What are ICE’s documented attrition rates year‑over‑year since 2015?
Executive summary
A clear, authoritative year‑over‑year series of ICE “attrition rates” from 2015 through the present is not published in a single official table; publicly available datasets and trackers provide the raw components to calculate attrition (encounters/arrests/book‑ins versus removals), but inconsistencies and gaps in DHS/ICE releases mean a definitive, vetted annual attrition series cannot be produced from the reporting provided here alone [1] [2] [3] [4]. Independent aggregators—TRAC, the Deportation Data Project, Migration Policy Institute (MPI), and Vera—compile ICE’s original data and estimate removals, but they also warn that selective publication and changing counting rules complicate straight year‑to‑year comparisons [1] [5] [6] [4].
1. What “attrition rate” means here and how it is measured
In enforcement reporting the term “attrition” is typically used to describe the share of immigration enforcement cases or encounters that do not end in a removal—operationally calculable as the difference between ICE encounters/arrests/book‑ins and removals/returns for a period—but it is not an ICE‑defined single metric published as a year‑by‑year series in these sources; instead, ICE and DHS publish component tables (encounters, arrests, book‑ins, book‑outs, removals) that an analyst must combine to derive an attrition percentage [2] [3] [1].
2. What the primary sources offer for reconstructing attrition
ICE’s Enforcement and Removal Operations statistics provide the foundational counts—book‑ins to detention, book‑outs from detention, and removals—which are the necessary denominators and numerators for an attrition calculation, and the Deportation Data Project republishes ICE’s original files to make such computations tractable [2] [1]. The DHS Office of Homeland Security Statistics (OHSS) posts monthly tables that break out arrests, book‑ins, and removals at finer time resolution, enabling more precise year‑by‑year arithmetic—if one has the time series and consistency to reconcile monthly to fiscal‑year totals [3].
3. What the independent analysts report and the gaps they flag
Aggregators have produced removal totals that suggest dramatic year‑to‑year swings—MPI estimated about 340,000 deportations in FY2025, and TRAC reported large removal counts continuing into FY2026—yet both MPI and TRAC note that DHS has been selective about which ICE datasets it posts publicly and that some FY2025 detail remains unavailable, undermining a straightforward, fully transparent attrition time series [4] [6]. Vera’s detention dashboard supplies long‑range midnight population and detention flow data useful for context, but it focuses on detention counts and trends rather than publishing a formally labeled “attrition rate” series [5].
4. Why a simple year‑by‑year attrition table is elusive
DHS and ICE have altered the frequency and completeness of data releases over recent years, producing lags and partial reporting that break comparability; Migration Policy Institute and other observers explicitly cite DHS’s inconsistent public reporting as the major obstacle to tracking interior enforcement outcomes reliably across administrations [4]. In addition, different analysts use different denominators—some count only formally charged removal cases, others include voluntary departures and returns—so computed attrition rates can diverge depending on methodology even when using the same base ICE files [1] [3].
5. How to get a defensible attrition series and where to look
A defensible year‑over‑year attrition series from 2015 onward requires downloading ICE’s raw tables from ICE’s statistics pages and the Deportation Data Project, reconciling monthly OHSS tables into fiscal or calendar years, and deciding upfront which outcomes to count as “non‑removals” (e.g., releases, returns, dropped charges, voluntary departures) before calculating attrition; these primary sources are the starting points recommended by independent researchers and are cited here [2] [1] [3] [5]. Because the reporting here does not include a precompiled annual attrition table, any precise year‑by‑year numbers must be disclosed as analyst‑constructed and subject to revision if DHS later releases fuller tables [4].
6. Alternate readings and institutional incentives
Advocacy researchers highlight that selective transparency can mask enforcement intensity or outcomes and that agencies may have incentives—political or operational—to release headline removal totals while withholding granular case‑level disposition data that would permit clean attrition calculations; these concerns are raised explicitly by MPI and echoed by watchdogs who republish ICE’s raw feeds to preserve analytic reproducibility [4] [1] [5].