The number of people living on the streets in San Diego County may be 50 percent higher than thought, according to a new study.
That means the annual count of homeless people overall could be much greater than numbers the federal government uses to fund housing programs.
The study, led by University of New Hampshire statistician Chris Glynn and sponsored by the real estate database company Zillow, factored in the relationship of housing costs to homelessness, a departure from the traditional head-count method in determining the number of homeless people.
“The relationship between housing costs and homelessness has important implications for the way that city and county governments respond to increasing homeless populations,” Glynn wrote in the introduction to his study.
Gordon Walker, CEO of the San Diego Regional Task Force on the Homeless, said Thursday that he had not heard of the Zillow study, but acknowledged that rent can play a role in homelessness.
“Does rent increase homelessness?” he said. “In a tight market, it’s probable.”
But Walker was quick to defend the merits of the so-called point-in-time count the task force oversees each January.
“We try very hard to count people,” he said. “Last year, we had 1,700 volunteers, and we work very hard.” Walker said the criticism of the count was based on a assumptions the researchers were making about the San Diego event.
The Zillow study examined numbers from the 2016 point-in-time count and included U.S. Census data and the Zillow rent index from 25 cities. It also factored in the number of “true” homeless by extrapolating from an earlier study that found the point-in-time count in New York missed 41 percent of unsheltered people.
In 2016, the annual count in San Diego found 8,669 homeless people, while the Zillow study estimated there actually were 11,149, a difference of 28.6 percent.
The real story was in the number of unsheltered people, said Skylar Olsen, a senior economic with Zillow.
In San Diego that year, the count revealed 4,940 people were living outside, while the Zillow study estimated there actually were 7,420, a difference of 50 percent.
(The numbers from this past January: 9,116 homeless people in San Diego County, including 5,621 who were unsheltered and 3,495 in shelters.)
Data from the annual count is plugged into a formula the U.S. Department of Housing and Urban Development uses to determine funding for homeless programs. In 2016, the count found 550,000 homeless people nationwide.
While the study showed the numbers of people on the street may be higher than what was found in the official count, San Diego compared favorably to other cities that had greater differences between the two.
Also of note, the study reported that San Diego has had decreases in homelessness at times when rent increased, an indication that the region may be doing a good job of serving the population. The study also predicted that future rent hikes wouldn’t increase the homeless population here as much as in other cities.
The HUD point-in-time count has never been considered an exact science. It includes a head count of homeless people in shelters taken one January night, when shelters are likely to be the fullest, and the tally is considered accurate.
The harder part is counting people outside. Volunteers with clipboards and flashlights search dark storefronts, alleyways, canyons and other places shortly before dawn, a time picked because most people are asleep, minimizing the chance of accidentally counting someone twice.
It’s always been assumed that some people are missed, and Zillow found miscounts were greater in cities with warmer climates, where more homeless people are likely to be outside in winter.
In Los Angeles, the 2016 count overall found 47,000 homeless people while the Zillow study estimated there were 60,000, a difference of 27.7 percent.
In Tampa, 1,800 homeless people were counted. Zillow estimated there were 3,204, a difference of 78 percent. And in Riverside, there were 2,200 homeless people tallied and Zillow estimated there were 3,400, a difference of 54.5 percent.
As economist Olsen explained, Glynn and fellow researcher Emily Fox used a method statisticians refer to as Bayesian hierarchical model. Simply put, the method involves mapping how different variables relate to each other, and it starts with a guess at a formula that is adjusted until the findings best reflect the observed data.
Critics say the method uses subjective information in making assumptions, while supporters argue that it uses relevant information in the calculations.
The Zillow study incorporated findings from a 2008 study in New York, where researchers blended in with homeless people on the night of the HUD point-in-time count. They found that only 59 percent of the unsheltered people were counted.
Olsen said the 59 percent finding was a starting point in determining the true number of homeless, but it fluctuated from city to city along with other variables.
An encouraging sign in San Diego was that the relationship between rising rent and homelessness was not as strong here as it is in other cities. In fact, at times homelessness dropped as rents rose between 2011 and 2016, Olsen said.
“Where there’s no strong relationship, there may be social policies to step in the way,” she said. “If those local agencies were not making an impact, you would have seen an even greater increase in homelessness.”
Rick Gentry, president and CEO of the San Diego Housing Commission, said he wasn’t surprised to hear a study had found that HUD point-in-time count may be low nor was he surprised that the study found some encouraging signs in San Diego.
“But for the work of a lot of people in homeless issues, it would be worse,” he said.
The Housing Commission operates 3,400 housing units and has 15,000 housing vouchers to get people, he said, while another effort is on track to find housing for 1,000 homeless veterans.
“I don’t think we’re doing anything wrong, necessarily, despite all the finger-pointing,” he said. “The problem is we’re not doing enough of the right things.”
Among those right things is to work more toward housing people with mental problems, he said.
Gentry agrees there is a relationship between homelessness and rent, and he said the cost of housing is having an effect on the Housing Commission’s efforts. Three years ago, the turnover rate for people moving out of Housing Commission units was 35 percent a year, he said, but now it’s 14 percent because rent increase and a lack of affordable housing has left them with few places to go.
Like Gentry, others involved in helping San Diego’s homeless don’t believe the point-in-time count is accurate, but not everybody is quick to abandon it.
“I would say that in some places like San Diego County, where we have a tremendous amount of volunteers, our point-in-time count gets better every year,” said Michael McConnell, a homeless advocate and former vice president of the Regional Task Force on the Homeless.
McConnell said that the count in San Diego may be more accurate than in other cities in part because many unsheltered homeless people are more visible, especially downtown, where they are clustered in large encampments.
It’s generally assumed that the point-in-time survey undercounts by at least 10 percent, said McConnell, adding that he does not think the unsheltered count is off by the 50 percent estimated by Zillow.
The point-in-time count isn’t the only way homeless are counted in San Diego.
As explained by homeless advocate Tom Theisen, the former president of the Regional Task Force on the Homeless, the two-year-old Coordinated Entry System tracks how many homeless people contact social services around the county. About 5,000 new names were added to that list this year, and the total has grown to 15,000.
But that isn’t completely accurate either. Theisen estimates the list includes about 90 percent of San Diego’s homeless population, but the bigger problem is that it also includes people who shouldn’t be on it.
Ideally, people would not be on the list for long, and their names would be purged after a social worker helped them find housing, he said. But because of the affordable housing shortage, they remain on the list for long periods, and may lose touch with the agencies trying to help them. Over time, Theisen said, they may find housing on their own, move away or even die, but because nobody minding the list knows, their names aren’t purged.
“I actually think the point-in-time count is most accurate,” Theisen said. “And I absolutely think it’s an undercount.”
Theisen said the real value of the point-in-time count isn’t its precision, but its consistency. If the same percent of people are missed each year, the data will accurately reflect any year-to-year change, giving cities a sense of whether the homeless population is growing or decreasing, he said.
Olsen said Zillow formed a government relations team last year, and Glynn’s method could be considered by HUD as a new way of doing the count.
Walker, however, said the task force feels relatively confident with the point-in-time numbers and the process.