The Extent and Nature of Veteran Homelessness in Canada

Aaron Segaert
Alana Bauer
Community Development and Homelessness Partnerships Directorate
Employment and Social Development Canada
March 2015

Veteran Homelessness in Canada

Quick Summary

  • First study to look at Veteran homelessness nationally
  • Uses Homeless Individuals and Families Information System (HIFIS) data collected at 60 emergency shelters across Canada between January 1, 2014 and December 31, 2014
  • It is estimated that 2,250 Veterans use homeless shelters each year in Canada
  • About 2.7% of annual shelter users were identified as Veterans
  • There is a higher rate of episodic homelessness for Veterans compared to non-Veterans
  • Homeless Veterans are older, on average, than homeless non-Veterans

Introduction

Homelessness among Veterans is known to be a significant problem, particularly in the United States. The 2014 Annual Homelessness Assessment Report (AHAR) to the United States Congress reported that Veterans account for 11.3% of homeless adults in the United States, but that the number of homeless Veterans has decreased by 33% since 2009 (AHAR, 2014). The Canadian context differs from that of the United States, and may be more similar to that of the United Kingdom or Australia. In the UK, Veterans make up approximately 6% of the homeless population. In Australia, there are about 3,000 homeless Veterans, which is around 3% of the homeless population (Ray & Forchuk, 2011).

The present study uses data collected by emergency shelters using the Homeless Individuals and Families Information System (HIFIS) to provide information about Veteran homelessness in Canada. The available data can provide details on the demographic characteristics of homeless Veterans, a comparison between homeless Veterans and non-Veterans, and a baseline estimate of the proportion of Canada’s sheltered homeless population who are Veterans.

According to Veterans Affairs Canada (VAC), there are an estimated 697,400 Veterans in Canada (Veterans Affairs Canada, 2014).  Of Canadian Armed Forces (CAF) Veterans aged 20 to 84 living in Canada, the majority (87%) are male, with CAF Veterans representing about 4% of Canada’s male population (MacLean et al. 2013).  Little is known about homelessness among Veterans in Canada, but there is some concern that Veterans may be overrepresented in the homeless population. The Metro Vancouver Region’s 2014 point-in-time count asked about military service, and 7% of respondents indicated that they had served in the Canadian Forces (Greater Vancouver Regional Steering Committee on Homelessness, 2014). Another recent homeless count in Waterloo Region found 5% of respondents reporting military service (Region of Waterloo, 2014). Similarly, Alberta’s seven-city (Calgary, Edmonton, Red Deer, Grande Prairie, Medicine Hat, Wood Buffalo, and Lethbridge) coordinated point-in-time count found 6% had military service (Turner, 2015). Seven percent of respondents reported military service in the 2013 Toronto Street Needs Assessment (City of Toronto, 2013). Baseline data for the At Home/Chez Soi project reported 4.3% of participants identifying themselves as Veterans (Bourque, 2014). The number of Veterans identified as homeless on VAC’s electronic database, the Client Services Delivery Network (CSDN), has increased over the past few years, from 131 in 2011 to 500 in 2015 (Veterans Affairs Canada, 2015).

In terms of the characteristics of homeless Veterans, Forchuk and Richardson (2014), following a group of 78 Veterans who have experienced homelessness in London, Toronto, Victoria and Calgary, found participants to be overwhelmingly male (92.1%), with an average age of 52.8 years. This is much older than what is found in the National Shelter Study (Segaert, 2012), which reports the average age of adult shelter users to be 37 years. A long gap between leaving the military and becoming homeless was noted among participants. On average, participants had been homeless for about eight years, and had been out of the military for an average of 28.4 years. Alcoholism, drug addictions and mental health issues were cited as major factors contributing to their homelessness. The At Home/Chez Soi project found Veterans and non-Veterans to have similar addiction and mental health issues (Bourque, 2014).

Methods

The data used in this study are from a national database containing non-identifying information about homeless shelter use collected by HIFIS. HIFIS is a software system developed by the Government of Canada which is used by homeless shelters to book clients in and out of the shelter, and by a broader pool of service providers for more detailed case management and housing placement. Shelters using HIFIS are now required to enter information about Veteran status, among other basic characteristics, for clients using the shelter. The software provides the options of “Veteran”, “Not a Veteran” or “Unknown”. Shelters have been instructed to encourage their staff to classify anyone who has been in the military as a Veteran, regardless of their official Veteran status, as this can sometimes be difficult to determine. Veteran status is a recent addition to HIFIS, only coming into full usage during the past year. Therefore, information about Veteran status is not available for many cases prior to 2014, with only a relatively small number of cases in 2014 having missing data. Accordingly, this study will focus on the period of January 1 to December 31, 2014, and cannot yet speak to trends in Veteran homelessness.

Sixty emergency shelters with the most complete HIFIS data on Veteran status during the one-year study period were selected for the study. The selected shelters are located all across the country, including the far north, and from communities ranging from small towns to large cities. The sample of 60 shelters with 2,254 beds comprises 15% of the emergency shelters in Canada, and 15% of available shelter beds. To ensure representative age and gender results, the number of beds at family, youth, male, female and general coed shelters in the sample is proportionate with the overall shelter system in Canada. These shelters were used by 16,096 unique individuals during 2014 (please see Appendix A for a summary of the sample).

The demographics of the sample matched closely to the results of the previously published National Shelter Study (Segaert, 2012) and other known studies of the homeless population, with 73.8% of clients male, 26% female and 0.2% other. For age groups, 3.1% of clients are children under 16, 21.9% are youth between 16 and 24, 71.8% are adults aged 25-64, and 3.2% are seniors over 65. Compared to the National Shelter Study, which was based on data from 2005 to 2009, there are slightly more seniors and slightly fewer children using shelters, but this is expected based on more recent demographic trends.

Results

In the sample, 2.1% of shelter users are classified as “Veteran,” 74.7% as “Not a Veteran,” 12.2% as “Unknown” and 11.1% are missing data (Table 2). Given the large amount of unknown or missing data, estimating the overall number of Veterans using shelters annually in Canada should be approached with caution. In absolute numbers, 338 individuals were identified as Veterans. Extrapolating this 15% sample to the full population of shelters suggests there may be 2,250 Veterans using emergency homeless shelters annually in Canada.

Table 2: Raw Distribution of Homeless Individuals by Veteran Status (HIFIS 2014)
N %
Veteran 338 2.1%
Not a Veteran 12,017 74.7%
Unknown 1,956 12.2%
Missing Data 1,785 11.1%
Total 16,096 100%

If we restrict the sample to cases with valid data (i.e., excluding missing and unknown), a slightly higher figure of 2.7% of shelter users were identified as Veterans. This is considerably lower than the 5% to 7% figures found in point-in-time counts conducted recently in a number of Canadian communities.

The percentage of Veterans reported in HIFIS was quite consistent across many shelters and across many communities. In terms of reliability, there was no significant difference in the percentage of Veterans reported at shelters with complete or near-complete Veteran status data compared to shelters with higher amounts of missing or unknown data. This suggests that the missing and unknown cases do not necessarily contain a greater proportion of Veterans than what is found in the cases with valid data. Across communities, there was some variation in the percentage of Veterans found (see Table 3), but none exceeded the 5% to 7% found in point-in-time counts. This table contains only the cities for which representative and fairly complete HIFIS data were available. Thompson, MB and Thunder Bay had the highest percentages, at 4.8% and 4.7% respectively, while Waterloo Region and Windsor had the lowest at 0.5%.

Waterloo Region and Greater Vancouver are the only two communities that conducted recent homeless counts and had sufficient HIFIS data available for comparative purposes. The HIFIS data for Waterloo, in particular, shows a much lower rate of Veteran homelessness than the point-in-time count (0.5% vs 5%). In the Vancouver point-in-time count, 6% of sheltered individuals reported military service versus 2.4% in the HIFIS data.

Table 3: Percentage of Shelter Users who are Veterans
HPS Designated Community %
Thompson 4.8%
Thunder Bay 4.7%
Winnipeg 3.6%
Victoria 3.5%
Halifax 3.3%
Regina 3.0%
Peel Region 2.4%
Vancouver 2.4%
Yellowknife 1.4%
York Region 1.1%
Waterloo Region 0.5%
Windsor 0.5%
Wellington-Guelph 0.2%

Despite the number of cases with unknown or missing data, this sample does provide an excellent opportunity to look at the characteristics of homeless Veterans vs. the non-Veteran homeless population. For the following analysis, clients with missing or unknown Veteran status were excluded, leaving a large and representative sample of 12,345 clients with a valid response for Veteran status.

Demographics

Among shelter users age 16 and over, Veterans were older on average (41.6 years) than non-Veterans (37.0 years). The proportion of seniors was much higher among Veterans (9.5%) compared to non-Veterans (3.0%), while the proportion of youth was lower among Veterans (18.6%) than non-Veterans (25.4%, see Table 4). Despite the Canadian Veteran population being overwhelmingly male, there was no difference in the gender distribution of Veterans versus non-Veterans among shelter users over the age of 15 (see Table 4).

Table 4: Age Group and Gender for Veterans vs Non-Veterans
Age Group 16-24 25-64 65+
Veterans 18.6% 71.9% 9.5%
Non-Veterans 25.4% 71.6% 3.0%
Gender Male Female Other
Veterans 71.8% 28.2% 0.0%
Non-Veterans 73.0% 26.7% 0.2%

Looking more closely at the demographic characteristics of female Veterans, 34.7% are under age 25 (see Table 5). Half of Veterans under 25 are female while less than one-quarter of Veterans 25 or over are female. The total number of female youth who are Veterans is quite small (only 33 cases) so this finding must be approached with caution. However, confidence in this finding increases due to the fact that young female Veterans were found in eleven cities in five different provinces.

Table 5: Cross tabulation of Age Group and Gender for Veterans
Youth (16-24) Adults (25-64) Senior (65+) Total
Male
- Count 30 186 26 242
- Percent within age group 47.6% 76.9% 81.3% 71.8%
- Distribution across age groups 12.4% 76.9% 10.7% 100%
Female
- Count 33 56 6 95
- Percent within age group 52.4% 23.1% 18.8% 28.2%
- Distribution across age groups 34.7% 58.9% 6.3% 100%
Total
- Count 63 242 32 337
- Percent within age group 100% 100% 100% 100%
- Distribution across age groups 18.7% 71.8% 9.5% 100%

Chronic and Episodic Homelessness

Clients with three or more episodes of homelessness during the year are considered episodically homeless. Clients who stayed in shelters for 180 or more nights are considered chronic. In the vast majority of cases, multiple shelter stays by an individual all occur within the same community. Although the sample consists of only 15% of shelters, stay histories are based on relatively complete data for several cities. However, incomplete stay histories may result in underestimates of chronic and episodic homelessness.

The majority of clients (73.2%) had only one episode of homelessness during the study period, which is consistent with the established literature on patterns of homeless shelter use (Aubry, Farrell, Hwang, & Calhoun, 2013; Kuhn & Culhane, 1998). Among clients with one or two episodes of homelessness, 1.9% were Veterans, while 3.9% of episodically homeless clients were Veterans.

Table 6: Rates of Chronic and Episodic Homelessness
Chronic Episodic
Veterans 0.6% 15.7%
Non-Veterans 0.7% 8.6%

The rate of chronic homelessness is similar for Veterans and non-Veterans, and is less than 1% for each (see Table 6). This is quite low, although not unusual for many communities, and could be partially explained by incomplete stay histories. There was a much higher rate of episodic homelessness among Veterans compared to non-Veterans. Among non-Veterans, the rate of episodic homelessness was 8.6%, which is typical of the rate of episodic homelessness found in most communities. The rate of episodic homelessness was 15.7% among Veterans, which is almost double that of non-Veterans.

A closer examination of the demographic characteristics of episodically homeless Veterans shows 69.2% were male and 30.8% were female, not much different than the gender breakdown for all Veterans. The average age of episodically homeless Veterans was 40.2, not significantly different than the average of 41.6 for all Veterans. However, comparing the rate of episodic homelessness between Veterans and non-Veterans (Table 7) for each gender and age group shows that the rate of episodic homelessness is particularly high among female Veterans, at 16.8% compared to only 6% among female non-Veterans. Among age categories, the rate of episodic homelessness is similar between youth and senior Veterans and non-Veterans, but much higher for Veterans aged 25-54 than for non-Veterans in that age group.

Table 7: Rates of Episodic Homelessness by Gender and Age Group
Veterans Non-Veterans
Male 14.9% 9.5%
Female 16.8% 6.0%
Youth 10.9% 8.3%
Adult 18.2% 9.1%
Senior 6.3% 6.2%

Looking at other stay-related variables, the average stay length in shelter was virtually the same for Veterans and non-Veterans, at 19.5 nights and 20.6 nights respectively. The distribution of short, medium and long stays was also nearly identical, with Veterans just slightly more likely to have more short stays.

Finally, HIFIS records a reason for service for each shelter stay. Table 8 shows the top ten reasons for stays by Veterans and non-Veterans. The lists are remarkably similar.

Table 8: Top Ten Reasons for Service
Reason for Service (Veterans) % of Stays Reason for Service (Non-Veterans) % of Stays
Lack of Housing 17.0% Lack of Housing 17.8%
Family/Relationship Breakdown 10.0% Family/Relationship Breakdown 13.5%
Transient Lifestyle 9.6% Transient Lifestyle 7.9%
Financial Crisis 7.4% Eviction by Landlord 6.8%
Eviction by Landlord 5.8% Financial Crisis 4.8%
Discharge from Correctional/Jail 5.1% Discharge from Correctional/Jail 4.7%
New Arrival to area 2.1% New Arrival to area 3.4%
Stranded in area 1.9% Eviction by other than landlord 2.8%
Eviction by other than landlord 1.7% Unsafe housing 2.5%
Unsafe Housing 1.7% Stranded in area 2.0%

Discussion

In the present study, the proportion of shelter users who are Veterans is significantly lower than what was found in recent point-in-time counts in Toronto, Vancouver, Waterloo and the seven cities in Alberta. There are several possible explanations for the differences. They could be attributed to the methods used, namely point-in-time versus annual shelter data. The differences could be due to undercounting Veterans in the HIFIS data, given the high rate of missing and unknown data. Or there may be over-reporting of Veteran status in point-in-time counts.

Point-in-time counts enumerate the homeless population on a specific day and count both the sheltered and unsheltered homeless. Point-in-time counts are thus useful for capturing individuals who do not use shelter services. For example, the Greater Vancouver point-in-time count found 2,777 homeless individuals, which is less than 20% of the unique individuals using a shelter over the course of a year in Vancouver. What this means is, point-in-time counts will tend to have a higher proportion of chronically homeless individuals, as they are more likely to be homeless at the time of the count.

Conversely, counts using annual shelter data will capture a greater share of temporarily homeless people. Annual shelter counts, such as the present study, only count those who use shelters, but due to the longer survey period, capture a much larger part of the homeless population. The literature in the United States and Canada clearly shows that the majority of homeless shelter users experience only a single, short-lived episode of homelessness (Aubry, Farrell, Hwang, & Calhoun, 2013; Kuhn & Culhane, 1998). In the present study, three-quarters of clients had only one shelter stay. The rate of veteran homelessness is low among these temporarily homeless clients, which contributes to the lower proportion of veterans found in annual shelter data compared to point-in-time counts.

In terms of sheltered and unsheltered homeless, there are some important considerations that may impact the number of Veterans reported using these different methodologies. While it is well known that there are homeless individuals who prefer not to use shelters, the number who do not use a shelter even once over the course of a year is believed to be small. In addition, many cities have made concerted efforts to reduce street homelessness, which increases the probability that homeless individuals will use a shelter at least once during the course of a year. With respect to Veterans, there is some evidence that they may be more likely than others to be found sleeping rough. For example, the Toronto count found a higher proportion of Veterans among the unsheltered homeless (11%) versus 7% in shelters. There corresponding numbers for Vancouver were 7% and 6% respectively.

Thus, the higher rate of Veteran homelessness found in the point-in-time counts could suggest that Veterans are more likely to be found outside of shelters. This is consistent with HIFIS data that shows a higher rate of episodic homelessness among Veterans.

Conclusion

The demographic characteristics and stay patterns of Veterans using shelters generally match expectations based on existing research. Most homeless Veterans are adult males, and overall they tend to be older than homeless non-Veterans. Veterans were three times as likely to be over 65 as non-Veterans. The proportion of female Veterans using shelters decreases for older age groups, consistent with other Canadians (Segaert, 2012).

An important finding is that there is a high rate of episodic homelessness among Veterans. This is consistent with point-in-time counts which suggest there may be more Veterans experiencing conditions such as sleeping rough. Interestingly, there is a particularly high rate of episodic homelessness among female Veterans. Again the numbers are very small, so percentages can fluctuate easily, but this raises important questions and new lines of potential research.

The results show about 2.7% of shelter users are Veterans. This is lower than the results of several point-in-time counts. As a baseline estimate, 2,250 Veterans use emergency homeless shelters each year in Canada. There may be underreporting due to missing and unknown data in the HIFIS database, however, similar percentages were found for shelters with lots of missing data versus those with very little missing data. The strong inter-shelter reliability for Veteran status in the HIFIS data means that it should be considered a solid baseline for the number of homeless Veterans. In addition, the HIFIS database has yielded important information about the characteristics of homeless Veterans. The amount of unknown and missing data should decrease significantly over time, and will allow future analysis to examine trends over time in Veteran homelessness.

Appendix A

Summary of shelters included in sample by community
Community Province Beds Shelters
St. John’s NL 29 3
Halifax NS 100 2
Montreal QC 169 1
Brantford ON 10 1
Greater Toronto Area ON 382 10
Guelph ON 48 3
Hamilton ON 220 3
Kenora ON 25 1
Niagara Region ON 20 1
Ottawa ON 163 4
Peterborough ON 15 1
Sault Ste. Marie ON 10 1
Sudbury ON 52 2
Thunder Bay ON 59 2
Waterloo Region ON 169 6
Windsor ON 67 2
Thompson MB 24 1
Winnipeg MB 198 3
Regina SK 153 4
Saskatoon SK 8 1
Vancouver BC 166 4
Victoria BC 109 2
Yellowknife NWT 58 2
Total 2,254 60

References

Aubry, T., Farrell, S., Hwang, S. W., & Calhoun, M. (2013). Identifying the Patterns of Emergency Shelter Stays of Single Individuals in Canadian Cities of Different Sizes. Housing Studies, 28(6), 910-927.

Bourque J, VanTil L, LeBlanc SR, Kopp B, Daigle S, LeBlanc J, Sareen J, Darte K, Landry LA, More F. Correlates of Veteran Status in a Canadian Sample of Homeless People with Mental Illness. Canadian Journal of Community Mental Health, July 2014, Volume 1 supplement of At Home/Chez Soi Study. http://www.cjcmh.com/doi/pdf/10.7870/cjcmh-2014-031.

City of Toronto (2013). Street Needs Assessment Results 2013.

Forchuk, C. & Richardson, J. (2014). National Veterans Project: Addressing Homelessness Among Canadian Forces Veterans.

Greater Vancouver Regional Steering Committee on Homelessness. (2014). Results of the 2014 Homeless Count in the Metro Vancouver Region.

Kuhn, R. & Culhane, D. P. (1998). Applying Cluster Analysis to Test a Typology of Homelessness by Pattern of Shelter Utilization: Results from the Analysis of Administrative Data. American Journal of Community Psychology, 26(2), 207-232.

Maclean, M.B., VanTil, L., Kriger, D., Sweet, J., Poirier, A., & Pedlar, D. (2013). Well-being of Canadian Forces veterans: Canadian community health survey 2003. Charlottetown: Veterans Affairs Canada, Research Directorate.

Ray, S.L.& Forchuk, C. (2011). The Experience of Homelessness Among Canadian Forces and Allied Forces Veterans (Research Report submitted to HRSDC Canada).

Segaert, A. (2012). The National Shelter Study: Emergency Shelter Use in Canada 2005-2009. Ottawa: Homelessness Partnering Secretariat, Human Resources and Skills Development Canada.

The U.S. Department of Housing and Urban Development. (2014). The 2013 Annual Homeless Assessment Report (AHAR) to Congres - Part 1: Point-in-Time Estimates of Homelessness.

Turner, A. (2015). 2014 Alberta Point-in-Time Homeless Count: Provincial Report. (Prepared for the 7 Cities on Housing & Homelessness).

Region of Waterloo (2014). Waterloo Region Registry Week Community Debrief (This presentation is no longer available online).

Veterans Affairs Canada (December 31, 2014). Service Delivery Management Report, Clients with Homeless Indicator Populated by District Offices and Areas.

Veterans Affairs Canada, Statistics Directorate (2014). Facts and Figures Book, September, 2014.