Every year, the Los Angeles Homelessness Services Authority (LAHSA) coordinates a volunteer-driven census of people living on the street. In the past two years, it’s gotten less accurate.
At stake is millions of dollars in federal and county assistance that could go to serve homeless Angelenos. The good news is that using professional counters to cross-check volunteers on the night of the official count could end the undercount.
The count has always been required as part of the calculus for federal assistance, but now also determines how roughly $100 million in local sales taxes will be distributed each year to fund homelessness services through Measure A.
With homelessness finally declining in LA County, properly directing these resources to sustain progress is critical. This means understanding the count’s accuracy, which is where new findings from our Los Angeles Longitudinal Enumeration and Demographic Survey (LA LEADS) raise concerns.
LA LEADS has been surveying the highly impacted communities of Hollywood, Venice, and Skid Row since 2021. Like LAHSA, we tally unsheltered persons (“rough sleepers”) and their dwellings (tents and vehicles). Unlike LAHSA, we rely on professional counters who scan each area six times per year instead of just once.
Comparing LA LEADS to the four LAHSA counts it now spans reveals three trends that could hurt progress by diverting resources away from needs.
First, LAHSA’s data represent a growing undercount. In 2024, LA LEADS found 35 percent more persons and dwellings than LAHSA in the same census tracts. This year, that rose to 47 percent. Extended across the entire city of LA, this offset suggests that 7,900 persons and dwellings may be missing from the official data—more than Orange County’s entire homeless population.
Second, LAHSA’s undercount varies from neighborhood to neighborhood. Hollywood’s official total was 81 percent of LA LEADS’s this past February, but Skid Row’s was just 61 percent. If this neighborhood-level unevenness reaches across municipalities—which seems likely—some cities will get one third more Measure A dollars per unsheltered person/dwelling compared to others.
Third and most troublingly, LAHSA’s undercount depends not just on where people live, but how.
Since 2021, the LA LEADS neighborhoods have lost about 600 tents and gained about 300 rough sleepers—the most isolated, highest-need individuals living without the protection of a structure or vehicle. The net effect of this substitution of easy-to-count tents with hard-to-count, more vulnerable individuals is to make LAHSA’s count less accurate and turn that inaccuracy into a systemic bias that could actively push resources away from where they are needed.
Others have argued that LAHSA’s count is typically a lowball, but our data don’t necessarily support that conclusion. In 2022 and 2023, LAHSA and LA LEADS were within five percent of each other, suggesting that there is no intrinsic issue with the official methodology. Further, LA County’s recent homelessness decline is not an artifact of an undercount—it shows up in LA LEADS, too, and is probably driven by the success of policies like LA City’s Inside Safe, which target tent encampments. Simply, our data are showing a growing geographic and demographic bias in LAHSA’s statistics that has the potential to divert Measure A resources and stymie that progress.
Fortunately, there is a fix: use professional counters to cross-check volunteers on the night of the official count. These independent data—the acquisition of which may be fundable through Measure A itself—would then recalibrate the official census to make it more accurate. The return on this investment in policy confidence would be substantial, to say nothing of the human gains from better-leveraging precious resources.
In an area in which easy wins are rare, this is one of them.
Louis Abramson is an Adjunct Physical Scientist at RAND.