How do estimates of unauthorized immigrants during 2017–2020 compare to previous administrations (2009–2016)?
Executive summary
Estimates show the unauthorized immigrant population was largely stable across the two periods: researchers put the stock at about 11 million through 2009–2016 and find nearly the same level around 2017–2020, with only modest year-to-year variation that often falls within sampling margins of error [1] [2] [3]. Differences between studies stem from methods—residual estimation, survey reweighting, administrative data on visa overstays and border encounters—so comparisons must be read through methodological caveats rather than as evidence of a dramatic population swing tied to any single administration [4] [2] [5].
1. Numbers on the ledger: flat stock, modest swings
Multiple leading analyses conclude the stock of unauthorized residents stayed roughly flat from the Obama-era years (2009–2016) into the Trump-era window (2017–2020); for example, Robert Warren and other analysts report “about 11 million” throughout the 2010s, and Pew finds 2017 and 2019 levels nearly identical, with 2021 only a modest uptick relative to 2019 [1] [3] [6]. Migration Policy Institute and other outlets later revised upward for years after 2019, but for the specific 2017–2020 comparison the consensus in earlier residual- and survey-based work is stability rather than a large decline or surge [7] [1].
2. Why estimates look similar: entries, exits, and offsets
The reason totals remain similar across administrations is that flows in are frequently offset by exits—voluntary departures, removals, deaths or adjustment to lawful status—and because the primary mode of becoming unauthorized shifted toward visa overstays rather than border crossings, a change that alters the composition of the unauthorized population without necessarily changing its size dramatically [1] [5] [6]. Analysts note that from 2009–2019 overstays were the dominant source of the stock [1], and border encounter statistics alone therefore give an incomplete picture of stock change [8] [5].
3. Methodology matters: survey residuals, administrative tweaks, and margins of error
Estimates rely on residual methods and survey reweighting that are sensitive to Census benchmarks, sampling variability, and the treatment of special populations (e.g., Cubans) and years affected by COVID disruptions; Pew cautions that small year‑to‑year differences—like a 200,000 apparent change between 2016 and 2017—can fall within the margin of error and are not necessarily real [2] [9] [4]. The Social Security Administration and Center for Migration Studies papers explain methodological revisions for 2017–2019 that can change historic ratios and therefore estimated counts, underlining that methodological changes—not policy alone—can drive apparent changes across administrations [1] [4].
4. What diverging signals say: encounters, apprehension rates, and later revisions
Border encounter totals and DHS apprehension rates declined from earlier decades into 2013–2020, which some interpret as reduced illegal entries, but analysts stress apprehension rates and expulsions (including Title 42-era practices) complicate interpretation of successful crossings versus attempted entries [8] [5]. Moreover, more recent analyses and institutional revisions—MPI and some Census-updated series—have produced higher post‑2019 population estimates, indicating that stock estimates are still being revised and that the 2017–2020 snapshot can look different depending on which dataset and revision one uses [7] [10].
5. Bottom line and caveats
The bottom line: authoritative, peer-reviewed and government-derived series largely show the unauthorized population remained close to ~11 million across 2009–2016 and 2017–2020, with composition shifting toward longer residence and more visa overstays; however, methodological differences, sampling error, and later upward revisions for post‑2019 years mean that simple administration-to-administration comparisons risk overstating precision [1] [2] [6]. Reporting that treats point estimates as exact without noting margins of error or changes in estimation procedures implicitly advances a clearer narrative than the data support; readers should privilege trend patterns and methodological transparency over single-year headline numbers [4] [2].