How the big data revolution can help the homeless
Homeless policy needs to join the big data revolution. A data tsunami is transforming our world. Ninety percent of existing data was created in the last two years, and Silicon Valley is leveraging it with powerful analytics to create self-driving cars and to revolutionize business decisionmaking in ways that drive innovation and efficiency.
Unfortunately, this revolution has yet to help the homeless. It is not due to a lack of data. Sacramento alone maintains data on half a million service interactions with more than 65,000 homeless individuals. California is considering integrating the data from its 44 continuums of care to create a richer pool of data. Additionally, researchers are uncovering troves of relevant information in educational and social service databases.
These data, however, are only useful if they are aggressively mined for insights, looking for problems to solve and successful practices to replicate. At that juncture California falls short.
Take Rapid Rehousing, a key program that offers rent subsidies and social services to help people regain stable housing. In 2018 it served more than 1,100 Sacramento clients. If this program’s data were mined for trouble areas, the status lights would be flashing red. In 2015 it successfully rehoused 88% of its clients, but by 2018 that rate plummeted to 47%. One could place blame on the tight housing market, but several programs continue to rehouse more than 80% of their clients. In contrast, county programs are struggling. So, data mining can identify a problem, narrow possible causes and point where to look for solutions. If managers acted on this knowledge, they could potentially move 250 or more additional families off the streets each year.
Integrating homelessness data with information from social service and criminal justice agencies is another promising avenue. In 2017, AB 210 created a framework by which multidisciplinary teams could share confidential data across agencies. The potential is great.
Communities need better programs to prevent homelessness before it begins. Unfortunately, targeting such services is challenging because 98% of people at risk of homelessness avoided homelessness without public assistance. To help identify the 2% truly in need, USC Homelessness
Policy Research Initiative researchers worked with Los Angeles County to mine more than 85 million service records from a range of county programs. Their models are up to 48 times better at identifying who is at risk of homelessness compared to the average. They predict that serving just 1% of county clients identified at the greatest risk could prevent 6,900 homeless spells a year.
Data mining can also sniff out hidden program inequities. People of color are much more likely to experience homelessness. A question is whether programs treat them equitably. To take a deeper look the Los Angeles Ad Hoc Committee on Black People Experiencing Homelessness combined data from four communities. They found that although black and white individuals received the same services, black people fell out of permanent supportive housing at faster rates. HPRI is currently researching the causes of these disparities, but better data that accurately identifies and tracks client outcomes are key to recognizing and addressing racial disparities.
Is big data a silver bullet? Absolutely not. Nonetheless, effective data mining can greatly increase the number of people moved from the streets into stable housing. In doing so, it can demonstrate program effectiveness, shoring up public support for an aggressive response to this problem. The constraints on joining the data revolution are not technical. Rather, they are bureaucratic and political. AB 210 was a critical first step. Next, California needs to identify and empower champions who can connect their strategies and operations with the promise of big data.
Gary Painter is a professor at the USC Sol Price School of Public Policy and director of the USC Homelessness Policy Research Institute. Christopher Weare is president of the Center for Homeless Inquiries.