How do removals and returns get counted differently in DHS statistics and independent analyses?
Executive summary
DHS distinguishes “removals” (formal orders carrying administrative penalties) from “returns” (departures not pursuant to removal orders) and also counts expulsions under some authorities like Title 42, and these legal definitions drive how numbers are categorized in official statistics [1] [2]. Independent analysts often produce different totals because they pull from different component reports (ICE vs CBP), apply different deduplication or inclusion rules, and treat operational practices such as voluntary withdrawals, expedited removals, and expulsions inconsistently [3] [4] [1].
1. How DHS defines removals and returns in plain terms
DHS’s public statistical guidance says a “removal” is the compulsory execution of an order of removal and carries administrative or criminal consequences on reentry, while a “return” is a confirmed sending of an inadmissible or deportable person out of the United States that is not pursuant to an order of removal — in short, removals = formal orders with penalties, returns = departures without those penalties [1] [2]. The agency also labels “expulsions” separately where public-health authority (Title 42) or other special processes apply, and those were reported distinctly between March 2020 and May 2023 [1].
2. How DHS components produce and aggregate those counts
DHS statistical offices extract data from component systems — Border Patrol’s BPETS snapshots, CBP’s BorderStat, and ICE’s Integrated Decision Support systems — and aggregate them into DHS datasets like the OHSS Persist Dataset or OIS yearbooks; these systems are updated, cleaned and standardized by statistical officials before publication [1] [4]. ICE also issues its own removals statistics that exclude CBP’s border-side counts, meaning DHS-wide totals can differ from ICE-only totals simply because different components are being tallied or excluded [3].
3. Data processing choices that create divergent numbers
DHS reports that its statisticians clean and validate agency operational reports, convert records to standardized formats, and use unique identifiers to deduplicate agency data — all steps that change raw operational tallies into the official published counts [4]. At the same time DHS admits CBP removals and returns figures are estimates based on dispositions, citizenship, detention records and other enforcement history fields in live systems “subject to change,” which creates room for later revisions and for independent analysts who freeze different snapshots to report different figures [1].
4. The specific gaps that independent analyses exploit or confront
Independent analysts diverge from DHS counts for predictable reasons: some combine ICE interior removals with CBP border removals despite DHS sometimes releasing component-only statistics; others include expulsions or Title 42 actions with removals or treat voluntary withdrawals and admissions-withdrawn-as-returns inconsistently [3] [1]. Congressional and academic reports have long noted there are at least two sets of statistics — OIS/DHS-wide tables and ICE-produced removals — and interpreting which number answers a particular policy question requires knowing which series was used [3].
5. Stakes, institutional incentives, and interpretive traps
Different counts matter politically and operationally: counting expulsions or returns as “removals” inflates the sense of formal removals; excluding CBP data understates border enforcement totals; and methodological opacity lets stakeholders select the series that best supports their stance — whether to claim enforcement is rising or falling — so readers must check which component and which definition underpins quoted totals [1] [3]. DHS presents its products as independently produced for transparency, but component-driven reporting and live-system estimates create implicit agendas for both internal performance narratives and external advocacy claims [4] [1].
6. What the current reporting can and cannot resolve
Official DHS metadata explains definitions, systems and cleaning procedures but also warns that live-system-derived estimates are subject to correction, and it documents separate data streams (ICE, CBP, OIS) that produce different aggregates — meaning precise reconciliation often requires access to component records and time-stamped extracts that DHS sometimes does not publish in full [1] [4] [3]. Independent analysts can narrow gaps by stating upfront which component counts and definitions they use, but any definitive crosswalk beyond that depends on DHS releasing harmonized, row-level datasets with provenance, something the public record here does not fully provide [4] [1].