Encampment clearances

How often sweeps happen, who carries them out, and what follows

Last data pull: July, 2026

This page describes how unhoused Angelenos experience encampment clearances, or “sweeps”, drawing on PATHS monthly surveys from April 2023 onward. A respondent is counted as swept in a month if they reported being told by an official to move from where they were staying because of a “sweep” or a no camping or sitting/sleeping/lying law in the past 30 days.

Exposure to policing

Sweeps are one part of a broader pattern of contact with law enforcement. In any given month, 25% of respondents reported a sweep and 15% reported a police encounter. These experiences add up: over their first year in the study, 64% of respondents were swept at least once.

In any given month

Share of all monthly survey responses in which the respondent reported each experience in the past 30 days (n = 14,431–16,456 responses). Citations and arrests are counted among all respondents, including those with no police encounter that month.

In the first year of follow-up

Share of respondents who reported each experience at least once, and more than once, during their first 12 months in the study. Restricted to respondents who completed at least 6 monthly surveys in that period (n = 580–648; on average they completed 9 surveys), so these figures reflect only the months in which a survey was completed.

How often people are swept

Across the full period, 25% of monthly responses reported a sweep in the past 30 days. The rate was 32% in the first six months of data and 25% in the most recent six months. Respondent-reported trends may differ from official counts of sweep operations because the number of operations does not necessarily reflect the number of people affected: when encampments are scattered or sites are repeatedly cleared, more operations may reach fewer people.

All PATHS respondents

Share of monthly responses reporting a sweep, by calendar quarter. Point size reflects the number of responses in each quarter; the solid line is a smoothed trend. Boxed labels show the rate pooled over the first two and last two quarters.

City of LA: PATHS respondents and official records

Left: share of monthly responses reporting a sweep, among respondents located in a City of LA council district. Right: number of CARE+ encampment cleanup operations recorded by the City of LA, by quarter. The two are measured differently (a rate vs. a count), so compare trends over time within each, not their levels. Solid lines are smoothed trends.

Who gets swept

This section shows which respondents report sweeps most often, first by where people stay and then by demographic group.

By type of unsheltered place

Among respondents who were mostly unsheltered in the past 30 days, sweep rates across the full period were 40% for people staying in a tent, 39% for people sleeping rough without a tent or vehicle and 22% for people staying in a vehicle.

Share of monthly surveys reporting a sweep, by where the respondent stayed most often in the past 30 days and calendar quarter. Limited to respondents who were mostly unsheltered. The three charts share the same scale. Point size reflects the number of surveys in each quarter; solid lines are smoothed trends. Boxed labels show rates pooled over the first two and last two quarters.

By demographic group

To compare groups on a more equal footing, rates here are among respondents who were unsheltered at least one night in the past 30 days. Across the full period, 30% of their monthly surveys reported a sweep. The rates are unadjusted: they show how often each group reported a sweep, without accounting for other differences between groups.

  • Each dot is the share of a group’s monthly surveys that reported a sweep. The whiskers are a 95% confidence interval, which shows how precise the rate is: wider whiskers mean fewer surveys behind the number. The interval accounts for the same person answering many monthly surveys.
  • The dashed line is the rate for all unsheltered respondents, so dots to its right are groups that reported sweeps more often than average.
  • A person contributes one survey per month, so someone surveyed for many months counts many times. Hover over any point for the numbers behind it.
  • Groups with fewer than 30 people are not shown.

Sweep rates were 32% for Hispanic or Latino respondents, 31% for Black respondents and 28% for White respondents.

Hispanic or Latino includes respondents of any race; the Black and White groups are non-Hispanic. Respondents of other races or who preferred not to say are not shown.

Sweep rates were 30% for male respondents and 29% for female respondents.

Respondents who preferred not to say are not shown. Not shown (fewer than 30 people): Neither male nor female.

What comes with a sweep

A sweep can come with an offer of shelter or housing, a citation, both, or neither. The two charts below track each separately.

Offers of shelter or housing

Across the full period, 11% of sweeps came with an offer of shelter or housing. The rate was 17% in the first six months of data and 9% in the most recent six months.

Share of sweeps in which the respondent was offered shelter or housing, among sweeps where the respondent said whether they were offered, by calendar quarter.

Citations

Across the full period, 12% of sweeps led to a citation. The rate was 9% in the first six months of data and 14% in the most recent six months.

Share of sweeps in which the respondent received a citation, among all sweeps, by calendar quarter.

In both charts, point size reflects the number of sweeps in each quarter, the solid line is a smoothed trend, and boxed labels show rates pooled over the first two and last two quarters. The two charts share the same scale.

From offer to shelter

Of the 3,401 sweeps where the respondent said whether they were offered shelter or housing, 387 (11%) included an offer. Of the 350 respondents who were offered and said whether they accepted, 227 (65%) accepted, and 65% of those were still there at the time of the survey. Altogether, 4% of sweeps led to a shelter or housing stay.

Same-month pathway after a sweep, counted in sweeps.

  • Each bar is one step, from being swept to still being in shelter or housing at the time of the survey. The label on each bar gives its share of all sweeps; hover over a bar to see its share of the step before it.
  • The unit is a sweep, not a person. Each monthly survey in which a respondent reported a sweep counts once, so someone swept in several months is counted each time.
  • The top bar includes only sweeps where the respondent said whether they were offered shelter or housing.
  • The acceptance step leaves out the 37 respondents who were offered shelter or housing but preferred not to say whether they accepted.
  • “Still in shelter or housing” means the respondent was still there when they took the survey, within about a month of the sweep. It doesn’t show how long they stayed after that.
  • Only offers made during the sweep are counted. Offers made through Encampment Resolution Programs or by outreach workers at other times are covered on other pages.

Acceptance by type of offer

Offer type Share accepted Number offered
Any shelter or housing offer 65% 227 of 350
Bridge housing or transitional housing program 80% 49 of 61
Temporary stay in a motel/hotel-like setting 78% 100 of 128
Supportive housing 78% 28 of 36
Private room in shelter 68% 30 of 44
A tiny home 55% 17 of 31
Congregate (group) shelter 46% 36 of 79

Among respondents offered each type of shelter or housing during a sweep who said whether they accepted.

  • The first row covers any offer of shelter or housing during a sweep, and matches the acceptance step in the funnel above.
  • Each row below it covers sweeps in which the respondent was offered that type and said whether they accepted it. Rows are sorted by share accepted.
  • Types overlap. A respondent can be offered more than one type in the same sweep, so the same respondent can appear in more than one type row.
  • Some types are offered rarely (as few as 31 times), so their shares rest on small numbers and can shift noticeably as new months of data come in.

Where people go after a sweep

Respondents who were swept and did not move into shelter or housing either stayed where they were, moved temporarily and later returned, or moved and stayed away. In the first six months of data, these shares were 39%, 32% and 30%, respectively; in the most recent six months, they were 27%, 47% and 25%.

What respondents did after being swept, among those who did not move into shelter or housing, by calendar quarter (from July 2023). The three shares add up to 100% in each quarter, and the three charts share the same scale. Point size reflects the number of sweeps in each quarter; solid lines are smoothed trends. Boxed labels show shares pooled over the first two and last two quarters.

Who carries out sweeps

Across the full period, respondents who were swept most often reported that police or other law enforcement (63%) were involved, followed by sanitation workers (38%) and homeless outreach workers (17%).

Among respondents who were swept, the share who reported each type of worker as involved, by calendar quarter. Respondents answered separately about each type, so shares can add up to more than 100%. Solid lines are smoothed trends. Boxed labels show shares pooled over the first two and last two quarters.

How sweeps affect people

This section compares each person’s survey responses in the month they were swept with their responses in the preceding month when they had not yet been swept. As a comparison, people who were not swept in either month show how much these outcomes change over a similar period in the absence of a sweep. The chart and table below present the raw percentages for both groups, rather than estimates from a statistical model.

  • Each row is one outcome. For each group, the dot is the share of respondents reporting that outcome at the earlier survey, and the arrowhead is the share at the later survey. The table below the chart gives the same numbers.
  • Swept (red) are respondents who were swept in the later month but not the earlier one. Not swept (grey) are respondents who were swept in neither month. The not-swept arrows show how much each outcome shifts over one to two months without a sweep, for example with the seasons or the passage of time.
  • Compare the arrows, not where they start. People who get swept can differ from those who don’t even before the sweep, so the two groups’ dots often start at different levels. What matters is how far, and in which direction, each arrow moves. When the swept arrow is much longer than the not-swept arrow, that outcome changed more among people who were swept than among similar people over the same kind of interval.
  • Rows are sorted by how much the swept group’s change exceeds the not-swept group’s change, largest at the top.
  • These are raw percentages. They don’t account for chance variation or for other differences between the two groups, and some outcomes are based on small numbers.

Swept: 827 pairs of surveys; not swept: 5,245 pairs. Restricted to respondents who were unsheltered at both surveys.

Outcome Swept: before Swept: after Not swept: before Not swept: after
Displaced 49.4% 65.1% 42.2% 42.3%
Had a police encounter 17.4% 32.4% 10.4% 9.9%
Verbally insulted or harassed 28.2% 32.2% 21.9% 20.9%
Experienced food insecurity 77.0% 80.5% 62.1% 61.2%
Experienced sexual or physical violence 23.7% 27.6% 13.3% 12.9%
Visited the ER or hospital 14.1% 17.6% 9.5% 8.9%
Property theft 33.4% 36.6% 21.0% 20.1%