How do DHS Yearbook removals compare to ICE’s own interior removals statistics (ERO) for 2017–2020?
Executive summary
The DHS Yearbook’s removals totals and ICE ERO’s interior removals for 2017–2020 are measuring overlapping but not identical phenomena: the Yearbook aggregates DHS repatriations across components (removals, returns, and — for 2020 — expulsions), while ICE ERO reports focus on interior removals and operations after final orders, so Yearbook totals will include actions that ICE ERO’s interior-removal line does not [1] [2]. The divergence stems from different counting rules, agency responsibilities, and the inclusion of Title 42 expulsions and CBP actions in DHS-wide tallies, meaning a like-for-like comparison requires adjusting definitions rather than reading raw totals as equivalent [1] [3].
1. What each dataset claims to count and why that matters
The DHS Yearbook compiles tables on removals, returns, and expulsions across DHS components and defines “repatriations” to include removals (which carry administrative penalties), returns (which do not), and Title 42 expulsions when applicable, drawing on data from ICE, CBP, and other offices [1] [4]. By contrast, ICE ERO statistics present the agency’s enforcement and removal operations — focused largely on interior arrests, detention, and removals that usually follow a final order of removal under Title 8 — and are built to describe ERO’s operational outputs, not the entire department’s repatriations universe [2] [5].
2. Counting rules and methodological shifts that widen the gap
DHS and ICE revised counting methodologies to align certain administrative arrest and removal tallies (notably in 2016), and the Yearbook explicitly warns readers that changes in counting rules and venue (interior vs. border) alter headline numbers; that revision affects comparability across years and between ICE ERO and Yearbook aggregates [6] [3]. Moreover, DHS began including CBP “expulsions” under Title 42 in Yearbook tables for 2020–2021, which are operationally distinct from ICE removals and can inflate DHS repatriation totals relative to ICE ERO’s interior-removal counts [3] [1].
3. The Title 42 wrinkle: expulsions vs. removals
Between March 2020 and May 2023, U.S. authorities used a public-health expulsions authority commonly called Title 42; DHS Yearbook tables count those expulsions in repatriations (when CBP performed them), but ICE labels Title 8 removals as the agency’s formal removals — actions generally following final orders and involving different logistics like charter flights and coordination [2] [1]. Thus, Yearbook totals for 2020 will include many rapid expulsions that ICE ERO’s “removals” series does not treat as ICE removals, causing a sharp apparent divergence in that year [3] [2].
4. What this means for interpreting 2017–2020 headline numbers
Raw Yearbook totals for 2017–2020 cannot be equated to ICE ERO interior removals without disaggregating returns, CBP expulsions, and ICE removals and applying consistent counting rules; the Yearbook’s broader scope makes its totals larger and more department-wide, while ICE ERO’s numbers are narrower, interior-focused, and tied to Title 8 removal processes [4] [2] [1]. Any claim that “one agency deported more” is thus a framing choice: comparing DHS Yearbook repatriations to ICE ERO removals mixes apples and oranges unless the analyst reclassifies events by venue and legal pathway first [6].
5. Where reporting and advocacy can mislead readers
Advocacy groups, media, or officials can selectively cite Yearbook or ICE ERO series to support competing narratives — e.g., “overall deportations rose” (using Yearbook repatriations that include expulsions) versus “interior removals fell” (using ICE ERO interior removals); the underlying agendas are often implicit: DHS Yearbook aims at comprehensive departmental accounting while ICE focuses on its operational performance, and both sometimes omit easy crosswalks for public reuse [1] [2]. The available sources document methodology differences and the 2016 alignment change, but the documents provided here do not contain a side‑by‑side numeric reconciliation of 2017–2020 totals, so precise year-by-year numeric comparisons cannot be produced from these sources alone [6] [3].