Introduction
Many assume people experiencing homelessness are not working. In Los Angeles County, which has the largest population of people experiencing homelessness in the country, a prior California Policy Lab report found that three-quarters of people who received homeless services had worked before, and about 1 in 5 worked while receiving them. But that study examined a pre-pandemic world, when both homelessness and the labor market looked different.
This data point updates the picture. Using more recent data, we revisit the employment and earnings of Los Angeles County residents enrolled in homelessness services. The new results reaffirm what we found before: most people who received homeless services had prior work experience, challenging the idea of long term labor market detachment. However, despite some earnings growth after enrolling in services, overall earnings remained low relative to area median income even two years later. This provides an opportunity for targeted workforce interventions to help workers increase their earnings to levels that would promote long-term housing stability.
Key Findings
- Most people who experienced homelessness were previously employed. Eighty percent of people who experienced homelessness had worked in California before enrolling in homeless services.
- Employment rates among people who experienced homelessness were moderate to low in the years just before enrolling in homeless services. About half of the people (48%) who experienced homelessness were employed in the four years before enrolling in homeless services and 21% were employed at the time they enrolled in services.
- For those who worked, earnings were 27% higher after receiving homeless services than before, but were low in all time periods. Average annual earnings two years after enrollment in homeless services were $18,468 for those who worked at least one quarter, or just 21% of Los Angeles County’s area median income.
- Employment rates varied substantially across demographic groups and program types. In the two years before enrollment in homeless services, members of families and transition-aged youth (18-to-24 year olds) had higher employment rates than other demographic groups. Among different types of homeless services, people enrolled in prevention and time-limited housing subsidy programs had the highest pre-enrollment employment rates.
Data Sources1
- Homeless Management Information System (HMIS) records from the Los Angeles Homeless Services Authority (LAHSA) show demographics, program enrollment, and service use for people enrolling in homeless services.
- Unemployment Insurance (UI) Base Wage data from the California Employment Development Department (EDD) show quarterly earnings from formal, unemployment insurance (UI)-covered employment in California. The data do not include earnings from self employment, informal work, or out-of-state employment.
Sample: Heads of household aged 18–70 who enrolled in homeless services between January 2019 and July 2022, including street outreach, interim housing, prevention, time-limited subsidies, and permanent supportive housing programs.2 The sample is restricted to people enrolled in homeless services whose data included the information needed for attempting to link with employment records (N = 95,386).3 Most people in the sample were male (58%) and Black (55%) or Hispanic (30%) (Appendix Table A1). Their average age was 43.
Research Partnership
This research was conducted through a partnership between the Labor Market Information Division of the California Employment Development Department and the California Policy Lab, a nonpartisan research center at the University of California, with sites in Berkeley, Los Angeles, and Sacramento.
Results
Work Experience Before Homelessness
Most people who experienced homelessness in Los Angeles had some prior connection to formal employment (Figure 1).4 Roughly four out of five people had worked in California at some point before enrolling in homeless services.
However, employment was far less common in the periods immediately preceding enrollment. Less than 40% worked at any time in the two years prior and about 1 in 5 (21%) worked in the quarter just before or during enrollment.
Employment and Earnings Before and After Homelessness
Figure 2 compares quarterly employment rates before and after enrollment in homeless services. Among all people experiencing homelessness, employment remained low and relatively stable, hovering around 20% both before and after enrollment. To better understand employment dynamics for individuals who may be more likely to be able to work, we separately analyzed individuals with recent employment. Employment rates were higher among these “recent workers” — those with earnings in the third and fourth years before enrollment — but their employment declined steadily as they approached enrollment and it did not return to earlier levels.
We also examined average earnings among those who worked. Figure 3 shows quarterly earnings for people with at least one quarter of positive earnings. For both the full sample of all workers and recent workers, average earnings were higher after enrollment in homeless services than in the period just before. While this analysis cannot answer why this is happening, it’s possible that enrolling in services helped people to stabilize their housing situation, and as a result, focus more on finding or maintaining work. Recent workers consistently earned slightly more than the overall sample, though their trends are similar.
Despite these post-enrollment increases, average earnings remained low. Two years after enrollment, people who worked in Q8 earned an average of $6,664, or 31% of Los Angeles County’s quarterly area median income of $21,727.5 Because households earning 30% of area median income or less are typically considered extremely low income, these results show that even people who were employed had annual earnings well below levels associated with housing stability in a high cost region like Los Angeles County.6
Employment and Earnings by Demographic Groups
Employment rates in the two years before enrollment in homeless services varied across demographic groups and household types. As a reference point, Figure 1 showed that 39% of all people experiencing homelessness worked during this period. Relative to this benchmark, women had above-average employment rates and were more likely than men to have worked. White and American Indian or Alaska Native individuals were less likely to have worked than people of other racial and ethnic groups. People in families and transition-aged youth had substantially higher employment rates than single adults.
We also compared average annual earnings across demographic groups for those who worked at least one quarter before enrollment to those who worked at least one quarter after enrollment. As a reference point, for the full sample, average earnings rose from $14,558 in the second year before the year of enrollment to $18,468 in the second year after enrollment, representing a 27% increase in average annual income. Earnings increased after enrollment for every demographic group, though levels varied.
Men tended to earn more than women, and earnings differed across racial and ethnic groups, reflecting broader labor market disparities. People in families generally earned more than single adults, while transition-aged youth started with the lowest earnings but saw the largest gains after enrollment. Even with these increases, earnings remained low for all groups — only 17–23% of the county area median income of $86,911.7
Employment and Earnings by Health Conditions
Employment rates in the two years before enrollment also differed based on health conditions. Although health conditions were associated with lower employment rates, many people with a physical or behavioral health condition still had recent work experience. A substantial share of people with a mental health condition (36%), substance use disorder (36%), or both (31%), worked in the two years before enrollment (Figure 6). It’s possible these rates could be higher if people had the right support and services. Employment rates were lowest for people with a physical disability and highest for people without a physical or behavioral health condition.
Among those who worked, average annual earnings were higher after people enrolled in homeless services, regardless of health conditions (Figure 7). Earnings varied somewhat by health status. People with no physical disability earned more than those with a physical disability and likewise, people with no behavioral health condition earned more than those with a behavioral health condition. However, all groups had low earnings even two years after enrollment, reinforcing that employment among people experiencing homelessness is associated with limited earnings, regardless of health status.
Employment and Earnings by Homeless Service Type
People in homelessness prevention (for example, one-time financial assistance to maintain housing) and time-limited subsidy programs were the most likely to have worked before enrollment (Figure 8). Employment rates were lower in permanent supportive housing, where many participants are older or have severe health conditions. Employment rates were also lower in street outreach and interim housing programs, though a meaningful share (30–42%) of people in those programs had worked in the two years prior to enrollment.
Average annual earnings among people who worked generally declined in the year before homeless service enrollment and then stabilized or increased afterward. This rebound was most visible for people in time-limited subsidy, street outreach, interim housing, and permanent supportive housing programs.
In contrast, people in prevention programs showed relatively stable earnings over time, with a small pre-enrollment decline and no immediate post-enrollment increase, but they did see a bump in the second year after enrollment. Despite these trends, earnings remain low across all programs, accounting for only 19–29% of county area median income.
Conclusion
This analysis shows that many people who experienced homelessness in Los Angeles County had a history of work in California — challenging the common perception that homelessness is tied to long‑term detachment from the labor market. Although employment rates and earnings were low in the period immediately surrounding enrollment in homeless services, a meaningful share of people experiencing homelessness — especially those with recent work histories, members of families, and transition-aged youth — had higher rates of employment. Among those who did work, average earnings increased after enrollment but remained far below what is needed for long‑term housing stability. We caution, however, that this analysis is descriptive and does not speak to the causal effects of homelessness, service enrollment, or specific programs.
These patterns point to an important opportunity: many people who experience homelessness have prior work experience and some continued connection to the labor market. This suggests that targeted workforce interventions — such as job-readiness training, transitional employment, and job search support — can build on these existing strengths to help people secure and maintain the employment and earnings growth needed to maintain housing.
Acknowledgements
This research was made possible through support from the Conrad N. Hilton Foundation. We also gratefully acknowledge our partners that supported this research, including the Los Angeles Homeless Services Authority and the California Employment Development Department. This publication reflects the views of the authors and not necessarily the views of our funders. All opinions and errors should be attributed entirely to the authors. Any statements in this publication should be attributed solely to the authors and do not necessarily reflect the views of the Los Angeles Homeless Services Authority or the California Employment Development Department. All calculations were performed by the California Policy Lab. Any errors or omissions are the responsibility of the authors.
Suggested Citation: Caprara, C., Gross, M., & von Wachter, T., (2026). Employment and Earnings Before, During, and After Experiencing Homelessness. California Policy Lab, University of California. https://capolicylab.org/employment-and-earnings-before-during-and-after-experiencing-homelessness/.
Appendix
Sample Characteristics
The first column of Table A1 describes the average characteristics of heads of household aged 18-70 who enrolled in homeless services between January 2019 and July 2022 (N = 156,540). Most people in this sample were male (61%) and Black (37%) or Hispanic (35%). Adults made up the majority of the sample (85%), with families and transition-aged youth representing 5% and 9%, respectively. The table also shows the distribution of age, prevalence of self reported health conditions, and the extent of prior homeless service enrollments.
Reweighted sensitivity analysis
Linking HMIS and EDD records required restricting the sample to the 95,386 people enrolled in homeless services with social security number (SSN) data. However, the second and third columns of Table A1 show that there were differences in the average pre-enrollment characteristics between people enrolled in homeless services with and without available SSN data. We made sure that our findings represent the typical experiences of individuals in the HMIS. To do this, we reweighted the SSN-linked sample (column 3) to have similar average, pre-enrollment characteristics to all people enrolled in homeless services (column 1). The fourth column of Table A1 shows that reweighting was successful in creating a more representative sample
To implement reweighting, we estimated a logistic regression model using the full HMIS population to predict whether a person had an SSN recorded in HMIS, using pre-entry observable characteristics including demographics, household type, and health indicators. We then predicted the probability that each person had an SSN based on the regression model. Next, we constructed inverse probability sampling weights for the 99% of the sample (N = 94,736) with complete characteristic data. We then applied the weights to the SSN sample when estimating employment and earnings.
Based on reweighted analyses, the main findings are likely to generalize to the full population of people experiencing homelessness in Los Angeles County. Figure A1 compares employment rate estimates for the reweighted and unweighted samples and shows differences of no more than one percentage point. Figure A2 presents quarterly employment rates for the full sample and recent workers, with no meaningful difference in trends for the reweighted results. Figure A3 presents average quarterly earnings for the full sample and recent workers. While reweighting slightly increases estimated earnings across time periods, these changes are small and do not alter overall trends or key findings.
TABLE A1: Characteristics of all people enrolled in homeless services compared to the subset of people enrolled in services without available SSN data, with available SSN data (before weighting), and with available SSN data after reweighting to be more representative of all people enrolled in services
| Category | ALL PEOPLE (%) | PEOPLE WITHOUT SSN DATA (%) | PEOPLE WITH SSN DATA — UNWEIGHTED (%) | PEOPLE WITH SSN DATA — REWEIGHTED (%) |
|---|---|---|---|---|
| Gender | ||||
| Female | 37 | 32 | 41 | 38 |
| Male | 61 | 65 | 58 | 61 |
| Race/Ethnicity | ||||
| Black | 37 | 25 | 44 | 37 |
| White non-Hispanic | 24 | 25 | 24 | 24 |
| Hispanic | 35 | 41 | 30 | 35 |
| Asian American | 1.7 | 1.6 | 1.7 | 1.7 |
| American Indian or Alaska Native | 2.4 | 1.9 | 2.7 | 2.5 |
| Native Hawaiian and Pacific Islander | 1.1 | 0.8 | 1.2 | 1.1 |
| Household Composition | ||||
| Adult | 85 | 88 | 84 | 85 |
| Families | 5 | 2 | 7 | 6 |
| Transition-Aged Youth | 9 | 10 | 9 | 9 |
| Age | ||||
| 18–24 | 10 | 10 | 10 | 10 |
| 25–34 | 24 | 25 | 23 | 24 |
| 35–44 | 23 | 26 | 21 | 23 |
| 45–54 | 20 | 21 | 20 | 20 |
| 55–64 | 18 | 15 | 20 | 18 |
| 65+ | 5 | 4 | 5 | 5 |
| Health Condition | ||||
| Substance Use | 15 | 11 | 17 | 16 |
| Mental Health | 26 | 11 | 35 | 26 |
| Physical Disability | 16 | 6 | 22 | 16 |
| Prior HMIS | ||||
| 5+ | 6 | 1 | 10 | 6 |
| 3 to 4 | 7 | 1 | 10 | 7 |
| 1 to 2 | 20 | 9 | 28 | 21 |
| 0 | 66 | 89 | 52 | 66 |
| Sample (N) | 156,540 | 61,154 | 95,386 | 94,736 |
Table note: All categories are mutually exclusive, except for race/ethnicity and health condition. Individuals are classified into all race and ethnicity categories to which they identify, with one exception. Consistent with LAHSA reporting practices to address a data collection issue, those who reported both Hispanic ethnicity and White race are only classified as Hispanic and excluded from the White category. Prior HMIS indicates the number of homeless service enrollments between the individual’s first enrollment during the sample period and the earliest available HMIS data, which begins in 2010. Percentages for mutually exclusive groups may not sum to 100% due to rounding.
ENDNOTES
- This study was reviewed and approved by the UCLA Institutional Review Board. ↩︎
- Throughout this brief, we use the term “people experiencing homelessness” to refer to all individuals enrolled in these homeless service programs. This includes individuals enrolled in homelessness prevention programs who may not have been experiencing homelessness at the time of enrollment. We use this terminology for simplicity and consistency throughout the report. ↩︎
- This involved excluding 61,154 people (or 39% of all heads of household aged 18-70 enrolled in homeless services during the study period) with missing social security number (SSN) data. SSN data could be incomplete in the HMIS, for example, because their SSN was not recorded correctly at the time of enrollment. There were differences in the demographic characteristics and service histories between people who enrolled in homeless services who did and did not have SSN data (Appendix Table A1). This raises the possibility that our analysis is not representative of all people enrolled in homeless services. To assess this, we conducted a sensitivity analysis that reweights the sample of people with an SSN to look more like the full population of people enrolled in homeless services. Findings from the reweighted data were nearly identical to the main findings, suggesting the main conclusions likely generalize to the full population of people experiencing homelessness. See the appendix for the full results. ↩︎
- Throughout, employment refers to jobs eligible for UI that we can see in the EDD wage data. It excludes informal employment. ↩︎
- This back-of-the-envelope calculation of Los Angeles County quarterly area median income uses the U.S. Census Bureau’s 2023 data, adjusts it to July 2022 dollars using the Bureau of Labor Statistics CPI Inflation Calculator, and then divides by four to get a quarterly estimate. Census source: https://www.census.gov/quickfacts/fact/table/losangelescountycalifornia/INC110217#INC110217. CPI calculator: https://www.bls.gov/data/inflation_calculator.htm ↩︎
- 2025 Affordable Housing Program Income Limits published by the Los Angeles County Department of Regional Planning. Source: https://planning.lacounty.gov/wp-content/uploads/2025/05/Income-Limits-2025.pdf ↩︎
- Los Angeles County area median income is based on the U.S. Census Bureau’s 2023 estimate adjusted to July 2022 dollars using the Bureau of Labor Statistics CPI Inflation Calculator. Census source: https://www.census.gov/quickfacts/fact/table/losangelescountycalifornia/INC110217#INC110217 ↩︎