After years of preventing homelessness, some local governments in Britain and America began to tackle the issue from a different angle. Instead of searching for the homeless to take them to shelters, they use modelling and data analytics to predict and assist those at risk of homelessness and help them before they lose their safety.

Between government glossaries and classifications and civil society organizations, one may find different definitions of homelessness, depending on language, social and economic conditions, cultural norms and the purpose of the definition.

It is difficult to capture the entire experience of homelessness, as it is not limited to the deprivation of shelter in the physical sense. For a true definition, it is necessary to reflect on the loss of social contact and the sense of belonging to ostracism and crime.

For decades, the governments of different countries, whether developed or developing, have tried to approach this challenge, which is being aggravated by climate change and the spread of the coronavirus. This caused an exponential increase in the number of homeless people in both Britain and the United States, as even the largest cities were not immune from the phenomenon of random camps due to economic inflation.

However, none of these attempts succeeded in finding a final solution, either because they did not give this issue sufficient attention, or they did not realise the tremendous impact that its solution would have. In the best scenarios, they did not possess the data or resources, or improve their management and participation, and they may still be following traditional mechanisms in providing social services, which prolong their duration and increase their complexity.

The biggest loophole in these systems remains that they are preoccupied with results, and the reasons are overlooked, in the sense that many local governments do not realise that someone faces the risk of homelessness until they are on the sidewalk after it is too late to intervene. 

Today, technology offers governments and human services agencies the opportunity to formulate a new concept in dealing with homelessness, which is simply to anticipate the problem by using predictive analytics that identify citizens at risk of losing their homes.

Maidstone Borough Council in the UK has pioneered a preventative approach to delivering social care using data analytics. With a number of partnerships, the council has developed a data and analytics tool called “One View” that collects data to identify local residents who are at risk of homelessness, allowing caseworkers to step in before those most vulnerable lose their homes.  

Looking ahead, there are plans for Maidstone to integrate its system with wider Kent, with the view to creating a more holistic understanding of those at risk of homelessness in the wider area.  

One View’s predictive analytic and natural language processing capabilities enables agencies in Maidstone to identify residents at risk of losing their homes and engage sooner. To do this, the tool brings together historically disconnected data sets within the council to provide early warning signs of those at risk of homelessness, such as missed utility payments or housing assistance.

A similar experience by South Bend, Indiana, which is using data analytics to get ahead of housing vulnerability. The city, in cooperation with Notre Dame University, is bringing together data on code enforcement, utility bill delinquency, and evictions and foreclosures to build a model that can derive insights on housing vulnerability in real-time and potentially predict which households are most vulnerable.

On the other hand, the Los Angeles County found scores of people through a predictive tool developed by University of California, Los Angeles (UCLA), researchers, which pulls data from eight L.A. County agencies, for example, how many people have been hospitalised, jailed, have experienced psychological problems, received cash or benefits, or free government services, to help outreach workers focus their attention and assistance on people believed to be at gravest risk of losing their homes. The predictive model being used in L.A. County uses an algorithm that incorporates about 500 features.

However, collecting data is not an easy endeavour, as obtaining some information – such as judicial information – can be a challenge, not to mention that any error in entering it will lead to an unrealistic and impractical model.

Another challenge highlighted by the U.S. experiences is that some evictions do not go through the courts and are not documented with any data. A related project focuses specifically on water delinquency – that is, people falling behind on their bills. The project is intended to identify the period of time in the water delinquency cycle where intervention is most helpful.

Governments are in the unique position of being able to utilise existing data that private organisations and non-profits cannot, and share this information with the right service providers to facilitate accurate, real-time decision making. One View was designed to pseudonymise and protect all personal data. If an individual is triggered on the system, their information is kept private and only the case worker assigned to that person has access to it.

As with any data approach, it requires staff buy-in to be successful and sustainable. For this reason, the One View platform was designed with a focus on the staff that would be interacting with it. 

Data alone does not provide a sustainable solution to homelessness; hence Maidstone updated its entire service delivery model, which meant retraining frontline caseworkers to interact with vulnerable families and individuals differently when they offer assistance now that they are seeing that bigger picture.   

During the pilot year, over 650 alerts were generated. Those who were identified as highest risk were provided with an early intervention service – of those, only 0.4% became homeless. Those who presented with lower risk factors went through the process that they normally would have gone through – of those, 40% become homeless.

During the height of the Covid-19 crisis, 100 households were prevented from becoming homeless and the overall rate of homelessness in Maidstone fell by 40%. The information available through One View enabled other benefits: the council generated £2.5 million in societal savings, the time spent on administrative tasks was reduced by 61 days and over 15 different data files were consolidated – providing a more comprehensive overview of residents while driving interagency collaboration. 

These tools provide social service agencies with valuable data to draw upon to help or guide residents to get help, keeping them in their homes and shielding them from one of the harshest human experiences. But by shutting down the source that annually generates increasing numbers of homeless people, governments will save many costs such as building and managing shelters.

References:

• https://www.govx.digital/data/how-predictive-analytics-reduced-homelessness-by-40  
• https://www.ey.com/en_ae/government-public-sector/how-can-data-stop-homelessness-before-it-starts 
• https://www.smartcitiesdive.com/news/cities-predict-homelessness-analytics/635242/ 
• https://www.latimes.com/california/story/2022-06-12/homeless-prevention-unit 
• https://cities-today.com/cities-are-looking-at-data-differently-to-fight-homelessness/