Economic and social factors associated with the use of homeless shelters
La version française de ce billet se trouve ici.
Researchers seeking to understand what causes homelessness often pay close attention structural causes—that is, factors that determine how much homelessness is to found in a specific geographical region at a particular time. To help shed light on the Canadian context, Ali Jadidzadeh and I recently undertook some statistical analysis on behalf of Infrastructure Canada (Canada’s lead federal government agency on homelessness).[1]
Here are 10 things to know:
1. We used shelter data prepared for us by Infrastructure Canada. Data, spanning January 2012-December 2019, encompasses three regions (Ontario, Atlantic provinces, and Western provinces). Infrastructure Canada was able to access this by virtue of data-sharing agreements it has in place across Canada.
2. Infrastructure Canada was intentional about wanting to make this research relevant to the broader research community. As we worked through the statistical analysis, they asked us to draft a stand-alone document that would provide an overview of where to find the kind of data we used for our analysis (particularly the data we didn’t get from them). That document can help other researchers wanting to do similar types of analysis in the future.
3. A literature review has also been made available through the project. It looks at what causes homelessness. This too can be a resource for other researchers (the literature review can be found here).
4. Consistent with the literature, we find that high rental housing costs are associated with more homelessness. Specifically, we find that a 1% rise in monthly real rent for affordable units (i.e., the first quartile of real rent) is associated with a 1.36% and 2.47% increase in total admissions to homeless shelters in both Ontario and the Western provinces, respectively.
5. Social assistance benefit levels (i.e., how much money recipients receive) matter for homelessness. For example, our study finds that a 1% increase in real social assistance benefit levels results in a roughly 2% decrease in single total admissions to homeless shelters in Ontario and the Atlantic provinces.
6. Lengthy spells of unemployment are associated with more shelter use. In Ontario, the duration of unemployment has a positive and significant effect on total admissions. When the duration of unemployment increases by 1% in Ontario, total admissions rise by 0.157% (with all other variables held constant).
7. Perhaps puzzling to some, a higher minimum wage is correlated with more shelter use. The real minimum wage exhibits a positive and statistically significant relationship in Ontario and the Atlantic provinces. When the real minimum wage increases by 1% in these regions, total shelter admissions increase by 0.449% and 2.102% in Ontario and the Atlantic provinces, respectively. It may be that minimum wage rates increase during times of strong economic growth, exactly the time when rent levels are higher (and during which time there are fewer available low-cost housing units).
8. Regional context clearly matters, but for a variety of reasons. In some cases, it may be a result of data aggregation. Data for our Western and Atlantic regions, for example, had to be aggregated. And while data only from single adult shelters are available for Ontario and the Atlantic provinces, the data for the Western provinces includes aggregates of both single adults and family shelters.
9. The analysis itself (i.e., the full report) provides disaggregations across household types. This includes breakdowns for single adults (aged 25 and older) without dependants, families and youth (aged 16-24).
10. Our analysis searched for association, not causation. Put differently, while factors found to be associated with higher use of shelters may have caused greater use of shelters, causation could have happened in the other direction as well (i.e., higher use of shelters may have been the cause of the association). Likewise, an external factor not captured in our model may have been the true cause of the phenomenon in question (recall point #7 above regarding the minimum wage).[2]
In sum. Infrastructure Canada’s collegial approach to data sharing allowed us to undertake this important exercise, which found the following factors can help us better understand what may drive increased use of homeless shelters in Canada: rental market factors; social assistance; duration of unemployment; and the minimum wage.
I wish to thank Ali Jadidzadeh, Sylvia Regnier and Annick Torfs for assistance with this blog post. Any opinions or interpretations in this blog post are those of the author and do not necessarily reflect those of Housing, Infrastructure, and Communities Canada.
[1] On 20 June 2024, Infrastructure Canada was renamed Housing, Infrastructure and Communities Canada.
[2] Careful researchers typically exercise caution when using the term causation. In fact, there are long-standing tensions among academic disciplines as to what methodological approaches are required to establish it.
