Social housing protects against homelessness – but other benefits are less clear
Social housing, managed by governments and the community sector, provides a safety net to vulnerable Australians. A person living in social housing is far less likely to experience homelessness than someone battling it out in the private rental market.
But our research failed to find evidence of social housing residents achieving better outcomes in any of these other areas than similar residents in the private market – at least in the short run.
Prior to our research, there was limited evidence of the impact of social housing on these outcomes. State governments spend hundreds of millions of dollars providing social housing with an incomplete understanding of what that investment delivers. This needs to change.
Social housing is particularly effective at reducing homelessness because it's affordable, as rents are typically set at around 25% of income. And Commonwealth Rent Assistance for those in the private market has been failing to keep up with increasing rental costs over recent years.
Also, social housing, particularly public housing, tends to be secure with long-term – and sometimes even life-time – leases available.
Like other welfare policies in Australia, social housing tends to be highly targeted to those with particularly low means. It has also increasingly become targeted to specific priority groups – such as those fleeing family violence or people with a disability. As at June 2017, around 395,691 households were in social housing across the country, at a cost of A$3.9 billion per year to state and territory governments.
Government policy generally assumes that providing social housing to vulnerable people will result in improvements across a range of life outcomes. As a 2017 Productivity Commission report on housing and homelessness notes: "a lack of adequate and affordable housing contributes to housing stress and homelessness, and is detrimental to people's physical and mental health. Homelessness affects life expectancy, with homeless people estimated to live 15–20 years less than the mainstream population."
In our research we analysed the effects of social housing on measures of these and other key outcomes as well as on the risks of incarceration and homelessness. We used the Journeys Home survey (which measures a range of characteristics over points of time of people vulnerable to homelessness placed in social housing as well as those of similar individuals not in social housing).
Comparing the outcomes for social housing residents with a statistically constructed control group, we did find placing a vulnerable person in social housing significantly reduced their risk of homelessness. In the period following placement, the person's probability of experiencing homelessness was 13 percentage points lower than similar individuals not in social housing, who have a homelessness rate of about 20%. This is equivalent to a 65% reduction in the risk of homelessness for social housing residents.
But there were no statistically significant effects on employment, education, health (which included measures of mental health, self-assessed physical health and having a long-term health condition) or incarceration.
Already too vulnerable
The most likely explanation for this is that access to social housing has long been targeted to society's most vulnerable members who already struggle with factors such as employment or education due to age, family commitments, disabilities and long histories of disadvantage.
Similarly, long-term health conditions – be they physical or psychological – which have been found to be associated with vulnerability to homelessness, remain even after entering social housing. Although social housing provides the stability necessary for people to seek treatment, our research tells us this doesn't seem to flow through to measurable outcomes.
It's also possible the effects of social housing differ across different groups of people and different types of housing. Young people, for example, may have very different employment outcomes to older people – but averaging out the results means these cohort-specific effects may be lost.
Similarly, it's the children of residents who may be most likely to benefit from the security social housing provides. Analysis in the US found that children, particularly younger children, gained the most from experiments to improve their families' housing situation. But we did not examine children in our analysis.
The US research also suggests it's not just the housing that matters, but the neighbourhood too. Children of families who moved to areas with lower concentrations of poverty and crime when the children were young (below age 13) showed significantly improved educational outcomes and earnings later in life.
Why does this matter?
All of this highlights the importance of separately measuring the cohort-specific effects of programs and taking into account the neighbourhood characteristics of where people live.
Our research was unable to sufficiently take account of these factors, as we had no access to suitable large datasets, or those that include the children of residents. Much of the data that would enable this kind of analysis resides in isolation across government departments. Poor links between these sources limits researchers' access to it.
Having access to data to conduct such analysis would give governments an opportunity to consider if the current approach of only providing social housing to the most vulnerable is optimal. Or whether they should, perhaps, expand the social housing program to capture other at-risk cohorts.
Cost-benefit analysis is routinely used to assess the value of major transport infrastructure projects. But business cases for social infrastructure projects tend to rely on qualitative assessments of benefits. We have the tools to quantify the impacts of social housing but these need to be applied to the right data.
This will help ensure governments make the right investment decisions and allocate resources more efficiently while still keeping equity concerns top of mind.