups—but DHS outputs are generally published at the state level and incorporate administrative adjustments that are not always disclosed in public detail [1] [2]. At the county level, direct DHS‑to‑ACS residual comparisons are constrained: researchers rely on ACS‑based residual methods (often pooling multiple years and using SIPP and other adjustments) to generate county estimates, but those local estimates carry substantially larger sampling and undercount uncertainty than state or national figures [3] [4] [5].
1. DHS method and its typical geographic reporting
DHS’s Office of Immigration Statistics builds its “illegal/unauthorized” counts by estimating the total foreign‑born (from the ACS) and subtracting legally resident populations drawn from DHS administrative records, producing estimates broken down by state, period of entry, age and sex [1] [6]. DHS’s documents and datasets emphasize state‑level tabulations and component accounting (e.g., LPRs, nonimmigrants, refugees), and DHS has historically adjusted for ACS undercount assumptions when producing its state residence estimates [7] [8].
2. ACS‑based residual estimates and how researchers downscale them
Residual estimation—used by Pew, MPI, CMS and researchers like Warren and Warren—starts with ACS counts of the foreign‑born at whatever geography is available and subtracts best estimates of legal residents (from DHS administrative data), then adjusts for undercount and nonresponse; to get county‑level numbers, researchers typically pool multiple years of ACS data and draw on SIPP or other surveys to assign legal/unauthorized probabilities by origin, arrival cohort, age and sex [9] [2] [3]. MPI explicitly combines pooled 2019–23 ACS with SIPP and DHS inputs to produce local estimates and characteristics, acknowledging the need for weightings and adjustments at the county scale [3].
3. Where DHS and ACS residuals align — and why they diverge
At national and state levels, estimates from DHS, Pew, MPI and other groups tend to fall in a relatively narrow range, which lends confidence to broad patterns and magnitudes [5] [10]. Differences that do exist are driven by timing (reference dates), sampling variability in the ACS, divergent undercount assumptions, and methodological choices about how to treat special statuses (TPS, DACA, parolees) and emigration—factors that can shift counts by hundreds of thousands nationally and by large percentages locally [10] [9].
4. Why county‑level comparisons are especially fragile
County estimates amplify the ACS’s sampling error and the sensitivity of residual logic: small absolute differences in legal‑resident allocations or undercount adjustments translate into large percentage swings in counties with modest immigrant populations, and ACS single‑year county samples are thin enough that researchers often pool multiple years to stabilize estimates [5] [3]. DHS does not routinely publish a full set of county residence estimates comparable to every ACS residual table, which limits direct apples‑to‑apples validation at the county scale [1] [6]. Where independent county residuals exist, they are best interpreted as model‑based approximations with wider uncertainty than state or national estimates [5] [8].
5. Alternative approaches and institutional blind spots
Some analysts argue for monthly CPS‑based methods to reflect more current dynamics, especially after 2020, and for greater transparency in DHS administrative treatment of emigration and temporary parole categories; proponents say CPS can capture short‑term inflows better, while DHS’s administrative edge lies in precise legal counts but limited public disclosure on emigration and undercount modeling [11] [5]. That dispute highlights an implicit agenda: agencies and groups favor methods that fit their operational needs—DHS emphasizes administrative fidelity and state reporting, researchers prioritize local detail and survey‑based residuals—so disagreements often reflect differing tradeoffs between timeliness, geographic granularity and public reproducibility [2] [11].
6. Bottom line for county‑level users
DHS residence‑state outputs and ACS‑based residual estimates use fundamentally similar arithmetic (foreign‑born minus legal residents) and align at broader geographies, but DHS does not systematically provide the same county‑level estimates that researchers derive from pooled ACS residual methods; consequently, county comparisons are model‑dependent and carry notably larger uncertainty due to ACS sampling variability, undercount corrections and methodological choices about special immigration statuses [1] [3] [5]. If rigorous county‑level comparisons are required, the best practice is to use pooled ACS residuals with explicit uncertainty bounds and to document how DHS administrative categories and undercount assumptions were applied, because DHS’s published state figures cannot be blindly downscaled without those disclosures [4] [7].